pixel cinematography

of 73 /73
P I X E L C I N E M A T O G R A P H Y A Lighting Approach for Computer Graphics John Kahrs, Animation Director, Blue Sky Productions Course Organizer Sharon Calahan, Lighting Supervisor, Pixar Dave Carson, Digital Effects Supervisor, Industrial Light and Magic Stephen Poster, A.S.C. Siggraph ’96 Course #30

Upload: oliver-spasovski

Post on 30-Oct-2014




9 download

Embed Size (px)


Pixel CinematographyA Lighting Approach for Computer Graphics


Page 1: Pixel Cinematography


A L igh t ing Approach fo r Compute r Graph ics

John Kahrs, Animation Director,Blue Sky ProductionsCourse Organizer

Sharon Calahan, Lighting Supervisor,Pixar

Dave Carson, Digital Effects Supervisor,Industrial Light and Magic

Stephen Poster, A.S.C.

Siggraph ’96 Course #30

Page 2: Pixel Cinematography
Page 3: Pixel Cinematography


Schedule 5

Speaker Biographies 7

Course Introduction 9

Storytelling Through Lightingby Sharon Callahan 11

Lighting from a Filmmaker’s Perspectiveby Stephen Poster 41

Pixel CinematographyLighting for Computer Graphicsby John Kahrs 43

Lighting for Compositing and Integrationby Dave Carson 69

Page 4: Pixel Cinematography

4 P I X E L C I N E M A T O G R A P H Y

Page 5: Pixel Cinematography

P I X E L C I N E M A T O G R A P H Y 5


8:30 am IntroductionKahrs

8:40 am Lighting from aFilmmaker’s PerspectivePoster

10:00 am Break

10:15 pm Storytelling Through LightingCalahan

12:00 noon Break

1:30 pm A Lighting Approach forComputer ImageryKahrs

3:00 pm Break

3:15 pm Lighting for Compositingand IntegrationCarson

Page 6: Pixel Cinematography

6 P I X E L C I N E M A T O G R A P H Y

Page 7: Pixel Cinematography


Sharon Calahan, Lighting SupervisorPixar Animation Studios

As the creative Lighting Supervisor for Pixar’s“Toy Story”, Sharon Calahan has been a mem-ber of the technical team at Pixar for the lasttwo years. Her background and education inart and design led her into advertising, broad-cast TV, video production, and eventually com-puter animation. With a focus on lighting direc-tion, Sharon has worked in computer anima-tion for over ten years. Besides “Toy Story” andvarious commercial work, other accomplish-ments have been as the computer animationLighting Director for Hanna-Barbera’s “TheLast Halloween” which won an Emmy for Spe-cial Effects.

Dave Carson, Visual Effects SupervisorIndustrial Light & Magic

Dave Carson has been at ILM for over 15 years,beginning as a storyboard artist and modelmaker on the second and third Star Wars films.He has worked in various roles on many re-markable films, primarily as a Visual EffectsArt Director and Visual Effects Supervisor. Hiswork in the digital realm includes acting as aDigital Artist on “Hook”, “Forrest Gump” and“Jurassic Park”. He also contributed characterdesign and animation on “Casper” where hewas credited as Character Design Supervisor.His latest projects include supervising the up-dating of work in “Empire Strikes Back” and“Return of the Jedi” for their new film releases.He is currently scheduled to begin work as aVisual Effects Supervisor on the next film in theStar Wars series when it goes into productionlater this year.

John Kahrs, Animation DirectorBlue Sky Productions

John has been directing lighting and animationat Blue Sky Productions since 1990. The focus atBlue Sky has been on a classic approach tocharacter animation, combined with the verybest rendering techniques. At the core of theproduction system is a proprietary raytracer,for which John has written much of the user’smanual. John has made a priority of refiningBlue Sky’s lighting techniques. His lighting andanimation appears in several commercials forclients including Braun razors, Chock-full-O’-Nuts coffee, and Brother laser printers. John de-signed and constructed the Blue Sky web site.He also has outlined the lighting direction forthe CG cockroaches in the upcoming featurefilm “Joe’s Apartment”. In 1993, John won aGolden Nica Award for his radiosity imagery atthe Ars Electronica festival in Linz, Austria.

Steven Poster, A.S.C.,Cinematographer

Stephen Poster has worked on dozens of films,including Ridley Scott’s “Someone To WatchOver Me”, “Big Top Pee-Wee” and most re-cently “Roswell,” about the reported crash of aUFO in New Mexico in 1947. Originally fromChicago, Poster was called upon early in hiscareer to shoot second unit photography on“Close Encounters of a Third Kind” and “BladeRunner”.

P I X E L C I N E M A T O G R A P H Y 7

Page 8: Pixel Cinematography

8 P I X E L C I N E M A T O G R A P H Y

Page 9: Pixel Cinematography


How do you “Teach” Lighting?

Software tools and complex lighting models forcomputer graphics are some of the most elegant,sophisticated technologies of our time, yet theattention paid to lighting and refining the im-ages is often minimal, and sometimes practicallynonexistent. Conversely, we also see subtle,beautiful, resonant images made with comput-ers. What accounts for this disparity?

An answer may lie in the fact that some com-puter artists have a deeper understanding oflight and material qualities, while others maynot even consider lighting as an issue. Theymay not have trained themselves to see and un-derstand how light works, especially with theoften incomplete lighting model in computergraphics.

This course focuses on the craft of lighting forcomputer graphics. Using a hybrid approach oftraditional cinematography and knowledgeabout composition, color, balance, and the be-havior of light and materials, it offers a compre-hensive approach for lighting specifically in thefield of computer graphics.

I think the idea for a lighting course specificallyfor CG is very timely. It’s almost to the pointwhere it’s hard to find a sizeable Hollywoodfilm without some kind of digital effect of somesort. The medium of computer animation is, Ithink, entering a Golden age. Software toolsmore powerful than ever, and elegant in theirsophistication. A beautiful film called Toy Storyhas been embraced in and outside the graphicscommunity. To just watch the Siggraph filmshows from the past decade is to see technol-

ogy evolve into artistry. We hear a lot abouthow there’s no ceiling, there’s no end in sight,we’re only just beginning, and all this limitlessoptimism can get on your nerves after a while,but the funny thing is that it’s the truth.

Part of what inspired the idea for a course onlighting is the animation courses at Siggraphthat seem to pop up every other year or so. Ihoped to do for CG lighting what JohnLasseter, Chris Wedge and others have done forcomputer animation. The influence of tradition-ally trained animators in the new medium re-flected a sea change that was occurring in thelate eighties: those who used the classic prin-ciples of animation applied them when usingthe new tools. This mood culminated in a 1987Siggraph course: 3D Character Animation byComputer, and more recently, 1994’s AnimationTricks. Suddenly CG animation had grown up.The cliche of slow, computer-smooth motionbecame less prevalent. Now it was entertaining,exciting, and the entire medium was beingtaken more seriously. They succeeded becausetraditional techniques, hammered out overyears of practical use and distilled down to alist of basic principles, were skillfully appliedto a new medium.

I hoped that a similar approach could be appliedto computer lighting: where the principles of tra-ditional cinematography could be applied to thenew tools. This is possible, but only to a certainextent. This is partly because, while there aremany strong parallels, traditional techniquesaren’t so easily portable to computer techniques,as they are with animation.

P I X E L C I N E M A T O G R A P H Y 9

Page 10: Pixel Cinematography

As I wrote the course notes for my part of thetalk, the idea of the course changed drastically.I had thought that the course speakers couldteach lighting, plain and simple. I reallythought, for some time, that in lighting too,much of the task could be distilled down to anessential list, and my ultimate model for such alist was “The Principles of Animation”, a chap-ter in the indispensable book, The Illusion of Life:Disney Animation, by Frank Thomas and OllieJohnston.

Then I was on the phone one day with StevenPoster, the cinematographer I asked to speak atthe course to offer a look at lighting from a tra-ditional angle. He said out loud what I hadbeen sensing deep down more and moreclearly. He said, “Oh, absolutely, no. No, youcan’t teach lighting. You can’t teach someonehow to light. You can only teach them aboutlight and how it works, and you can give thema few guidelines, but you can’t teach anyonehow to light.”

I realized my folly in presuming this. It waslike figure drawing class in art school. No onecould teach us how to draw. Only we, the stu-dents, could teach ourselves to draw better. Theinstructor was merely trying to get us to seemore clearly: to observe and measure with oureyes and compare what we saw with what wehad drawn. If the instructor was good, he wastrying to teach us to see.

The process of computer graphics work is likeworking with a kind of complex diorama-ma-chine. We’re creating little worlds, and we canbuild everything almost as if from scratch, be-

1 0 P I X E L C I N E M A T O G R A P H Y

cause the level of control can be so basic. Thereare so many different skills to be proficient inwhen we do this. We have to be Renaissancepeople.

This course isn’t going to magically transformanyone’s images into flawlessly refined pic-tures. All it can really do is offer a few guide-lines, provide some important things to remem-ber, and hopefully point you in the right direc-tion with a solid footing about where to start.

The artistry of computer lighting has to comefrom your own vision and intuition about whatyou want to see. If it succeeds, it may help yousee light in a way you hadn’t before, and en-courage you to teach yourself how to createtruly great images.

So how to approach the task of lighting on thecomputer…

John KahrsNew York, May 1996

Page 11: Pixel Cinematography


A Compute r Graph ics Perspec t i ve

By Sharon Calahan


This course is designed as a beginning, non-technical course to discuss the how lighting incomputer graphics can be used to enhance vi-sual storytelling for cinematic purposes. It col-lects knowledge and principles from the disci-plines of design, fine art, photography, illustra-tion, cinematography and the psychology ofvisual perception. Although much of the con-tent of this course is not solely applicable tolighting on the computer, its special needs arealways in mind.

1. Introduction

The desire to write these notes and to present acourse on lighting for storytelling in computeranimation arose from the shortage of availableliterature on the subject. Frequently I am askedto recommend a book or two on lighting, and al-though several good books are available, noneare ideal. Most tend to focus on the equipmentand mechanics of live-action lighting withoutexplaining how to achieve the fundamentalprinciples. The commonality between live-actionlighting and computer lighting is chiefly thethought process, not the equipment. Computertools vary with implementation, are continuallyevolving, and are not limited by physics. Toolsin the future will be driven by the desire to seeon the screen what we are able to visualize inour minds. This course is designed to focus onthese thought processes, while providing notonly practical information, but also the desireand resources to continue exploring.

The use of words alone is inadequate to de-scribe visual concepts. Most books includemany repetitive visual examples to drive thepoint home. Although a few crude visual ex-amples are included in these notes, they aremerely intended to serve as a reminder of thepresentation of this course. These notes are alsonecessarily succinct, and may contain conceptswhich could not fit into the hour-and-a-halftime allotment.

The term lighting in computer animation oftenincludes the task of describing the surface char-acteristics of objects (often referred to asshaders), as well as compositing and the inte-gration of special effects. For the purposes ofthis course, lighting is defined more in live-ac-tion terms as the design and placement of thelights themselves, but in a purely computergraphics environment.

Visual storytelling is a vast topic that reaches farbeyond the realm of lighting. Most of it is notnoticeable on a conscious level to the viewer,but adds depth and richness to the story and thevisual experience. The lighting principles andtechniques presented in this course are dis-cussed in isolation from other visual storytellingdevices. Ideally the lighting would be designedwith these in mind, but would extend far be-yond the scope of this course.

Cinematic lighting literature typically empha-sizes live-action lighting issues and techniques,and in this discussion of lighting for syntheticcinema we will find that many live-action con-cepts apply. However, there are some differ-ences in the approach, roles and responsibili-ties, the size of the crew, and the sequence inwhich tasks are accomplished.

P I X E L C I N E M A T O G R A P H Y 1 1

Page 12: Pixel Cinematography

1 2 P I X E L C I N E M A T O G R A P H Y

In live-action, the lighting design, the staging,and framing of a shot are a collaborative and si-multaneous effort between the director and cin-ematographer. Each activity affects the other,and it is important that they are fine-tuned to-gether. Actors can rehearse the scene, stagingand framing can be altered, and props can beredressed to take best advantage of the lightingdesign. This differs from the pipeline approachoften employed in synthetic cinema, where themodeling, surface design, staging, framing, setdressing, and acting are accomplished sequen-tially, each usually established before the light-ing designer begins to work. It should be keptin mind that the sooner in the production pro-cess the lighting can be designed, the more in-volved it can be in the storytelling process.

Another important difference between live-ac-tion and synthetic lighting can be the role of theart director. In live-action, the art director is ab-sorbed in designing sets and props and is notusually involved in staging, framing, and light-ing design. On the other hand, computer gener-ated animation has a more stylized, illustrativequality, with its roots more in hand-drawn ani-mation than live-action cinema. In addition todesigning sets and props, the art director is alsooften heavily involved in the staging (layout)and lighting decisions. With the director, the artdirector is often responsible for determiningthe lighting style for individual sequences aswell as the film as a whole.

2. Objectives of Lighting

The primary purpose of cinematic lighting isstorytelling. The director is the storyteller and itis his vision that the lighting designer is attempt-ing to reveal. To that end, it is important to un-derstand the story-point behind each shot, andhow it relates to the story as a whole. It is notenough that the lighting designer simply illumi-nate the scene so the viewer can see what is hap-pening, or to make it look pretty. It is the light-

ing designer’s task to captivate the audience byemphasizing the action and setting the mood.

The following six lighting objectives are impor-tant fundamentals of good lighting design.They also break down the thought process intoa good course outline. They are borrowed andadapted from the book Matters of Light andDepth, by Ross Lowell.• Directing the viewer’s eye• Enhancing mood, atmosphere and drama• Creating depth• Conveying time of day and season• Revealing character personality and situation• Complementing composition

3. Directing the Viewer’s Eye—The Study of Composition

The primary objective of good lighting is toshow the viewer where to look. Shots are oftenon-screen only briefly, which means thestorytelling effectiveness of a shot often de-pends upon how well, and how quickly, theviewer’s eye is led to the key story elements.

Learning to direct the viewer’s eye is essen-tially the study of composition. Composition is aterm which is used to collectively describe agroup of related visual principles. These prin-ciples are the criterion employed to evaluatethe effectiveness of an image. They are notrules to be followed, but define a structure bywhich to explore creative possibilities. They de-scribe a visual vocabulary, and provide meth-ods for breaking down a complex image intomanageable characteristics for subjective analy-sis. Besides being of interest to artists, theseprinciples are also an important aspect of visualperception and cognitive psychology research.

The seemingly simple act of placing lights canradically change the composition and focal pointof a shot. Good lighting can make a well-com-

Page 13: Pixel Cinematography

posed image stunning. It can also help rescue aless-than-perfect composition. The principles ofcomposition are the tools with which the light-ing designer can analyze a scene to devise waysto accentuate what is working and to mini-mize what is not. They are effective in bothstatic or moving scenes. Pauses in cameramoves and character poses are perfect op-portunities to evaluate a kinetic composi-tion using static techniques.

Rather than simply referring the reader atthis point to consult a book on composi-tion, a brief discussion of the primaryprinciples needed to the lighting designerare presented here. Although each prin-ciple relates to the others, they are presented inisolation for clarity.

3.1 Unity/HarmonyThe name of this principle suggests that theelements of the composition appear to belongtogether, relate to each other, and to otherwisevisually agree. Where other principles of com-position break down the image into spe-cific topics for study, the principle of unityreminds the artist to take a step back andlook at the image as a whole.

Although most artists rely on intuition todecide if a composition is working, thecognitive psychologists offer a somewhatless subjective alternative. They study theeye and brain processes that lead to theartist’s intuitive decisions. The cognitivepsychologists have developed the Gestalttheory to help explain our perceptual tenden-cies. The term Gestalt means “whole” or “pat-tern.” Gestaltists emphasize the importance oforganization and patterning in enabling theviewer to perceive the whole stimulus ratherthan discerning it only as discrete parts. Theypropose a set of laws of organization that re-flect how people perceive form. Without theseorganizational rules, our world would be visu-ally overwhelming. They include:

• The brain tends to group objects that are closeto each other into a larger unit. This is espe-cially true with objects which share proper-ties such as size, shape, color or value.

• Negative or empty spaces will likewise beorganized and grouped.

• Elements are divided into planes, such asforeground and background planes.

• Patterns or objects that continue in one direc-tion, even if interrupted by another pattern,are perceived as being continuous. The brainwants to perceive a finished or whole uniteven if there are gaps in it.

• The brain attempts to interpret the world byfinding constancies. If a person is familiarwith an object, he remembers its size, shapeand color and applies that memory when hesees that object in an unfamiliar environ-ment. This helps him to become familiarwith the new environment, instead of be-coming disoriented, by relating the objects inthe new environment to the known object.

P I X E L C I N E M A T O G R A P H Y 1 3

Page 14: Pixel Cinematography

By ignoring these principles, an artist risks cre-ating an image which challenges the eye to or-ganize it with little success. The viewer’s eyewill quickly tire and lose interest. Conversely,too much unity can be boring; if there is noth-ing to visually resolve, the eye will also quicklylose interest.

By understanding how the eye tends to groupobjects together, the lighting designer canhelp unify a disorganized or busy compositionwith careful shadow placement, or by minimiz-ing or emphasizing certain elements with lightand color.

3.2 EmphasisTo direct the viewer’s eye, an image needs apoint of emphasis, or focal point. An imagewithout emphasis is like wallpaper, the eyehas no particular place to look and no rewardfor having tried. Images which are lit with de-fault or uniform lighting similarly feel draband lifeless. By establishing the quantity,placement and intensities of focal points, thelighting designer directs the attention of theviewer by giving him something interesting tolook at, but without overwhelming the viewerwith too much of a good thing.

A composition may have more than one focalpoint, but one should dominate. The more com-plicated an image is, the more necessary pointsof emphasis are to help organize the elements.Introducing a focal point is not difficult, but itshould be created with some subtlety and asense of restraint. It must remain a part of theoverall design.

By first understanding what attracts the eye,the lighting designer can then devise methodsto minimize areas which distract the viewerby commanding unwanted attention, and in-stead create more emphasis in areas whichshould be getting the viewer’s attention.

1 4 P I X E L C I N E M A T O G R A P H Y

3.2.1 Emphasis Through ContrastsThe primary method for achieving emphasis isby establishing contrast. Contrast can beachieved with shape, size, color, texture, bright-ness or even motion. A focal point results whenone element differs significantly from other ele-ments. This difference interrupts the overall feel-ing or pattern, which automatically attracts theeye. With one dark dot among thirty bright ones,there is no question which dot gets noticed, thedark one, for two reasons: it has the most contrastwith its background, but also because it is theonly one of its type. Unique or minority elementswithin larger groups tend to attract our attention.

Contrast in value (brightness) is easy for the eyeto see, which is why black and white imagery issuccessful despite its lack of color. It also illus-trates why lighting is a major tool in the estab-lishment of emphasis and directing the eye ofthe viewer.

3.2.2 Emphasis Through TangentsTangents, where two edges just touch eachother, can produce a strong point of emphasis

Page 15: Pixel Cinematography

by creating visual tension. The eye is not com-fortable with tangent edges and wants tomove them apart. With care, tangents can becreated intentionally to attract viewer interest;however, most of the time they are accidentaland distracting. If a tangent is creating un-wanted emphasis, it is best to try to move oneof the shapes. It may be necessary to move anobject in the scene if it falls tangent to anotherobject. Another potential compositional prob-lem is when an edge of a shadow or light fallstangent with an object or other geometricedge. In this case, it is preferable to move theshadow or light to avoid the tangency.

3.2.3 Emphasis Through IsolationEmphasis by isolation is a variation of the Ge-stalt grouping concept. When an object defiesgrouping, by not being near or similar to anyother object, it calls attention to itself and be-comes a point of emphasis through tension.This tension is created by the feeling ofunpredictability caused by the lone element notbelonging to the group.

If this emphasis is undesirable, finding a wayto link it to the larger group may help mini-mize attention. Using an edge of a shadow topoint to the isolated element is one way tolink it to the group.

3.2.4 Emphasis Through AnglesA subtle form of emphasis can be achieved byusing perspective angles and other edgeswhich lead the eye to the focal point. However,they can just as easily lead the eye away fromthe intended subject. If perspective angles areleading the eye away from the focal point, it isnecessary to attract or contain the eye morestrongly using another method.

3.2.5 Emphasis Through ShapeThe brain tends to characterize shape as eitherrectilinear or curvilinear. Most images are notcomprised of strictly one or the other. By cre-ating an image with primarily one type, theother type becomes a point of emphasis. Inthe simple example to the right, the trianglestands out from the field of circles because ofits shape is unusual in this context.

As another example, a long straight shadowin an image with a lot of curves may need tohave less contrast or a softer edge than usualto keep it from drawing too much attention. Abusy shape among many simple ones, or viceversa, will also attract attention. This concept

P I X E L C I N E M A T O G R A P H Y 1 5

Page 16: Pixel Cinematography

may be helpful in recognizing why an objectmight be attracting more attention than other-wise expected.

3.2.6 Emphasis Through RecognitionBecause of the human need for self-recognition,human or anthropomorphic characters willnaturally attract more attention than inanimateobjects. Furthermore, in our attempt to recog-nize a character, we naturally are attracted tolook at his face, and especially to his eyes if he isspeaking, to see what he is thinking and feeling.

3.2.7 Emphasis Through MotionA static image has static points of emphasis andall principles of emphasis apply, but a movingimage has the added bonus of being able to cre-ate emphasis through motion. Camera motionand character acting are topics unto themselves(see [Lasseter87]), but it helps to understandwhen the eye is attracted to moving objects andwhen it is not. If all objects are moving exceptone, the eye will be drawn to the one which isnot moving. The opposite case, of only one ob-ject moving, is more common and even moreeffective in attracting attention.

3.3 BalanceWhen an object is unbalanced, it looks as thoughit will topple over. Instinctively the viewerwants to place it upright or straighten it. An un-balanced object is distracting and calls attentionto itself. An entire image which is off-balancewill make the viewer uncomfortable because hewants to balance it, but cannot. This discomfortcan be desirable if it enhances the mood orstorypoint. By knowing ways to balance or in-tentionally unbalance an image, the lighting de-signer can affect the mood of the scene.

A scale is balanced by putting equal weight onboth sides. It doesn’t matter how large or densethe objects placed on the scale are, they will bal-

1 6 P I X E L C I N E M A T O G R A P H Y

ance as long as they have equal weight. The bal-ancing of a composition is similar except that vi-sual interest becomes the unit of measure. Visualinterest comes in many shapes, sizes, values, col-ors and textures, each with varying density. Theprinciples of emphasis and balance are thereforerelated since points of emphasis carry visualweight which must be considered when evaluat-ing the balance of an image.

Visual balance is achieved using two equations.The first balances the image around a horizon-tal axis, where the two halves, top and bottom,should achieve a sense of equilibrium. Al-though it is desirable to have a sense of equaldistribution, because of gravity, the viewer isaccustomed to this horizontal axis being placedlower than the middle of the frame.

Besides helping to create a pleasing image, thetop/bottom weight ratio can also have astorytelling effect. The majority of constant fac-tors in our visual life experience tend to behorizontal in nature—the groundplane beneathour feet, the horizon in the distance, the sur-faces of water. Where these horizontal divisionsare, relative to where we are, tells us how tallwe are, how far off the ground we might be, orwhether we might bump our heads on some-thing. Because we are accustomed to makingthese comparisons, the placement of a characterwithin the image format and the angle that thecamera sees him can imply the height of a char-acter. And since we tend to associate height as adominating physical characteristic, it can saysomething about the importance of the charac-ter in his current situation. In one shot a shortcharacter is placed high in the frame, in thenext shot a tall character is placed lower in theframe. The shorter character in the first shotfeels taller and more important to us than thecharacter who is actually taller but is visuallysubservient. A character’s eyes are usuallyplaced above the center line, unless the charac-ter is looking up.

Page 17: Pixel Cinematography

The second equation of visual balance dividesthe image around a central vertical axis. Thehorizontal format of cinema is most affected bythis left/right ratio. And with the possibilitiesof action entering and exiting the frame, orcamera pans and dollies, this ratio has the po-tential to be very dynamic.

The simplest type of left/right balance is sym-metrical balance, where the two sides are mir-ror images of each other. Symmetrical balanceis discussed here primarily because it is easyto understand and to achieve. Heavily used inarchitecture, symmetrical balance feels veryformal, permanent, strong, calm and stable. Inother forms of art, perfect symmetry is rarelyseen. One distinct advantage of symmetry isthe immediate creation and emphasis of a fo-cal point. With two similar sides, there is an ob-vious visual importance to whatever element isplaced on the center axis. Another asset is itsability to easily organize busy, complex ele-ments into a coherent whole. In film, symmetri-cal balance is sometimes used to help portray aformal, official, or religious environment ormood. The Ingmar Bergman film “WinterLight” uses symmetrical balance to impart stiff,claustrophobic formality to the church settingin the opening sequence.

In contrast to symmetrical balance, asymmetricalbalance is more commonly used, more naturalin feeling, and much more challenging toachieve. Although asymmetry appears morecasual and less planned than symmetry, its vi-sual ease belies the difficulty in its creation.Balance must be achieved with dissimilar ele-ments by manipulating the visual interest ofeach. Some of the variables to manipulate arevalue, color, shape, texture, position and eyedirection. Each are discussed here individuallyfor clarity, but keep in mind that the interplayof these variables will affect the end result.Color can balance value, or texture can bal-ance shape, infinite combinations are possible.

3.3.1 Balance by ValueWe have already discussed that the eye is at-tracted to contrasts, particularly that a high con-trast area attracts more interest than one of lowcontrast. To balance the scale, a small area ofhigh contrast will command an equal amount ofattention as a large, low contrast area.

When it comes to projecting film in a theatre,the value scale isn’t necessarily level to beginwith. A theatre is dark to draw the viewers at-tention to the screen. In general, the eye is at-tracted to bright areas more than it is darkones, and in a dark theatre, with our pupils di-lated, a bright area will attract even more atten-tion since it contrasts with the darkness of thetheatre environment itself.

3.3.2 Balance by ColorLike value, color can be a balancing element.The eye is more attracted to a color than to aneutral image, the more saturated the color, themore attention it grabs. A small area of bright

P I X E L C I N E M A T O G R A P H Y 1 7

Page 18: Pixel Cinematography

color can balance a much larger area of aduller, more neutral color. How colors areplaced next to each other can also affect colorbalance. Complementary colors which areplaced adjacent to each other will weigh morethan two analogous colors would in the samesituation.

If a surface color is attracting too much atten-tion, its color saturation or diffuse light re-sponse may need to be toned down, or perhapsthe surface can be placed in shadow.

3.3.3 Balance by ShapeA large, simple shape can balance a smaller,more complicated one. Similarly, a large unclut-tered area can balance a smaller busy area con-taining many shapes.

Busy areas can be minimized by placing themin shadow or enhanced with directional lightplacement. Large simple areas can be brokenup with shadows or evenly lit depending onthe visual need.

3.3.4 Balance by TextureTexture and surface are similar to value, colorand shape. A busy, high contrast texture on asmall shape, will balance a larger shape witha smooth, matte surface.

1 8 P I X E L C I N E M A T O G R A P H


Texture is an invitation to the audience’s sense oftouch. Even though they cannot reach out andfeel the image, the textures in the image can trig-ger a sensory reaction. The surface quality of ob-jects can help define the mood of the scene,where soft, fuzzy objects summon a warmermemory than do smooth, polished objects. Sur-face texture also gives the audience cues as to

how close they are to the object surface, sincetexture becomes less apparent as the objectmoves farther away. A highly textured surfacewill attract more viewer interest than a non-textured surface, and these surfaces can be litto highlight or minimize their respective sur-face qualities. A highly textured surface can beaccentuated with hard light from the side orback, or it can be minimized with soft frontallighting.

3.3.5 Balance by PositionOn a scale, a heavy weight can be balanced to alighter one by moving the heavy weight closerto the scale center point, or by moving thelighter weight further away from the center.This principle is also true in composition. Alarge element placed close to the center of theimage can be balanced by a smaller elementplaced near the edge.

Page 19: Pixel Cinematography

Although it may not always be possible to movean object in the scene to balance the composi-tion, affecting an object’s visual weight can beachieved with lighting, either by changing thevisual weight of the object itself, or by counter-balancing it with more visual weight elsewhere.

3.3.6 Balance by Eye DirectionEvery edge, shape or group implies a directionof some kind, either straight or curved. It is thecounter-balancing of one direction with anotherwhich stabilizes an image.

Directional lines, such as perspective angles, area simple example of how eye direction can beused to balance a heavier side by pointing to-ward the lighter side, transferring visual impor-tance. Perspective angles are an example whichtend to be fairly subtle. A linear object, a shadowedge or the edge of a light can achieve a strongerdirectional effect. Eye direction also occurs whenthe eye is led from one side of an image to theopposite side when it finds the same color, orwhen the eye follows the gaze of a character.

3.3.7 Balance by Physical WeightThe perceived physical weight of objects in thescene also contributes to the sense of visualbalance. An object known to the viewer to beheavy in physical weight will impart morevisual weight than it would otherwise due toour tendency to factor mass and gravity intothe equation.





.4 Scale and Proportionhe term scale refers to overall size, whereasroportion is used to describe relative size. Scalend proportion principles help the viewer orga-ize an image. They can also be used to creater minimize points of emphasis.

cale and proportion are connected to empha-is and balance. If an object is out-of-scale or isddly proportioned, it can create too much em-hasis. Large scale elements, especially those

arge in proportion to the other elements, alsocreate obvious visual weight. The brain auto-matically reads larger objects to be nearer andmore important than smaller objects. As anexample, in a close-up shot of a character, theviewer is rarely distracted by the background.But in a wide shot containing many elements,the viewer must create relationships betweenelements to establish perspective and to find afocal point. In addition to mere size grabbingviewer attention, the brain will notice an ob-ject which appears out of proportion, or un-natural in size, compared with other objects.he lighting designer may need to minimize

arge objects which should not be the focus ofttention.

n a purely computer generated environment,he viewer works harder to establish and main-ain size relationships since there is no directomparison available to the real world. In “Toytory”, the architecture, furniture, props andharacters were modelled and textured withn appropriate amount of detail for their rela-ive size as well as to the distance they woulde viewed from the camera. Great care waslso taken in the staging and editing processeso avoid disorienting the viewer with unex-ected changes in apparent size relationships. consistent use of lenses for close-ups, me-ium shots and long shots helped to ensurepacial continuity.

P I X E L C I N E M A T O G R A P H Y 1 9

Page 20: Pixel Cinematography

Besides being useful for describing relativesize, proportion also pertains to the pleasingdesign of objects and their placement in theoverall image frame. Placing the focal pointdead-center in the frame is not usually a veryinteresting or pleasing placement. The eyenaturally prefers the focal point to be locatedwithin one of the four quadrants. This is espe-cially true if the image contains another strongpoint of emphasis. This is only a general rule ofthumb, however. Sometimes the lighting de-signer will be given a shot where the focalpoint is intended to be in the center of theframe. If the shot is a wide shot with many ele-ments, the eye may have difficulty remaining inthe center of the screen. It may be necessary tominimize other points of emphasis on the pe-riphery to keep the eye from wandering.

Much study has been devoted to deriving idealproportions for an individual entity as well ashow to portion an image in the most pleasingform. The use of aesthetic formulas has a longhistory. Even before the Renaissance, whenthere was more overlap between the arts andsciences, artists and mathematicians perfectedwhat they believed were the ideal proportionsfor human form, design and composition. Oneof the formulas they derived is the golden sec-tion*. Once used heavily in architecture, paint-ing and sculpture, it is described as “Tradi-tional proportion which is supposed to expressthe secret of visual harmony. In its simplestform it consists of a line divided into two sothat the smaller part is to the larger as thelarger is to the whole.” This proportion is ex-pressed in mathematics as a Fibonacci sequenceof numbers, or as a ratio roughly equivalent to8:13 (j). The golden section is frequently foundin nature and is very pleasing to the eye. Whendeciding where to place an edge of a light or

**p. 90; The Thames and Hudson Dictionary of Art Terms,by Edward Lucie-Smith: Thames and Hudson, 1984.

2 0 P I X E L C I N E M A T O G R A P H Y

shadow, formulas like the golden section maybe useful to keep in mind.

3.5 Repetition and Rhythm3.5.1 Repetition of ShapeThe use of similarly shaped elements in an im-age, however subtle, is a strongly unifyingforce, as a product of the Gestalt grouping prin-ciple. Repetition is an aspect of visual unitywhich is exhibited in some manner in every im-age. The human eye is very good at makingcomparisons and correcting minor differencesto equate two shapes as being essentially thesame, and then grouping them together as aunit or connecting them in some way. Becauseof this tendency, two similar objects, even atopposite sides of the image, will lead the eyefrom one shape to the other.

Also, the shape of an object can acquire morevisual weight if it is repeated, by its ownshadow for instance. By looking for ways to re-peat shapes with light and shadow, the lightingdesigner can help unify an image or redistrib-ute visual weight

3.5.2 RhythmVisual rhythms are plentiful in nature: a standof trees, a flock of birds, ripples in the sand,raindrops in a puddle. A person is so accus-tomed to natural rhythm that they feel it morethan they see it.

As a design principle, rhythm is based on rep-etition, although just because something re-peats itself doesn’t mean it has rhythm. A smallnumber of repeated or similar elements becomevisually grouped together to form a unit. Toachieve rhythm, a larger number of elementsare required, enough elements so as to discour-age grouping as a single unit, but several.Groupings of three or more start to introducerhythm, but only if they are not exactly thesame. Rhythm also requires variation within itsrepetitive groupings.

Page 21: Pixel Cinematography

Rhythm can be established using colors or tex-tures, but shapes and their arrangement aremore common repetitive elements. Rhythmicpatterns can help lead the eye through a com-position, but they can also evoke an emotionalresponse. A succession of curvilinear forms canbe calming, whereas a pattern of angular linesmay be stimulating.

3.6 Line and ContourThe human eye is very sensitive to change, andcan find even very subtle distinctions. It thenmentally connects these demarcations to createa line or boundary. The imagination and pastmemories then quickly fill in missing details inan effort to recognize these lines as a knownshape. In addition to filling in the missing bits,the brain also ignores visual stimuli which maybe competing for attention.

The careful placement and emphasis of edgesand linear elements play an important role inleading a viewer’s eye through a composition,directing it to the intended subject. A practicalexample of how lighting can be used to controlthe placement and emphasis of edges might bethe angle of a shadow, highlighting the edge ofan object, or even deliberately allowing theedge to blend with the background.

There are three types of edges or lines, an actualline, an implied line, and a psychic line. An ac-tual line is the easiest to recognize because itusually defines a shape. A line is implied bypositioning a series of points so that the eyetends automatically to connect them. A psychicline is not a line which is seen, but one which isfelt as a mental connection that exists betweentwo elements. The eyeline of a character is agood example. If a character looks toward anobject, the viewer will also look. These psychiclines are very powerful and can be difficult tominimize if they are distracting.

How and where a line is terminated can also af-fect its importance. A line which points to anobject, but doesn’t quite touch it will create ten-sion and attract attention more than one whichcontinues.

The boundaries of shapes and other linear ele-ments portion a composition for proportionalanalysis. They also have an emotional role.Horizontal lines imply stability, vertical linesimply potential motion, and diagonal lines im-ply dynamic motion and depth. When workingwithin the rectangular cinema format, horizon-tal and vertical lines work as stabilizers and re-duce feelings of movement since they mirrorthe format boundaries. A common camera tech-nique is to roll the camera to introduce a feelingof instability.

3.7 ShapeThe computer animation environment is threedimensional as it exists within the computer.Three dimensional objects move and deformfreely, changing shape and position, in theirthree dimensional world. And although thesculptural form and motion of the objects affecthow light is reflected and shadows are cast; ul-timately, it is the placement and definition ofthe resulting two dimensional shapes, withinthe image frame, that becomes the final prod-uct. Camera placement and lighting are whatcontrol this transition from the original designspace to the image the audience sees projectedon the screen. This may seem obvious, but italso appears to be forgotten at times.

A composition is primarily an arrangement ofshapes. The brain not only strives to recognizeshapes, it also attempts to organize them intofigure and ground relationships, or positive andnegative space. This happens on several levels.Just as the brain distinguishes between back-ground and foreground planes, it also looks forpositive and negative relationships within eachplane. The focal points and busy areas of theplane become the positive space, while theother areas become relief for the eye, or nega-

P I X E L C I N E M A T O G R A P H Y 2 1

Page 22: Pixel Cinematography

tive space. Negative spaces are not necessarilyempty flat areas, but they do not tend to attractattention. In a well crafted image, as much careis given to the shape and placement of thenegative spaces as is given to the subject itself.However, the shape of a negative area, particu-larly an enclosed one, can be so interesting thatit may take on a life of its own, attracting atten-tion instead of deflecting it.

A lighting designer is constantly balancing theneed for readability and the need for integra-tion. An image which has all of its shapesclearly defined is easy to understand. However,it is not as interesting as an image whereshapes fall in and out of definition, by falling inand out of light and shadow. Similarly, cleardefinition between foreground/background,and positive/negative space is easy to read, butis not a particularly interesting spatial solution.It is often desirable to blend together, or inte-grate, the spaces in some way to avoid theharsh juxtaposition of forms as is evident in abad matte. The use of a similar color or valuealong an edge can help the eye travel more eas-ily between the spaces.

An important concept for the lighting designerto keep in mind, is that the brain is very goodat recognizing shapes with a minimal amountof information, especially if this shape is al-ready familiar to the viewer. By just hinting at ashape with a minimal amount of light, theviewer’s imagination becomes engaged, and amood of mystery and suspense is evoked. Thisis a concept that is apparently foreign to adver-tising agencies who want to see the entire prod-uct label evenly lit.

Shape distortion can be a powerful emotionaltool. The viewer is so accustomed to seeing theworld in a natural fashion that when shape isdistorted in an image, it signals an altered stateof reality. An emotional response will rangewidely depending on the shape being distortedand its context. The baby in “Tin Toy” is dis-

2 2 P I X E L C I N E M A T O G R A P H Y

torted, using refraction through a cellophanewrapper, with comic relief to the plight ofTinny. In another context the same techniquemay be eerie and unsettling. The individualparts of the mutant toys in “Toy Story” are notthemselves distorted, but in combination theyrepresent a distorted vision of a life-like toy.The combined effect is disturbing and repul-sive, which helps us believe that they may in-deed be cannibals.

3.8 ValueA black and white image can often work aswell as a full color image because enough vi-sual information exists for the viewer’s imagi-nation to fill in the missing color information.In fact, a black and white image can sometimesbe more powerful than color precisely becauseit requires the use of imagination.

Contrast in value provides spatial cues fordepth perception. Areas of greater contrast ap-pear to advance into the foreground, while ar-eas of lesser contrast tend to recede. The angleand direction of a shadow helps define the lo-cation and orientation of the surface on whichit falls. If a shadow does not fall as expectedfrom its source, the result can be disorienting asthe viewer tries to reconcile the surface plane tothe shadow.

Lighting is ultimately how the values of thescene are controlled as the camera will seethem. In live-action, the aperture of the cameralens will also affect the final result by selectinga tonal range within the capabilities of the filmstock.

The interplay of light and shadow and the rela-tionships of tonal values is a major contributorto the style and mood of the scene. This is dis-cussed in more detail in Section 4.

3.9 ColorValue and color are related to each other sincethe light which falls on reflective surfaces, orshines through translucent materials, producesvarious brightnesses. On black and white film

Page 23: Pixel Cinematography

they are reproduced as gray values. On colorfilm, the apparent brightness is greatly influ-enced by the hue and saturation of the colors,but the final outcome is still a range of values.Every color has a value, but color, which isbased on wavelengths of light, offers a muchbroader field of visual differences and contrasts.

The color of a surface is determined by how itreflects the light that illuminates it. The appar-ent color of a surface depends upon the light-ing situation. Unfamiliar objects appear just asthe eye perceives them, that is the apparentcolor and value are determined by the actualwavelength of the reflected light. For familiarobjects, the principle of color and brightness con-stancy takes effect. Here the brain uses previousexperience to augment the strictly physical per-ception of the eye. If the color of a familiar ob-ject differs from that in memory, the brain as-sumes that the color of the object is affected byits environment. For example, if the viewer seesa purple apple, chances are they have neverseen an actual purple apple, and will assumethey are viewing a red apple as seen under bluelighting or through a blue filter.

A color is also perceived as a certain hue, satu-ration and brightness as it relates to the colornext to it. A color on a neutral background mayappear very different than it would in contextwith other colors. Similarly, two complemen-tary colors, when juxtaposed will accentuateeach other and appear more intense than theywould if either were placed adjacent to ananalogous color. Neutral colors can be heavilyinfluenced by a stronger color next to them,where the neutral color will tend to go towarda hue which contrasts the strong color. In otherwords, a grey square next to a red one will tendto go a little greenish.

Color can play a big part in visual storytelling,both in terms of the set, props, and wardrobedesign, but also in the lighting. Section 4.7 con-tains a brief discussion on color and our emo-tional responses to it. An exhaustive discussionof color, however worthy, is beyond the scope

of this course. Many books on color theory andthe psychology of color exist for the reader whois interested in exploring these subjects in fur-ther detail.

3.10 Compositional MoodA pleasing composition evokes a sense of well-being, a feeling that everything is happy andgoing to stay that way. A composition which isa little unbalanced, or otherwise feels awk-ward, can create a feeling of tension and appre-hension. This feeling can be useful if the intentis to build story tension or to portray the emo-tional state of a character. A progressive build-ing of visual tension can foretell that somethingbad is going to happen whether it actually doesor not. A sudden change in visual tension canaccentuate the shock of a dramatic change.Sometimes the composition and lighting designwill intentionally be in contradiction with thesubject matter. Soft, warm, beautiful lightingcan be used to light a violent, ugly subject mat-ter. This contradiction can aid in viewer dis-comfort because it feels especially out of con-text and shocking.

Whether or not the composition is busy orsimple will also have some emotional impacton the viewer. A well-composed simple scenewill feel elegant and noticeably beautiful com-pared to a cluttered scene. An element in asimple scene will also feel more important thanthe same element in a busy scene.

3.11 TransitionsWhile it is necessary to direct the viewer’s eyeto the intended subject within a shot, it mayalso be desirable to “lead” the viewer’s eye intothe next shot. An effective transitional device isto use lighting and compositional elementsfrom the current shot to transition into similarelements in the next shot. The audience is al-ready visually prepared for the next shot beforethe cut or dissolve occurs.

P I X E L C I N E M A T O G R A P H Y 2 3

Page 24: Pixel Cinematography

2 4 P I X E L C I N E M A T O G R A P H Y

3.12 Putting it TogetherEven with the intent of discussing the prin-ciples of composition as isolated topics, it iseasy to see how they interrelate. Figures 3.12a,3.12b and 3.12c illustrate a few examples of us-ing a few of these techniques together to directthe attention from one character to another. Fig-ure 3.12a shows a simple shot of Woody andBuzz over a plain grey background. The samebasic lighting is applied to both Buzz andWoody, the direction of which is arbitrary. Ifyou were to guess which character is supposedto hold our attention, which would youchoose? Initially we may think that it is sup-posed to be Buzz because Woody is looking athim which naturally draws our attention toBuzz. Ordinarily we would consult with the di-rector to find out his intention, but for the pur-poses of our example, we will light one to high-light Buzz and then relight it to focus our atten-tion on Woody. In these examples, only thebackground has been relit, because the result isclear to see.

Figure 3.12b uses several techniques to helpdraw our attention to Buzz. A shadow line wasintroduced on the background to reinforce theeyeline from Woody to Buzz. It is sharper be-hind Buzz and softens as it reaches Woody toprovide more contrast near Buzz. The contrastof the shadow line is low enough so that it doesnot attract attention to itself. This dark shadowarea behind Buzz allows him to “pop” from thebackground while Woody tends to blend intoit. Also, this shadow line is positioned so that itis almost tangent with Buzz’s head to add ten-sion to that area. Woody does not have a rimlight to aide in his blending into the back-ground, while Buzz has a nice bright one whichhelps him separate.

Figure 3.12c shows the shift of attention towardWoody. For this image, we will use the proxim-ity of the grey wall to help us. The shadow ofWoody on the wall provides us with more con-trast areas around Woody’s head and the rep-etition of his shape on the wall gives him morevisual weight and drama. The wall behind

Buzz is lit to help him blend into it. The rimlight on Buzz has been toned down while therim on Woody has been brightened. As you cansee in this image, the psychic line created byWoody looking at Buzz is very difficult to over-come.

Exaggeration is a familiar device used by char-acter animators to enhance the life-like qualityof a character. Purely natural or physically cor-rect lighting is often not interesting enough tocreate drama and captivate the audience. Push-ing the limits of reality can create magic andbeauty that connects the imagination with thestory being told.

It would be possible to fill volumes with ex-amples, but let’s move on. The next five light-ing objectives draw upon our knowledge ofthese principles of composition and illustratefurther how they can be used in lighting forstorytelling.

4. Enhancing mood,atmosphere and drama

Many aspects of an image affect its mood anddramatic qualities. The sets; the costumes; theactors and their acting; the staging; the score;the weather; the time of day; and the lighting,are all components which can illustrate themood of the story being told. Lighting designcan pull it all together into a cohesive mood, orit can provide a contradictory mood of its ownfor the sake of contrast.

The emotional role of lighting is not always ob-vious, even though it is profound. This emo-tional effect can be accentuated by juxtaposingcomplementary lighting situations. An overcastday will evoke vague feelings of oppressionand melancholy, but might not be noticeableuntil it is compared with the uplifting qualityfelt at the first brightness of a spring sun. Itprobably wouldn’t be appropriate to light a sadand gloomy scene with lots of bright light, nor

Page 25: Pixel Cinematography

would it seem right to light a comedy withharsh high-contrast light.

Lighting vocabulary is based on studio light-ing. This long-established vocabulary breaksdown the infinite possibilities of light into man-ageable topics for discussion. It is therefore use-ful to take a look at the studio method of ex-plaining lighting design and relate it to oursynthetic tools.

The establishment of mood and drama throughlighting is the sum of the properties of thelights themselves as their motivation, purpose,placement, direction, range, color, quality,quantity and intensity. This section is primarilydevoted to the discussion of these light proper-ties, how they relate to each other and howthey influence the emotional qualities of thescene. Each property is presented in the logicalorder that lighting decisions are usually made.

Often, the first decision to be made is to deter-mine the style of lighting to be used for thescene.

4.1 Lighting StyleAn infinite number of combinations of lightingproperties can be created for a wide range of vi-sual and emotional effects. In order to simplifythe endless possibilities, some generalizationshave been drawn which are used to label anddescribe lighting in terms of style. Lightingstyles are described by their tonal range, whichis the range of values from the darkest dark tothe brightest highlight and the grey values inbetween. Lighting styles are also described interms of the overall color, motivation, place-ment, and quality of the lights and shadows.

The character and mood of an image is dramati-cally affected by the range of tone values fromlight to dark and by their distribution within theframe. This tonal range is decided early in thelighting process. This decision is usually moti-vated by the dramatic quality of the story andcan be consistent throughout the entire movie orvary with the location and time of day.

A light-hearted or comedic story might dictatea high-key lighting style. High-key lighting ischaracterized by a scene that is mostly well litwith a lot of soft fill light and no heavy or hardshadows. The sets and costumes also tend to belight in color. This doesn’t mean that therearen’t any dark areas, but the overall brightnesstends to be light, contrast is low, and the darkareas are soft and few. The result minimizessuspense since nothing is left to the imagina-tion of the audience.

At the other end of the spectrum is low-keylighting. In a low-key lighting situation, mostof the scene is darkly lit, with the emphasis onthe few areas which are brightly lit. The setsand costumes are also usually dark in color.The overall impression is dark but not murky.What is seen is equally important to what is notseen. The detail only hinted at is much richerthan it would be if it were well lit. Light is usedto direct the viewer’s attention, the darkness tostimulate his imagination. Of course, these arethe polar opposites, with many possible tonalranges in between.

Aside from the overall brightness or darknessof the style, its contrast range can evoke moodand meaning. Unlike a low-key scene wheremost of the frame is dark, high-contrast scenescontain a wide range of light and dark areaswith a narrow middle range of greys. A high-contrast image, with many hard edges of lightand shadow, has a dramatic graphic qualityand can evoke a sense of energy or unrest. Alow-contrast image, composed of a range ofshades of middle tonality, can convey a feelingof calmness or bleak oppression.

Even before the viewer has understood thestory-point, the lighting style can suggest afeeling for a scene, especially in comparisonwith adjacent scenes. Or within a single shot,one character may be modeled in bright tonesand another in shadows and dark tones to sug-gest their individual personalities or their emo-tional or dramatic situations.

P I X E L C I N E M A T O G R A P H Y 2 5

Page 26: Pixel Cinematography

2 6 P I X E L C I N E M A T O G R A P H Y

4.2 Quality of LightThe creation of varying degrees of softness anddirectionality are important aspects in creatingmood through lighting. In addition to consider-ing the tonality of image, lighting style is alsooften defined by the quality of the lights, espe-cially the key source. The quality of a light iscomprised of three characteristics, the primaryone being its hardness or softness, with theother two being its angle of throw, and its color.A soft source is diffused which scatters light inmany directions and creates very soft shadows,whereas a hard source is not diffused and castsvery crisp shadows. A light source, even a softone, will become harder as it moves fartheraway from the subject. The apparent size of thesource becomes smaller, and as its rays becomemore parallel, causing its highlights and shad-ows to become more crisp.

In addition to the actual hardness or softness ofthe sources themselves, the contrast range ofthe resulting image also contributes to the over-all feeling of hard or soft lighting. Subjects oflimited tonal range, with middle tones of gray,appear softer than subjects with deep blacksand brilliant whites.

The quality, or character, of light will initially beinfluenced by the scene location and the time ofday but may vary to accentuate mood or story-points. Daylight scenes are usually softly lit ex-cept for direct sunlight. Night scenes, especiallyexteriors, tend to be lit with harder lights. Thecharacter of a light is also heavily influenced bythe motivation of its source.

4.3 Lighting MotivationOnce the character of light is chosen for a givenscene, the next task of the lighting designer isto decide the practical and hypothetical sourcesof light and their orientation to the set and sub-ject. These decisions will be influenced by thescript, the set, and the camera locations withinthe scene. Lights are characterized as being ei-ther logical or pictorial. A light is logical if it ap-pears to be motivated by an actual source oflight (practical source) that the viewer can see

or is implied, such as a window or table lamp.Logical lighting, also called naturalistic, moti-vated, or method lighting, generally follows thenatural, logically established visible sources ina scene. On the other hand, pictorial lightinggenerally uses lighting directions simply be-cause they produce a pleasing picture.

Most of the time, there is a compromise be-tween the logic of the source and the composi-tional requirements of the frame. Sometimesthe light direction is established by what feelsnatural, even if the logic of the source is slightlyviolated. It is the overall character of the light,its color, softness and direction, that matters.The exact angle and intensity of the light willnever be scrutinized by the audience as long asit is not disorienting.

Practical sources which are visible to the viewerneed to be well placed. If there is a visiblesource of light within the image frame, theviewer expects the overall light direction toemanate from the source they see, even if thelighting originated from a different source inthe previous shot.

4.4 Quantity of LightsThe number of logical sources chosen will alsohelp determine the mood of the scene. A softone-light scene, for instance from a candle, canfeel very warm and romantic. A big bank offluorescent lights can feel sterile, cold and over-lit. There are many possibilities in between, butin general, the number of logical sources areusually kept relatively few to be able to estab-lish overall direction.

The number of logical sources may be smalleven though the actual number of lights usedto achieve a look may be many. This is true inlive-action and even more so in synthetic light-ing since no ambient or bounce lighting comesfor free (unless of course a radiosity renderer isused). In general a light should not be added toa scene without an intended purpose, and thetemptation to use one light to serve two pur-poses should be avoided.

Page 27: Pixel Cinematography

4.5 Quality and Quantity of ShadowsThe number and quality of light sources also helpdetermine the number and quality of shadows.This is one area where synthetic lighting has agreat advantage over live-action. In a live-actionsituation, each hard light casts a distinct shadow.Great care is taken to minimize multiple shadowsand to establish one dominant shadow. Con-versely, in synthetic lighting, it takes effort to cre-ate multiple shadows and penumbra effects, andthe tendency is to overdo it. One shadow shouldstill dominate, but it is usually necessary to intro-duce a secondary or contact shadow to helpground a character onto the shadowed surface.

Also in live-action situations, the quality of theshadow is determined by the placement andquality of the light. A hard or distant light willcast a crisp shadow. The softest light will notcast any shadow. The density of the shadow isdetermined by the amount of bounce and filllight in the scene. In synthetic lighting, shadowdirection, quality, color and density controls canbe independent of the light attributes, but theyshould still feel naturally motivated by thesources in the scene.

4.6 Light Type, Purpose, Placement,Direction and IntensityThe desired tonal range of the image, and qual-ity and motivation of illumination have nowbeen established. The next task to consider is thespecific type, purpose, placement, direction andintensity of each light.

4.6.1 Types of LightsThree basic types of lights are used for live-ac-tion lighting. The spotlight, the floodlight, andthe area light. The spotlight has a narrow beamand is usually used as a hard light source. Itcasts crisp shadows and a crisp bright highlight.The floodlight has a broader beam and is usu-ally diffused and used as a softer source. It castsa soft shadow and a broader diffused highlight.An area light is either fully diffused through adiffusion material or is bounced off another sur-face such as a white card. It casts very faintshadows, if any.

P I X E L C I N E M A T O G R A P H Y 2 7

The use of diffusion and focusing materials cre-ates a wide range between a very soft scatteredlight and a very hard directional light. Lightplacement also affects the apparent hardness ofthe light, since even a soft light will appearharder as it moves farther away from its sub-ject, as its rays become more parallel. Thesethree types of studio lights are designed toemulate nature. Sunlight is an example of par-allel rays which cast very crisp shadows. Anovercast sky is an example of very diffusedlight casting soft shadows. And the shadow ar-eas under a clear blue sky is an example of abig area source which casts faint shadows.

The synthetic lighting designer attempts to rec-reate these real light properties with a differentset of tools. Many renderers currently supportall three basic light types. Spotlights (point)and solar (parallel, infinite, or distant) lights arepretty standard stuff, area lights are more diffi-cult to find since they are morecomputationally expensive.

Without the availability of area sources, theoverall soft look can be emulated using morestandard techniques. The parallel rays of a so-lar light have a softer feel than a spotlight be-cause the illumination is more evenly distrib-uted over a surface. And since the shadows canbe controlled independently, a blurred shadowwith partial density can help this cheat. Thespecular quantity and quality from a sourcealso describes its apparent softness. By using anenvironment map, with blurring and color fil-tering controls, to define the highlights, thecheat can be taken a step further.

Many renderers support control for diffusingthe specular highlight for each source indepen-dent from the surface roughness, but specularhighlights are really just a cheat. It is oftenmore convincing to use an environment mapwhich contains representations of the sourcesinstead. Gratuitous specular highlights imparta plastic, computer-generated quality; high-lights should be placed, shaped and coloredwith care.

Page 28: Pixel Cinematography

The quality of a light’s shadow provides themost information in describing the quality ofthe source. In computer lighting, much controland creative freedom can exist for manipulat-ing shadows. The color of a shadow can be en-hanced, and the density and softness of ashadow can be set uniformly or be variedacross its surface.

The softness of the shadow suggests the soft-ness of the source as well as the distance of thesource from the subject. But, even though ahard distant light will cast a hard shadow, soft-ening the shadow can suggest atmospheric dif-fusion. The softness of the shadow also gives usvisual cues about the distance from theshadow-casting object to the surface on whichthe shadow falls. The nearer the subject to theshadowed surface, the more dark and crisp itsshadow will appear.

4.6.2 Light Functions and PlacementThe function of a light is independent of itstype, its quality and even its placement. Alight’s function is particularly meaningful fordescribing how it is used on a subject. For thisreason, light function and placement are dis-cussed here together. Ambient—or base lightingThe overall brightness of the shadow areas isusually determined first by the use of baselighting. In live-action this might be achievedby suspending a white cloth over the set andtop-lighting it, bathing the entire set in a washof light. In the computer this is accomplishedby using a combination of an ambient light anddiffuse lights. The ambient light adds a slightamount of light everywhere to prevent any ab-solute black areas, but is extremely flat. The useof a few diffuse lights can add a little bit of di-rectionality to ambient base lighting. Aradiosity renderer eliminates the need for add-ing a flat ambient light, but does not necessar-ily eliminate the need for base lighting.

2 8 P I X E L C I N E M A T O G R A P H Y Key light—for modeling of surfaceand setting of moodThe key light is the brightest light striking asubject. It defines the brightness and shape andtexture of the subject.

As the dominant source, the placement, color,intensity, and textural quality of the key lightare important attributes in setting the mood fora scene. But it is the placement of this light thatmost affects the mood and dramatic quality ofthe image by controlling the direction of thelight as it strikes the subject. The direction ofthe light can vary the apparent shape and vol-ume of the subject, by accentuating or minimiz-ing certain features. This is referred to, in light-ing terms, as surface modelling. The characterof this modelling is also affected by the softnessof the light and its shadows.

Although there are no hard and fast rules for theplacement of the key light, it is conventionallyplaced 30-45 degrees to the side and above thesubject relative to the camera axes. However,this light can be effectively placed as far back as135 degrees from the camera as a three-quarters-back key light. Another convention is to place thekey light so that it is on the opposite side of theactor’s eyeline from the camera. These conven-tions are interesting, but only serve as a looseguideline since the direction of light is usuallydictated by the relationship of the subject to themotivation of the source, the chosen style oflighting, and the mood of the scene.

By controlling the direction and quality of thekey light, it is possible to change the appear-ance of the subject as well as to suggest some-thing about the subject’s personality or dra-matic situation. A beauty-shot of the heroinemay have a softer, more frontal key light thanthe villain who is chasing her.

In live-action lighting, the addition of a fillerlight is often added to fill in the key light shad-ows. This light is placed near the key light, buta little lower. On the computer this light is not

Page 29: Pixel Cinematography

usually added since the same effect can beachieved by lessening the density of the keylight shadow. Fill light—for subject contrast controlA fill light is a low-intensity, diffuse light usedto fill in shadow areas. This light does not callattention to itself. In pure terms, it does not castnoticeable shadows, nor does it produce a no-ticeable or sharp specular highlight.

Although a fill light can be placed almost any-where, it is traditionally placed nearer to thecamera axes than the key light. Since the filllight is often near the camera, it tends to fill inthe key light shadows and reduce the surfacemodelling created by the key light.

The ratio of the key light plus the fill light tothe fill light alone is called the lighting ratio andis one way of controlling the contrast range ofthe scene. In a high-key lighting situation, a lotof soft fill light is used to bring up the overalllevel of illumination. In low-key lighting situa-tions, the fill light is frequently omitted. Backlight— for separation frombackground and setting of moodAlso referred to as rim, hair, or separationlights, backlights are traditionally used in blackand white cinematography for foregroundseparation. In color cinema they are needed lessfor separation, but they are also effective forcreating a romantic mood.

A true backlight is traditionally placed behindthe subject so that it is pointing directly at thecamera resulting in a thin rim of light aroundthe edge of the subject. They are also placed athigher angles to highlight hair and clothing.Back cross-lighting is frequently used to put arim on both sides of the subject. A soft back-light can look natural even if it has no obviousmotivation. A hard backlight, unless it is moti-vated by direct light, will look unnatural, butthey are still often used anyway. Backlightingshould be used with thought since it tends tobe overused.

Backlighting is easy to achieve in live action, of-ten with one light. However, many shading al-gorithms ignore light which strikes the back ofan object and do not try to simulate the halo ef-fect that results from backlight hitting a surfacewhich is not mirror smooth. Without specialshaders which comprehend backlighting, creat-ing this effect in CGI requires a bit of cheatingand patience. It helps to have roughness con-trols on the specular highlight of each lightsource, so that the backlights can have as broada highlight as possible. Exact light placementcan be tricky to control especially with movingor multiple characters, and multiple sources. Ifa localized effect is desired, a spotlight may beeasier to control than a solar light which ismore effective over a broad area. Consistentbacklighting is also tricky to achieve in a wide-angle shot, especially if a character is movingacross the frame. It may be necessary to ani-mate backlights in these situations. Kicker—for surface modelingand character enhancementA kicker light is an optional light usually used todefine the non-key edge of a subject. This lighttypically works from a three-quarters-back po-sition and is placed nearer to the floor than thebacklight. This light can be soft and diffuse orhard and specular, depending on need and theintended lighting style.

A kicker light is a more general name for a vari-ety of lights which perform slightly differentfunctions. The three main types are a kicker, aliner, and a glow light. When used to create asheen (specular light), on a cheek for instance,they are frequently referred to as a kicker light.When far enough forward to contribute morediffuse light, it is sometimes referred to as aliner light. A glow light is a little farther for-ward still and is softer, non-specular, andshadowless. On people, it is usually desirableto avoid having a hot kicker light hit the tip ofthe nose.

P I X E L C I N E M A T O G R A P H Y 2 9

Page 30: Pixel Cinematography Specials—to accent an area for eithersubject or backgroundA special is any type of light that is used forbackground and set lighting, or for highlight-ing an area on the subject. Bounce Lights—to simulateradiosity effectsIn computer graphics lighting, unless aradiosity renderer is used, extra lights usuallyneed to be added to simulate the light that nor-mally bounces off nearby surfaces. These lightsare usually localized, non-specular, low-inten-sity and colored to mimic to the surface theyare reflecting.

4.7 The Color of Light4.7.1 Color PalettesA lighting style is described as a chosen tonalrange, but it also includes a color style as well.Color style is often discussed in terms of pal-ette, consisting of hues and tones. In order toset a style, a fairly small selection of colors arechosen according to how they relate to eachother. This selection, or palette, may consist ofcomplementary colors, analogous colors or an-other of an infinite variety of combinations. Thesets and costumes already have an establishedpalette which the lighting designer may try toaccentuate or minimize.

Naturalistic lighting mimics the complemen-tary palette found in nature. The range is fromyellow/purple to orange/blue to red-orange/blue-green in varying degrees of saturation. Fora daytime scene, the key light is warm, simulat-ing the sun, while the fill light is cool, simulat-ing the natural fill of blue sky. A nighttimescene might reverse this sense with a strongblue key light acting as moon light with a softwarm fill emanating from incandescent indoorlighting. The eye is accustomed to seeing thiswarm-cool relationship in a wide range of colorintensities. The contrast between warm andcool is minimized during the early to middlepart of the day, and grows as the day nears

3 0 P I X E L C I N E M A T O G R A P H Y

dusk as dust particles in the atmosphere filterthe color of the light A natural feeling still canbe maintained even when using a strongly col-ored light which falls outside of this naturalpalette, as long as it appears to emanate from avisible practical source.

The similarity or contrast between lightinghues and saturation can help suggest the moodof the scene. Scenes which are lit with similarcolors tend to be more somber than sceneswhich use extremes. The color of individual ob-jects, sets and costuming evoke emotional re-sponses of their own. The combination of theseelements into a whole image also presents anoverall color for emotional consideration.Lighting can be used to accentuate or minimizeindividual areas of color as well as setting thetone for the overall scene.

The placement and intensities of the lights alsohave an effect on the overall color. A low-key,almost black and white effect, can be achievedby minimizing object color saturation with theuse of strong directional lighting. The emphasisfalls on the shapes of objects rather than theirsurface colors.

Many new lighting designers are intimidatedby using and mixing brightly colored lightsources because they don’t do what the de-signer expects. Most people have had experi-ence with subtractive color theory since they havemixed colors together as paint or other pig-ments. It is called subtractive because if youmix the three primaries (red, green, blue) to-gether, the result would be black. If you mix thethree primaries of light together, the result iswhite light. This is called additive color theory. Aquick study of additive color theory wouldgive the new lighting designer confidence toexplore the mixing of colored lights.

4.7.2 Color and EmotionsEarly man’s use of color was largely symbolicand emotional, based in mysticism and reli-gion, and not necessarily chosen for aestheticreasons. The palette for a culture was estab-

Page 31: Pixel Cinematography

P I X E L C I N E M A T O G R A P H Y 3 1

lished and adhered to within that culture, andwas used to identify themselves by dynasty,race or tribe. Not until the Renaissance wascolor appreciated as an aesthetic choice.

Colors evoke physiological, psychological andemotional responses. These responses are a re-action to associations we make with our pastexperiences and cultural heritage. Two peoplecan have very different reactions to the samecolor, and one person can have a varied reac-tions to the same color depending upon its con-text. Even so, there are enough common life ex-periences and contexts within which to drawsome generalizations about how color affects usemotionally, especially in American culturewhere many of them have been stereotypicallyreinforced by advertising.

Colors are often referred to as being warm, coolor neutral. Warm colors are generally agreed tobe those which fall within the red-orange-yel-low spectrum, and cool colors to be within thegreen-blue-violet range. Neutral colors arethose which are near grey in saturation value.Cool hues tend to recede and induce calm.Warm hues stimulate the nervous system andraise the heartbeat. Pure, saturated colors tendto advance and excite, while duller neutral col-ors tend to recede into the background.

Specific colors evoke more specific associationsand responses. Red, for example, is an emotion-ally charged color which has many associa-tions: anger, passion, fire, blood, violence, sun-set, sex, adultery, aggression, power, creativity,embarrassment, and courage. It is also used asa universal symbol to stop or to denote whenan error is encountered.

Green recalls calmer memories: nature, water,trees, mountains, meadows. It is an introspec-tive, reserved color which evokes feelings ofsecurity, constancy, normalcy, balance, civility,convention. It is a suburban color for activehealthy people. It is the color of money. Greenis generally a positive color, although itdoes have negative associations, we have all

heard the expression “green with envy”. Greenlighting can look eerie, chemical, artificial andunhealthy.

Blue can feel heavenly and religious and is as-sociated with Western culture weddings. Itfeels spacious as it reminds us of the sky andoceans. It is a rational, conservative color whichsymbolizes authority, loyalty, order, peace, con-formity, success, caution, and patience. Bluelighting can look gloomy, electric, and cold ifthere is no warm light to counterbalance it.

Violet and purple have been associated withroyalty since the Roman Empire when only theemperor was allowed to wear it. It can feelmagical, exotic, sensitive, sophisticated, idealis-tic, and cultured. Violet lighting in shadow ar-eas can be very beautiful.

Yellow feels sunny, happy, and reminds us ofsummer days and flowers. It is also associatedwith intellect, wisdom, timidity, cowardice, andhunger. Yellow lighting is associated with mid-day and interior settings.

Orange is the social color, full of fun and cheer-fulness. It is urban and outgoing. It has also re-cently become known as the safety and con-struction color due to its visibility. Orangelighting is associated with evenings, and inte-rior lighting.

Brown is a homey and down-to-earth color, fullof duty and responsibility. It is often associatedwith poverty and the lower class and is easilydisliked. It is also associated with the pastsince objects tend to turn brown with time andexposure.

Pink packs more punch than other pastel col-ors. It can immediately portray someone asfeminine, silly, delicate, floral, pampered, ten-der, healthy, wealthy, vain, and indulgent.

Page 32: Pixel Cinematography

Black can look formal, elegant, and sleek. It canfeel evil, empty, mysterious, anxious andfearful. It is associated with night, death, andinevitability.

White can feel pure, virginal, innocent, classi-cal, and youthful; but it can also feel sterile andemotionless. White lighting is associated withhigh afternoon sunshine and daylight interiors.

Grey is the color of oppression and isolation. Itcan feel institutional, indifferent, sad, cold andheartless.

A person’s response to a color is immediate, butis usually short-lived. After continued exposureto a color, the effect wears off or sometimes evenreverses itself. It is the change from one color toanother which triggers an acute response.

Researchers who study human response to colorhave established that people remember skintones as being warmer or pinker than they reallyare. Human skin (real or computer generated) ismore appealing in warm light and we like to re-member it that way. Films are usually lit andcolor-corrected during printing to make skintones look “rosy”, and in general, films are usu-ally color-corrected for the skin tones rather thanfor other colors or objects in the scene. Overallskin tones which are colored more realisticallytend to give an image a documentary feel.

4.7.3 An Example Palette and UsageThe movie “Crimson Tide” is a terrific exampleto illustrate these points because it takes themto the extreme. The chosen color palette is puresaturated primary colors, red, blue, yellow,with the addition of pure green. The set andcostumes are dominated by neutral tones withaccents of these primaries. The main part of thesubmarine is lit with nearly neutral light aswell. The overall intensity is fairly dark, withhotter rim lights, giving it an enclosed, sub-merged feeling. The intensely colored lights areused to place the viewer in different locations

3 2 P I X E L C I N E M A T O G R A P H Y

within the sub. Green light is used near the so-nar screens, blue light in the transition areas,red light in the weapons room, and yellow haz-ard lights flashing everywhere to heighten thesense of urgency. The brightly colored scenesintercut with the neutral ones, which can besomewhat jarring, but you never get confusedabout where the shot is taking place (even withthe sound turned off watching it for the firsttime). This intercutting also tends to heightentension, mimicking the attitude of the flashinghazard lights.

4.8 Shaping and Controlling LightOf equal importance to the placement and di-rection of light, is shaping and controlling it, il-luminating the intended subject without spill-ing into unwanted areas. It is sometimes alsodesirable to create a defined light shape, eitherto mimic a logical source, or for dramatic orcompositional purposes.

Many of the techniques used to control andshape light synthetically can be similar to thoseof live-action because the problems are alsosimilar. These problems include controlling thelight’s size, shape, distribution, isolation, andcoverage over distance.

It is desirable to break up large or even surfaceswith varying light and shadow. Sometimes thiscan be achieved with deliberate, recognizableshadows from actors or props and sets, othertimes a more subtle overall variation will be ap-propriate. Barndoors, the four flaps attached tothe sides of a light, can be used to shape andtrim a source. The use of freestanding flags toblock lights, can be used to shape light at vari-ous distances from the source, which allowscontrol over the hardness of softness of theshaping. A cucaloris (or cookie), a rigid boardwith squiggly holes cut into it, can be used infront of a light to create a more random, or-ganic light pattern.

A good computer lighting toolkit will offer avariety of light shapes along with sizing andsoft barndoor controls. Further light shaping

Page 33: Pixel Cinematography

can be accomplished with the use of blockerflags, slides and cookies, as well as light attenu-ation over distance.

Computer lighting has several benefits whichdo not exist in live-action. Lights and flags canexist anywhere in space without interferingwith the camera or subject. It is also possible toilluminate a subject and not have the light af-fect other characters or the background, nega-tive intensity lights can used to softly subtractillumination and shadows can be indepen-dently colored, blurred, and given an arbitrarydensity value. The apparent softness of a lightcan be independent of its distance from thescene, and resulting shadows do not necessar-ily have to be projected from the source pointof the light.

Shadow quality and shadow placement to re-move light from certain areas are methods forcontrolling light, creating emphasis, and estab-lishing mood. The “Film Noir” lighting style of-ten used superlative examples for controllinglight to establish mood and influence composi-tion. The film “Mildred Pierce” offers wonder-ful examples of not-so-subtle plays of light andshadow on both subject and background.

5. Creating Depth

The film medium is a two-dimensional surfacethrough which light is projected. Any feeling ofdepth and three-dimensionality achieved ismerely an illusion. This implied depth must becreated; it does not happen automatically sim-ply because the subject matter is three-dimen-sional.

5.1 Planes of LightThe separation of planes is a significant aspectof achieving depth through lighting. Most lit-erature on live-action lighting discusses the ne-cessity of creating planes of light, often withoutreally explaining what it means or how toachieve it. A lighting plane is essentially a col-

lection of objects or subjects which are parallelwith the camera plane and are lit as a unit tocontrast with other planes. These planes can beany distance from the camera and are definedwith light for the purposes of creating the illu-sion of depth through layers.

As an example, a simple scene might nicelybreak down into four planes, from back tofront: the view out a window, the wall of theroom, the subjects, and a near-foreground ob-ject such as a vase of flowers. Lighting each ofthese planes in reference to each other will en-hance depth. The foreground vase of flowerscould be dark in silhouette, the subjectsbrightly lit, the background in partial shadow,with a bright view out the window. Each planeis clearly defined by its brightness.

This separation can be further accentuated bycolor variations between planes, with the sub-jects being in warm light while the backgroundwall is lit with cool light. The subjects can alsobe further separated from the wall with a littlebacklighting motivated by the window. And asanother step, the vase of flowers and the viewout the window might be out of focus.

By finding ways to create planes of light, ascene becomes easier to visually organize. Italso gains much added richness and beauty.

5.2 Volume and SpaceDepth is also implied by feelings of volumeand space. Volume can be created for indi-vidual subjects or other objects in the scene.Space is created by how these subjects and ob-jects are compared to each other.

For individual subjects, volume can be accentu-ated using back and side lighting, by highlight-ing prominent features, and leaving the rest inshadow. In contrast, any light which emanatesfrom the general camera direction tends to re-duce the illusion of volume.

P I X E L C I N E M A T O G R A P H Y 3 3

Page 34: Pixel Cinematography

The three-dimensional quality and the feelingof space in a set can be augmented by usingpools of light separated by dark objects or ar-eas. For example, a long hallway has more ap-parent depth if only a few parts of it are lit withshadow areas in between. A subject whichmoves in and out of light imparts a greater feel-ing volume as the light travels over its surface,and a greater feeling of space as the subjecttravels over distance.

5.3 Perspective and Depth of FieldThe motion picture image is a window into aworld where the viewer makes intuitive com-parisons to determine what they are seeing.Some of these comparisons are easy. If an objectis bigger than another, it is probably nearer. Ifan object overlaps another, it is decidedlycloser. At this point, the comparisons start tobecome more complicated. Does the object ap-pear small because it is a small object or be-cause it is far away? The brain looks for othermonocular and binocular clues to establish sizeand distance. The converging lines of linearperspective are very helpful, but can be alteredwith the use of various camera lenses. As an ex-ample, a long telephoto lens tends to condensedistance, making faraway objects appear verynear. Focus and depth of field are also impor-tant clues, but are dependent on the lens focallength and aperture used. Aerial perspective ishelpful, but it can vary depending upon atmo-spheric conditions. The brain also uses the anglebetween a light source and where its shadowfalls to help determine object-to-object proxim-ity. The viewer uses all of these comparisons toorganize the image and to establish depth.

Depth of field is an inherent feature of binocu-lar eyesight, hence it feels natural in a monocu-lar camera lens. With a live-action camera,depth of field is determined by the focal lengthof the lens as well as the aperture used. Eachlens has its limits within which it can operateeffectively. With these limits in mind, a lens ischosen depending on the story-point, mood,

3 4 P I X E L C I N E M A T O G R A P H Y

filmstock, available lighting intensity, theactor’s features, and compositional reasons.The choice is a technical decision as well as anartistic one. A synthetic camera has the techni-cal restrictions removed, which makes thechoice a purely aesthetic one. This doesn’tmean that the choice becomes any easier. Thelens and its depth of field determines how theviewer interacts with the subject. Two close-upswith similar subject framing can have dissimi-lar effects resulting from the perspective andfocal depth caused by the choice of lens. Onelens can place the viewer uncomfortably closeto the subject while the other places the viewerat a more detached distance simply throughdepth of field. A close-up where the back-ground is out of focus will feel more intimatethan one where the background is sharp.

Depth of field is also important in establishingdepth. A shot with varying levels of focus willhave more apparent depth than a shot which isuniformly crisp. In addition, depth of field isuseful for directing the viewer’s attention tothe subject which is in focus. Depth of field ren-dering is very convincing, although it can beexpensive and is not supported by all render-ers. Pseudo depth of field can be created byrendering the scene in planes and then blurringthem by varying amounts during compositing.This solution is adequate for many close-upswhere it is more frequently used. Another solu-tion is to use a depth buffer image of the cam-era view and blur the final pixels based on theirdistance from the camera. It is even possible topull a convincing rack-focus with these tech-niques, although it should be kept in mind thatsome blurring filters are more convincing thanothers.

5.4 AtmosphereThe use of atmospheric effects can enhanceboth depth and mood. Except for a windy dayor immediately following a rain storm, somedust and water particles hang in the air whichreflect and scatter light. These particles are par-ticularly noticeable in direct sunlight wherethey create shafts of light. The mood these

Page 35: Pixel Cinematography

shafts create depends largely on context andlight color. Bright, warm shafts of light feelwarm and cozy while cool shafts feel mistyand foggy. Dingy, warm, irregular shafts feelsmoky.

Atmospheric effects are also noticeable overdistance where they create aerial perspectiveby minimizing color saturation and contrast.Smoke and fog machines are heavily used inlive-action cinema to heighten these effects. Inthe computer aerial perspective can be createdsimply by adding a percentage of a fog colorinto the shading calculation for each pixelbased on surface distance from the camera. Formore complicated, three-dimensional effects,volumetric light shaders can be used.

5.5 Lens EffectsOther effects that enhance atmosphere anddepth which are often used in live-action arelens effects. Diffusion filters or nets can be usedto slightly soften a scene and glow the high-lights. These filters can be used overall or justaround the edges leaving the center clear. Onthe computer, gaussian filters and other imageprocessing techniques can be used to reproducethese effects during compositing. Lens diffu-sion effects create a soft, sensuous, romanticquality. They are often used in close-ups tobeautify actors, with less diffusion on mediumshots, and none on long shots. Other filters fre-quently used are grad filters which darken oralter the overall color or density as a rampacross the lens, or as a vignette.

5.6 Depth Using Color and ValueDepth can be enhanced with chromatic and lu-minance separation. A warm subject over a coolbackground will impart more apparent depththan a subject and background with the samecolor temperature. Similarly, if the brighter sideof a subject is placed against a dark back-ground area, it will help define the shape of thesubject, and keep it from blending with thebackground.

A further feeling of depth can be achieved bylighting the background more brightly than theforeground, with a blending of tonal ranges be-tween. This is especially effective in a dark the-atre environment.

6. Conveying time of day and season

Conveying the time of day and season is im-portant to place the story and to illustrate pas-sages of time. The time of day and season aremajor factors in determining the quality, quan-tity, motivation, direction and color of lightsources for a scene. They also are major compo-nents in setting the mood.

For daylight scenes, being too literal aboutplacing lights for the hour’s sun position is notnecessary and is often not aesthetically desir-able. A general feeling for dawn, morning, af-ternoon or dusk is usually sufficient since lightpositions often need to move to be able to lightsubjects attractively, and the subtle time dis-tinctions would be unnoticeable by the viewer.

At dawn, the light is blue, warming as the min-utes pass. Slightly later, in the early morning,the rising sun casts long shadows, and the colorof the light is slightly blue. At noon, when thesun is overhead, the lack of shadows tends toflatten the images and make color appear lessvivid and interesting. The light tends to be verywhite in color, becoming more warm as the dayprogresses. In the evening, at sunset, the shad-

P I X E L C I N E M A T O G R A P H Y 3 5

Page 36: Pixel Cinematography

ows are long, and the color of the light appearsmore red. The low sun at both dawn and sunsetrims figures and objects with light that sepa-rates them from the background and, depend-ing on the position from which the scene isphotographed, creates dramatic or romantic ef-fects. For winter scenes, the light is usuallycolder all day and the light angles remain morehorizontal. Light nearer to the poles tends to becooler than light nearer to the equator.

Morning light feels optimistic and cheerful. Theday is beginning with high hopes for what the itwill bring. The air is fresh and you feel restedand rejuvenated. Evening light is romantic, butit is also a little melancholy. The day is coming toan end, slowing down, and you are getting tired.These daily cycles are repeated on a yearly scaleas well; spring is a new beginning and the lightis clear and cool, autumn light is warmer andnature is winding down for the year. An old per-son portrayed in a morning spring setting willfeel more energetic and youthful than when thescene is set in afternoon autumn.

Day interiors are usually lit with soft lights ex-cept for any direct sunlight which may be shin-ing through a window. The light sources areusually motivated by windows rather thanpractical sources in the scene. Day exteriorlighting is motivated by the weather and thesetting. There is much more license to modulatethe light outdoors. Dappled lighting effectsthrough trees or a cucaloris are frequently used,especially on the background.

For night scenes, hard, directional lighting ismore justified, although soft light is also fre-quently used. The practical sources which arevisible in the frame should determine the qualityof the light. The illusion of night is created bythe angle and the distribution of light. The angleof the light tends to be less frontal for a night ef-fect, a three-quarters-back key light is frequentlyused with little frontal fill. Because there is lessambient and fill light in general, the percentageof well-lit areas in the frame is smaller, andspecular glare becomes more noticeable.

3 6 P I X E L C I N E M A T O G R A P H Y

7. Revealing characterpersonality and situation

The quality, color, and direction of light can in-dicate to the audience impressions about thepersonality or character of the subject. It canalso say something about the dramatic situa-tion, or emotional state of mind, in which thesubject currently finds himself. When a charac-ter is narrating a scene, he can describe histhoughts and how he sees the world; visualrepresentation of his thoughts is not necessary.Most of the time, however, we require the useof our imagination to decipher the motivationsand feelings of the characters by observingtheir actions, watching them emote, listening tothem interact with other characters, and sur-veying their surroundings. It would be confus-ing to listen to a movie with the picture turnedoff, we rely so much on our vision to tell uswhat is happening, which is probably why we“watch” television or go to “see” a movie.

Visual clues are an aide for the viewer to helphim understand the story more quickly or com-pletely, getting him emotionally involved withthe characters and their predicaments. Visualclues are comprised of tangible elements suchas location, sets, props, wardrobe, time of day,time of year, that are almost taken for granted,but without which the viewer would have nocontext. Is it Elizabethan England in the deadof winter, high noon on the chaparral, or a hu-mid summer night in New Orleans?

Besides establishing context for scenes, visualclues can also impart an emotional impressionon the viewer by employing symbolism. Myon-line computer dictionary describes symbol-ism as “expressing the invisible or intangibleby means of visible or sensuous representa-tions” as “artistic imitation or invention that isa method of revealing or suggesting immate-rial, ideal, or otherwise intangible truth orstates”. Some of this is absorbed on a consciouslevel (the good guys wear white hats, right? )while much of it is subliminal.

Page 37: Pixel Cinematography

P I X E L C I N E M A T O G R A P H Y 3 7

Light itself expresses symbolism as life, freedom,clarity, hope, enlightenment, truth, and guid-ance. Darkness represents the opposing forces.The source type can also express emotion.Candles, for instance, are associated with happyoccasions such as weddings and social dinners,as well as contemplative locations such aschurch. They are also nostalgic since they re-mind of us of times before electricity. Warm, soft,flickering candlelight is sensuous, flattering, se-ductive and romantic. Windows and doorwaysrepresent transitions. Our hopes “fly out thewindow”, and “opportunity knocks” on, andcomes in through, the doorway.

Light placement and direction impart emo-tional significance as well. Hard underlightingis commonly used to signify an evil or criminalcharacter or situation. Soft underlighting canlook very sensual. Lighting from directly over-head can look dreary when the subject is look-ing down, but spiritual, uplifting and hopefulas the subject looks up toward the light. Com-pletely illogical lighting is often used in dreamsequences or hallucinations, the more illogical,the better.

Whether we realize it or not, we attach sym-bolic meanings and react emotionally to virtu-ally everything. Some reactions are innate, oth-ers are dependent on our culture, and still oth-ers are uniquely personal.

As a lighting designer (and visual storyteller)we can take advantage of these emotional reac-tions in how we choose to portray characters ina scene. The best way to learn how to do this isto study films (with the sound off) to experi-ence how you are emotionally affected by whatyou see.

8. Complementing composition

The seemingly simple act of placing lights canradically change the composition and focalpoint of a shot. Good lighting can make a well-

composed image stunning. It can also rescue aless-than-perfect composition. As an example,start with an unlit, staged scene and add justone light with its shadow. Move this lightaround the scene and change its direction, ob-serving how the composition changes. As thelight moves, shapes transform into differentshapes as they become defined by light or lostin darkness.

Every shot is unique and requires its ownanalysis, particularly considering that each shothas its own story-point to convey. What workswell for one shot might not be the answer foranother shot. Most of the time, a shot will re-quire the lighting designer to address severalproblems. The background is too distracting,too busy or too plain, the main compositionallines all point to a different character than theone with the dialog, or maybe it needs specialcare for it to cut well with the next shot. Tryingto figure out where to start is often the mostdaunting dilemma for a new lighting person.The establishment of the focal point is the bestplace to start. In the process of creating empha-sis for the focal point, distracting elements willneed to be minimized. Once this has beenachieved, how the remainder of the scene is litwill largely be determined by the lighting style.

9. Continuity

It would be wonderful if there were enoughtime to craft each shot as its own masterpiece,capable of surviving scrutiny on a gallery wall.But there isn’t enough time, and sometimes it isalso not appropriate. A complex compositiontakes time to study, and the eye can take itstime meandering to various points of emphasis.A thirty frame shot needs to direct the eyequickly. The audience does not have time toguess where to look, the shot needs to have im-mediate impact.

Page 38: Pixel Cinematography

3 8 P I X E L C I N E M A T O G R A P H Y

The desire to craft each shot as a masterpiecealso needs to be balanced with the necessity of aconsistently lit sequence. It is very important toconstantly check the lighting progress on othershots being lit in a sequence, especially if theyare being lit by other people. Lighting will inevi-tably and necessarily vary from shot to shot, butthe overall feeling of the shot should be consis-tent with its sequence and especially with its ad-jacent shots. Sometimes this means that lightsneed to be brighter, darker, warmer, cooler, oreven repositioned to achieve a unified feeling ascamera angles change. However, the more simi-lar two shots are that cut directly together, themore important it is for the lighting to be same.

An establishing shot which shows a wide viewof the set for a sequence, will usually requirebroad lighting strokes and often more light inbackground areas to establish the set. When thecamera moves in for the close-ups, it is oftendesirable to darken the background slightly ormove a shadow line to help the foregroundsubject to separate from the background. Thesechanges should be unnoticeable to the viewer,and fortunately changes in camera angle andcut-away shots help hide these alterations. Youcan get away with more than you might think,but the only way to know for sure is to be ableto view the shot in context. The computer light-ing environment offers immediate feedback, aswell as preview and comparison capabilitiesnot found in live-action.

10. Film Considerations

Once a shot is lit on the computer, it is renderedand exposed onto film. It is then developed,printed and projected onto the screen, some-times with surprising results. Colors and val-ues can sometimes change drastically. Film canonly capture a small range of the available lightand color range of the real world. Film record-ers may not even be able to reach the range ofthe filmstock. Video monitors vary widely andmay not match the final result. For these rea-

sons, it is preferable to view lighting tests onfilm whenever possible.

Although it is desirable to get as close as possible atexposure time, the printing process offers great lati-tude in altering color and density (brightness). Thisprocess is called timing the print. Each shot can betimed individually, but not varied over time. Thetiming is applied over the entire image, it is not yetpossible to tweak isolated color areas as is possiblein digital film-to-tape transfer sessions. It should bekept in mind that it is usually much easier to time aprint darker than brighter. By brightening a print,more light is pumped through the negative whichcan result in a washed-out “milky” quality.

11. Conclusion

One of the most creative aspects of lighting is infinding ways to unify a screen-full of characters,objects and sources into a readable and believableshot. This is where everything discussed thus farcomes together and starts bearing fruit. But, ittakes practice to be able to look at an image andquickly know how to make it better. It helps tolearn from other people, and to learn to look atimages objectively. In art class critiques, instruc-tors frequently turn all of the students’ paintingsupside down or sideways and then start the cri-tique. The point is to teach the students to seethings they wouldn’t otherwise think to look for.It also helps to study classical painting, illustra-tion, Disney animation, still photography, and towatch movies with the sound off. Observe theworld around you at various locations, times ofday and weather conditions.

Learning to light is learning to see, but requiresmore than keen observation. The most difficultand most important aspect to learn is visualiza-tion, the ability to see the final result in yourmind before you begin. It is also important to becomfortable with what your tools can do. The fi-nal consideration is one which cannot be taughtas inspiration. This you must find within yourselfto nurture with experience and experimentation.

Page 39: Pixel Cinematography

P I X E L C I N E M A T O G R A P H Y 3 9

12. Suggestions

• Think about the story-point of the shot andthe overall mood you are trying to evoke.

• Think about what logical sources might be inthe scene which can help motivate the lightsyou are using.

• When attempting to complement a composi-tion with light and shadow, it might help tofirst break up the composition into its planes.If the foreground or main character elementis working well, it may help to try to take ad-vantage of it as much as possible.

• A large flat shape does not always needadded interest, it depends on the rest of thecomposition, but it often adds depth tomodulate or ramp off the light.

• Very saturated colored lights will particularlyalter a composition as they reflect unex-pected hues back to the viewer. Somecolors will go very dark and murky whileothers will leap from the screen and steal theshow.

• Avoid the temptation of lighting dark shotstoo dimly. At least part of the image needs tobe well-lit.

• Avoid placing a bright light which emanatesfrom near the camera lens.

• Remember that one light cannot always solvetwo problems.

• Think about shot-to-shot continuity, but don’tbe too limited by it.

• Consider that lights which animate in posi-tion, unless they are attached to a movingobject, will not look natural, use with care.

• Don’t be too literal about the physics of it, gowith what looks and feels right.

• Remember that there is always more than oneright way to do anything.

• Investigate new ways to do things.• Ask for new features in your tools.• Experiment and get a second opinion.• Share your successes and discoveries.• Keep it as simple as possible.

13. Bibliography

[Lasseter87] Principles of Traditional AnimationApplied to 3D Computer Animation, by JohnLasseter, Pixar; SIGGRAPH 1987.

Matters of Light & Depth, by Ross Lowell; BroadStreet Books; 1992.

Logic & Design in Art, Science & Mathematics,by Krome Barratt; Design Books; 1980.

Painting with Light, by John Alton; University ofCalifornia Press; 1995.

Film Lighting, by Kris Malkiewicz; Prentice HallPress; 1986.

Cinematography, Second Edition, by KrisMalkiewicz; Simon & Schuster Inc.; 1989.

Design Basics, Third Edition, by David A. Lauer;Holt, Rinehart and Winston Inc.; 1990.

Psychology in Action, Second Edition, byHuffman, Vernoy, Williams and Vernoy; JohnWiley & Sons; 1991.

Elements of Film, Third Edition, by Lee R.Bobker; Harcourt Brace Jovanovich; 1979.

Introduction to Film, by Robert S. Withers;Barnes & Noble Books; 1983.

Color & Human Response, by Faber Birren; VanNostrand Reinhold; 1978.

Design and Composition, by Nathan Goldstein,Prentice Hall; 1989.

Page 40: Pixel Cinematography

4 0 P I X E L C I N E M A T O G R A P H Y

Page 41: Pixel Cinematography


A Trad i t iona l F i lmmaker ’s Appr oach

By Steven Poster

I think that I had my first conscious thoughtsabout light as a pre-teenager. I had become in-terested in Photography at ten years old. Bytwelve I knew that this would be my life’swork.

I learned at first how light reflects off of sur-faces.

I knew that light was crucial to my life, but Ididn’t learn how to control…no, how to see ituntil my first six weeks at Los Angeles Art Cen-ter College of Design. These were extraordinarytimes for me. A man named Charles Potts wasmy lighting instructor. From my first lecturewith him I felt that I was having a religiousepiphany. His explanation of what light wastaught me how to see for the first time. If I cangive you just a small portion of what he gaveme I will feel that this was a successful seminar.

The first thing the soft spoken Mr. Potts taughtus was that light was a law: There were fivecomponents that always existed when a lightsource was shined on a subject.

These were:1.The Highlight Side2.The Shadow Side3.The Core4.The Cast Shadow5.The Incident Highlight

The qualities of these components on the sub-ject helped defined the direction of light, thesize of the source of light, the shape of the sub-ject, the material and surface of the subject.

Charlie, as he was affectionately known, alsosaid another startling statement during out firstlecture; “There are only two kinds of light.When It’s sunny and when It’s cloudy”. Thisdeceptively simple statement defines the qual-ity of all light sources in the universe.

We will spend the next couple of hours examin-ing how these five components of light andthese two qualities of illumination are the basisof learning to see light. Once this visceral con-nection is made, you will have to tools to beable to then learn to control lighting in all ofyour work. Whether it is in the real world orthe many fantastic virtual worlds that you willcreate. And once you can begin to control lightyou may be able to develop the spiritual con-nection to seeing how this seemingly simple el-ement of our lives controls much more then ourjust our ability to see.

P I X E L C I N E M A T O G R A P H Y 4 1

Page 42: Pixel Cinematography

4 2 P I X E L C I N E M A T O G R A P H Y

Page 43: Pixel Cinematography


Ligh t ing fo r Compute r Graph ics

By John Kahrs


As Computer Graphics increases its sophistica-tion, the craft of creating the images has comeinto it’s own. The tools have gotten easier touse, allowing the user to concentrate less ontechnology, and more on content. In the sameway a writer forgets the computer keyboard heor she types on, animators, technical directorsand artists can think less about operating sys-tems and Cartesian coordinates. They can startto concentrate on the lighting, the animation,the refinement of the image itself.

With computer lighting, there has been a thrusttowards creating ever more accurate models oflight, recreating the effects of light, and improv-ing the tools for lighting, but there has beenlittle attention paid to how to best use these of-ten very sophisticated toolsets. We also see agreat disparity in the quality of the images.There is work that is simply astounding in it’sexcellence (the T-Rex night attack in JurassicPark, Toy Story), and yet conversely, one canopen any magazine on 3D computer graphicsand see images in which the lighting seems tohave been given little or no attention. Hope-fully, these notes pinpoint why some imagessucceed, help identify common mistakes.

The idea is not to concentrate on the physicsand science of lighting (although this has to beaddressed, it’s unavoidable), but to get a betteridea of how light works, and also how lightand materials are reproduced with computers.This information is directed toward the users ofsoftware. The focus is on the craft of image-making with computers.

A logical place to start would be to find tradi-tional principles in similar disciplines that ap-ply to the new medium. This would includinglive-action cinematography, still photography,and even painting and illustration. These artforms evolved to a high degree long before CGwas even a concept, so it should pay to exam-ine these areas closely.

The special considerations for computer light-ing are numerous, and an attempt to addressthem has been made for these notes. One of themost troublesome issues is that computer mod-els of light usually give only a raw, rough out-line of the way light behaves. The complexityand subtlety of light that is the free, real timetoolset of a live-action director of photography(DP) is, in the computer model, just a slow,crude approximation. It is the digital artist’stask to bring some semblance of this richness tothe antiseptic CG world. Also, artists must alsospecify and tune material qualities. This task ofcreating materials is just as crucial as placingthe lights, yet it often happens that materialqualities are given only minimal attention, andwithout knowledge that they have limits andrelationships to one another.

Some of the concepts covered here may seemvery basic, but in practice they open up to acomplexity of variables. Overall, the aim is toconcentrate on concepts and principles: prob-lems and solutions, not how to use this or thatsoftware’s menu cells.

There is a misconception that you need a high-end, expensive renderer to do really good light-ing. Actually, a good renderer does help a lot. It

P I X E L C I N E M A T O G R A P H Y 4 3

Page 44: Pixel Cinematography





makes things easier, but it’s not necessary. If thesoftware is merely adequate, experienced userscan push the software further than the design-ers ever envisioned. Where you run intotrouble is when some basic capability is simplymissing. Then the workarounds become anonerous task, and a good workman can blamehis tools.

Whether you drive a Yugo or a Rolls Royce,you can still get to places. The Rolls will be apleasure to drive, and the Yugo you might haveto jerry-rig along the way, but you can still goplaces with it. And conversely, you can drivethe Yugo with style and skill, and abuse and ig-nore the quality of the Rolls Royce.

Anyone who claims that they are “stuck” withthe limitations of low-cost software should takea look at a few images from the CDROM gameMYST, which used an inexpensive, off-the-shelf

4 4 P I X E L C I N E M A T O G R A P H Y

acintosh software called Strata Vision 3D™.ith limited tools, MYST’s creators defined the

tandard for CDROM games to follow. Thisrobably had less to do with the software itself

han with the authors’ solid grasp of image con-rast, composition and sensitivity to ambientight issues.

o illustrate the points in this course, many ofhe images were rendered with Softimage’s origi-al renderer, a sort of middle-ground renderer:

he Ford F-150 of raytracers. Otherwise, Blueky’s CGIStudio was used to create the images.

ideo 101

henever I’ve tried to learn about lighting from book on filmmaking, I come across a chaptern “Basic Lighting Techniques”, and I’ve beenresented with a “basic lighting setup”. This in-variably consists of a key light, a fill light, and aback light, or “kicker”.

When I walked into Video 101 in my first yearof art school, this three-light technique wasdropped on us right away, and when we triedit, it looked awful. It looked cheap and plastic.Now that I think back on what we were doingin that little studio, I realize that my class, andthose books were all just scratching the surfaceof a more complex set of issues.

When I began working with computer graphics,I had no experience with any kind of studiolighting except for that bad experience. I hadsome background in painting and photography,so to a degree, I understood how light worked. Ishunned this standardized 3-light approach be-cause of it’s canned quality. So, at that point, Ididn’t have any approach, I just looked at realityand tried to observe what happened, to learnfrom it and appropriate it. I decided, not only inCG, but in general, that forcing every situation

Page 45: Pixel Cinematography



down a constrained pathway was a mistake, andthat it made everything look essentially thesame. I recently discovered that many of thesebooks and classes were being written and taughtby people who worked in television from the60’s and 70’s, and their concerns with light wereless about motivation than about ease of use,simplicity and low production costs. I wanted tostart from a different state of mind: which I laterdiscovered was to let the characteristics of thescene dictate the lighting.

The standard 3-lght studio setup:key light, fill light, and kicker.

Which usually gives a TV-studiolook to even the most innocent,unsuspecting subjects.

s I worked over time, I saw that the way aighting setup evolves in computer graphics isery different from one on a live-action set.ost of my time was spent trying to balance the

mbient light, or just trying to get the overallmage to have some kind of balance, or maybe itust felt right. A lot of effort was spent on theackgrounds and environments. I saw that theigure/ground relationship was completely dif-erent, because, for example, ambient light wasontrolled solely (and often arbitrarily) by theser. The overall balance of the image dependedheavily on how I adjusted the ambient light. Inthe computer lab back in school, many of myclassmates didn’t realize it was there to be ad-justed.

When I started lighting at Blue Sky, I began touse reference more and more. When I strayedfrom the reference, the image quickly got lost.When I didn’t have any reference, I never evenknew what I was looking for. I kept playingaround with our radiosity model when I hadfree time, (a switch in our renderer that recre-ates the complex diffuse reflection that occurswith real light). This allowed me to see howlight behaved when the computer model wasmore in line with what the real-life scene wouldlook like. It was like taking picture of a real-lifemodel of you scene. Slowly I got an understand-ing of how light works in the CG realm, in thesame way that a photographer can call out theƒ-stop and shutter speed of a scene simply bylooking at it. Many of the things I learned madetheir way into these notes.

Most interestingly, without even intending it,this basic key/fill/kicker combination began toappear in the lighting setups: both in my ownscenes, and in live action environments thatwere duplicated for integration. I’d be workingon a scene all afternoon, and suddenly, there itas: key light, fill light and backlight. The more

looked at film images and tried to understandhat was being done on the set, the more I saw

hese same elements of key, fill and backlightover and over again.

P I X E L C I N E M A T O G R A P H Y 4 5

Page 46: Pixel Cinematography

But this time, there was much more motivationand subtlety behind the placement and tuning ofthese lights. The lights had a logical reason forbeing there, not because a book or formula toldsomeone that they were supposed to be there.The degree to which the quality of each light’scontribution was refined was orders of magni-tude more than I’d experienced in school. Thiswas light years ahead of Video 101. I also sawthat this general configuration was common,but by no means universal. Good lighting set-ups were developed not from some formula,but from the needs of the situation.

There were more than a few lighting rigs thathad several lights, reflectors cards, mirrors andother little adjustments. And there were somelighting setups that were beautifully simple.Some used only one light.

Clearly, there’s no standard, basic way to lightany scene. The needs of the scene define theway it should be lit. To go even further, it’s amistake to think that this course, or some per-son, can teach someone how to light computerimagery. Just as in figure drawing, there is nostandard, or “correct” basic approach, but thereare ways of beginning, and there are common,successful techniques. The key/fill/kickersetup is one of these techniques and definitely away to begin. I no longer see it as a confining


Thisin cosolvYou you arenwangroulightrule kindstan

4 6 P I X E L C I N E M A T O G R A P H

ula or book of rules, but as a toolkit withh to express the image. It is part of the lan-e of lighting. In that video 101 class, we

didn’t know what to say yet.

The basics

Since we now have a friendlier view ofthis toolkit, we can take a closer look at it.The traditional live-action approach tolighting seems to be an endlessreconfiguration of the classic key light/filllight/ back light approach. Again, it is of-ten a good starting point, but a motivatedlighting setup rather than the defaultshouldalways be considered. Some scenes havenothing to do with these “standard” light-ing configurations.

classic approach has a definite applicationmputer graphics, and the problems it

es are essentially the same as in live-action.need general illumination for your subject,need to fill in the shadow areas so they’t too dark, and often, but not always, yout to separate your figure from the back-nd, and rimlighting, backlighting or kickers help to achieve this. Again, this is not abook, it is more a way of naming certains of commonly used lights and under-ding what they do.


Page 47: Pixel Cinematography


Wemait’s selfallyto t

FillA ficrealiveandscen

lighthisusedowof cdarligh

Key LightA key light is the primary source of illumina-tion. It may be several lights, to cover extendedspaces that a subject passes through. A com-mon key light placement provides 3/4 illumina-tion from above. “3/4” refers to the fact that themajority of, say, someone’s face is illuminated.The key light doesn’t have to come from anyspecific direction. It may come from behind thesubject, where it might be called a 3/4 rear key,or directly from the side: a side key. It is usuallythe first light on the digital or live-action set,and other lights are added after the basic illu-mination of the key light is roughed out.

Beginners often place the key light close to thecamera, because they simply want to see whatthey’re doing, which creates very flat, frontal

Key Light, shown here as a broadspotlight in a 3/4 placement.




light. A key light isn’t necessarily a frontallight. Just because the light hits the subject,doesn’t mean we’ve lit the subject. On the con-trary, it is light and shadow we’re workingwith. Frontal lighting is just one kind of light,and it can take skill and experience to make itwork well and not look boring and flat. Thequality of the shadows is at least as importantas the quality of light, and it’s it the two ele-ments working together that make an image.

s is why the “default” /4 placement is men-ed and used so often as a starting point: itsn’t just illuminate the subject, it models it,g light with shadows.

could go into the qualities of key light likeny lighting books seem to (such as whetherhard or soft), but those qualities should be-evident. Descriptions of this sort are usu- redundant explanations of what is obvioushe eye.

lightll light softens and fills the shadows that areted by the key light. Without fill light on a action set, the shadows would be too harsh dark, since the natural ambient light in thee is usually not enough to do the work.

On a real set, for the purposes of thesenotes, we could say there are two differ-ent categories of fill light. One kind of filllight could be called natural ambient. Thatis, the light that reflects off of objects litby general illumination. The other kindof fill light is added fill, or lights that areadded specifically by the filmmaker tofill the shadows. This light from addedfill bounces around the scene as well, sonatural ambient is really, from a live-ac-tion point of view, a by-product of thewhole process.

In the standard computer model of real-ity, the categories and complexities of fill

t are drastically reduced. Unfortunately, also means that a lot of the subtlety we’red to in real life ends up going out the win-. There is no natural ambient in the world

omputer graphics. The shadows are evenker, because there is no natural ambientt occurring in the scene.

P I X E L C I N E M A T O G R A P H Y 4 7

Page 48: Pixel Cinematography

To solve this problem, natural ambient light incomputer graphics often takes the form of gen-eral ambient light, which merely provides abase of unidirectional illumination in the scene,or it can be another point or parallel lightsource. Sometimes this ambience is specified inthe object’s material. Unidirectional ambientlight is the most disappointing kind of “light” incomputer graphics, and it’s often what makescomputer graphics look so awful. It’s really notlight at all. It provides no modeling in the shad-ows, contributing to the stereotype of flat, com-puter-looking images. And even though thiskind of light is wholly insufficient, any subtletyit may have to offer is usually nixed, since it’soften the least-refined lighting component in aCG scene. On the contrary, fill light and ambientlight probably should be the most tuned, mostrefined component of a scene.

Computer generated ambient light has to fill avery big gap in the computer’s model of reality,because without a radiosity renderer, this modelis seriously incomplete. The CG artist’s task ismore complicated as both kinds of fill light: natu-ral ambient and added fill, must be duplicated.

Fill light, created with acombination of ambience anda low intensity spotlight.




4 8 P I X E L C I N E M A T O G R A P H Y

Fill light on a live action set often comes froman area close to the camera, leaving no shadowuntouched, at least from the camera’s point ofview. Other times, a general set illumination isprovided to raise the base level brightness ofall objects in the scene. Screwball comediesfrom the forties often have this kind of high-key fill light.

In the real world you can get away with the in-herent flatness that this tends to add to the im-age, because there is that natural ambientbouncing all over the place, adding modelingand interest. In the CG realm, this kind ofsubtlety has to be added by the artist, and thereare different ways of tackling this.

The concept of ambient and fill light is dis-cussed again in these notes, but for now, a goodplace to start when adding fill light would be acombination of unidirectional ambient com-bined with one or more additional point lightsources from logical directions, which some-times ends up being from the opposite direc-tion of the key. If you can eliminate the occur-rence of specular highlights on objects fromthese sources, the ambient can look very believ-able. Sometimes, this simple base ambient maybe all you need. Sometimes the scene may needa complex arrangement of fill lights.

Back lightA backlight provides highlights and rim light-ing around the edge of the subject. The back-light is often what differentiates the subjectfrom the background, and it’s also often whatseparates ordinary life from Hollywood. Whenit happens naturally, it can be beautiful. But thesituations where backlighting occurs naturallyare just as often as any other arrangement. Inmovies, it seems to occur continuously, and isone of the things that makes movies look likemovies. Backlighting tends to bring a stylizedquality to the image. When it has no relation-ship to what might occur naturally in the scene,it looks very stylized.

Page 49: Pixel Cinematography

At sombels band roonly tlightsis logthe simay h“Thistance.rimligis it h

A “kicker”, or rimlight, usedtypically to separate the sub-ject from the background.

The key-fill-kicker lightscombined together. Whenthe lighting is unmotivated,it looks stylized and clichéd,but this combination can beused to create thousands oflighting environments.




I Stu


Generally, it’s a good stylization, and it has apurpose, which is to separate the figure fromthe ground, and define the edge of the subject.

Typical back lighting comes from behindand above the subject. A “kick” lighttypically refers to a light positioned di-rectly opposite a 3/4 frontal key, a tech-nique that was very common in the1940’s. Another term called rimlightingprovides illumination around the edge ofa the subject, not necessarily comingfrom above or directly behind.

Different lights can switch roles, with thekey light eliminated, for instance, andedge lighting as the dominant source ofillumination. This is prevalent in nightphotography, since it defines the clearlyagainst the groundplane, but the overallimage remains dark, so we still think“night” when we see it. The rimlightbecomes the “key” light, and where thekey light used to be is now working as afill light.

e point, categorizing the lights with la-ecomes pointless as the numbers increaseles become more vague. But it’s useful, if

o communicate ideas, to have names for and what they do. If we use lighting thatically motivated and has a logical link totuation, then formulas become obsolete. Itelp to think of lighting in the sense of:

light represents the street light in the dis- It happens to be functioning as aht.”, and not: “Here is the rimlight, why


P I X E L C I N E M A T O G R A P H Y 4 9

Page 50: Pixel Cinematography

Supplementary lights

EyelightsA related kind of fill light that is used often ona live action set with actors is the eyelight. Itseems that whenever a subject’s eyes have bril-liant little highlights in them, they seem morealive, more sparkly, they “come to life”. Peoplewho make movies have known this practicallysince movies started, and the way it’s oftendone is by placing a small, low intensity pointlight source right next to the camera and point-ing it at the subject. A dim light will still pro-vide highlights in shiny surfaces such as eyes,and won’t illuminate the diffuse surfaces verymuch. There’s no reason this won’t work forthe same reasons for computer generatedscenes and characters.

PracticalsA practical light is a light source that occurs inthe frame. A lamp on a table, a candle, a flash-light, and yes, the blinding light from themothership in Close Encounters are all practicallights.

These kinds of lights are often not powerfulenough to shed light on their own in the scene.You might have a memory of watching oldmonster movies, and figuring out that, “Hey!There’s some guy with a spotlight off camera!”,who is shining it on the wall next to the womanwith the candle on the stairway. Film stockswere much slower back then, so light fromflashlights, candles and table lamps wassupplemented with off-camera lights. This of-ten looked obvious, but as lenses and filmstocks became more sensitive, the actual lightfrom these lights could be used, probably cul-minating with the film Barry Lyndon, when thedirector, Stanley Kubrick, had special extra-fastlenses developed so he could film scenes usingactual candlelight. The techniques that “fake”the light from a practical light are more stillused often, and are much more convincingnowadays.

5 0 P I X E L C I N E M A T O G R A P H Y

As an example, let’s take a step back from this scene and figureout what’s happening. First of all, the light has now becomemuch more localized, so whatever ambient contribution therewas from the walls and floor is mostly eliminated. Our first re-sponse for this scene would be to eliminate the ambient lightaltogether.

With a basic, 3/4 key light shining on the subjects, the environ-ment around them illuminates pretty evenly, so the default uni-directional ambient does an adequate job.

If the lighting moves outside of this basic situation, suddenlythe basic ambient light becomes irrelevant. In the shadow ar-eas, there’s no modeling at all.

Thinking about ambient light



e re









e re



Page 51: Pixel Cinematography

When the two lights are combined, the scene begins to have aresonant quality to it: we start to believe in it, or at least muchmore than we believed in the first spotlight image. Instead ofsettling for simply illuminating the scene, we tried to isolatewhat’s happening in terms of the ambient light, and create amore specialized ambient that has some relationship with whatwe’ve observed.

So now there’s no light reaching the subjects outside of thespotlight. Again, what would happen here if this was happen-ing in reality? The bright spotlight illuminates a portion of thefloor beneath the subjects. This light bounces back up into thescene from below. We could consider it a light source. Wecould place a point light source right where the floor is illumi-nated, and it does the trick.



e re







In computer graphics, whoever’s doing thelighting has a distinct advantage in this regard,since he/she can control lights of any bright-ness inside the camera’s view without thesource being seen, eliminating the need to fakethe practical’s light with an off camera light ofgreater brightness.

Creating bright lights in the image frame bringsup some points about qualities that make anobject seem truly brilliant, such as diffraction orflaring. There has been some interest in recentyears in ways to overexpose and flare out superbright objects in computer images. Blooming,flaring halos around bright lights offer a visualcue to the eye that makes the difference be-tween an odd white shape on the screen and abrilliant, blinding light.

Local spotsWherever the brightness of a localized areamust be controlled independently of the entirescene, a local spotlight is sometimes used.Computer artists have the capability (if thesoftware does also) of isolating which objectsare affected by certain lights, so this local con-trol is actually much more flexible than in liveaction. If those doing CG lighting would kill forthe kind of interaction that live-action DP’shave, they’d probably kill to have our controland isolation capability with CG lighting.

Local spots are especially handy in productshots for commercials, to insure that the labelsare exposed properly, and often to simply illu-minate the face of a subject, or to draw atten-tion visually to a certain place in the frame. It’simportant not to let these added lights becometoo numerous and complex, as it make the en-tire lighting setup slower and harder to control.

With all our ability to twist or own models ofreality into customized images, it may seemodd for live action DP’s to break the rules. Itturns out they can do anything, such as usingtiny, isolated spotlights pointed at a tiny strip

P I X E L C I N E M A T O G R A P H Y 5 1

Page 52: Pixel Cinematography

of mirror attached to a motion control rig. Thiscan create a local highlight on a subject justwhere they want it, and nowhere else. Powerfulbut very focussed spots can backlight just thesteam from a cup of coffee, or a diamond ringon someone’s hand.

This sort of very specific tuning of the lightshould be exercised with care. If you go too farwith it, it’s much like overworking a painting.There is a kind of “forest for the trees” phe-nomena where every little object is tuned andbalanced perfectly, yet somehow the overall im-age lacks impact. This topic is covered in moredetail in the section called “Image impact”.

Ambient light

Ambient light should be considered the mostcrucial part of computer lighting, and thoughthere’s been a lot of improvement in the wayit’s used is computer graphics, it’s role stillseems the most often under appreciated. Thereason it’s so important is that, in the com-puter world, it has to fill the role of duplicat-ing very subtle and complex phenomena thatoccur in real time, all the time in the realworld: indirect illumination, or radiosity.

As you read this, take a look at the environ-ment around you. Some of it will probably belit directly by light sources. But chances arethat many areas are illuminated by reflectedlight. In some cases, the entire scene, such asthe shady side of a daytime city street, is lit byindirect light.

Computer artists have enough trouble just rec-reating direct lighting, not to mention thecomplexities of indirect illumination. Itdoesn’t come automatically the CG world, andit’s such a big part of the environment aroundus every day.

5 2 P I X E L C I N E M A T O G R A P H Y

Using local fill lights to create aricher ambient environmentWhen light illuminates a space, it reflects off surfaces in thatspace. Using judiciously placed spotlights with very soft,broad cones, we can approximate this diffuse inter-reflection.

The default, unidirectional ambience offers little in the way ofsubtlety. As a first step, let’s get rid of all the lights and con-sider what’s going on in this room.

Before we work with the sunlight, what about the sky? Thebrightness of the light coming in the windows is an importantcontributor to this scene. For this room, the planes of thewindows are, effectively, local light sources. Two very softspots (over 180˚ of spread) are placed in each one, with sig-nificant dropoffs. The colors are on the cool side, since thelight is sky blue.

Now for the sunlight. When the sun hits that floor, it reflects upinto the room. For efficiency’s sake, two spotlights along thelength of the patch of light and just below floor level are usedinstead of an expensive and (for this particular renderer, un-available) area light. The have a similar dropoff and spread tothe window spots, only they are warmer in hue. Much warmerthan the sunlight itself, too.



e re









e re



Page 53: Pixel Cinematography

Remember, all these fill lights don’t cast any shadows.With a total of only four lights, this scene won’t take toolong to render.

When all the elements are combined, the image is miles aheadof where we started. Also keep in mind that these lights wereisolated to show each component separately. That’s not the wayto work, though. Usually everything has to be up and runningsimultaneously to understand the image as a whole. Nothing’sset in stone, either. Several materials were tweaked along theway. The floor highlights were knocked down, and the doormaterial was made much darker, to get a more mysterious feel-ing that contrasts with the bright room.

To exaggerate that contrast even more, and get a more open,bright feeling, some of the dreaded unidirectional ambiencewas added back into the scene, in a small dosage (10%). Inthis role, it works well. Tweaking the hue of this ambience onlyslightly can drastically affect the hue of the entire image. A bitof warmth is plenty for this scene.

There are some areas in this image that feel almost likeradiosity rendering. With more detailed modeling and sometexture-mapping on the walls and floor, it could approachphotoreal, and all with a few, ordinary lighting tools.



e re







Live action directors spend plenty of time andeffort either trying to get rid of the ambientlight, or trying to direct it into the right place.On a typical set, there are large frames withblack cloth stretched across it to kill the ambi-ent reflections from unwanted areas. There arealso similar contraptions with white material toreflect soft illumination back into the scene.They try to balance the look of the scene withthese and other tools, and that’s exactly what it’sall about: balance. This balance is going tochange significantly according to the situation.Usually, the more ambient light there is in yourscene, the less contrast there is. The ambientlight has to be adjusted to complement and logi-cally fit into the needs of the lighting situation.

Every scene will have it’s own quality of ambi-ent light, and each situation should be tackleddistinctly, in terms of it’s own needs.


A crucial characteristic of light is that fact that itsintensity drops as the distance from the light in-creases. When you hold a piece of white paper 2inches from a 100 watt bulb (the light has to beturned on for this complex experiment to work),you can hardly look at it, it’s so bright. Hold thepaper it 6 feet away and it seems dim by com-parison. Except for the sun (which is so distantthat its dropoff is negligible here on Earth) thiseffect occurs everywhere. Candlelight wouldn’tbe candlelight without it. In live action, it’s a ma-jor consideration in terms of exposure.

Any self-respecting renderer or raytracer oughtto have dropoff controls for it’s light sources,it’s often up to the user to find those controls,because it tends to be turned off by default. It’scheap to compute, so why not add it to yourscene? This is one lighting switch that, whenyou start using it, pushes the believability andsubtlety of your images to another level.

P I X E L C I N E M A T O G R A P H Y 5 3

Page 54: Pixel Cinematography

It’s often necessary to compensate for the de-crease in brightness at the subject’s location bypumping up the intensity of the light at itssource. It’s also useful to note that the smallerthe light source, the more pronounced thedropoff is. A tiny strobe light flashing on aplane wing at night drops off harshly over avery short distance.

When lights have a pronounced dropoff, con-sider saving render time by excluding theselights as illumination on distant objects. (For in-stance, those mountains in the distance at nightaren’t visibly affected by the porch light of thecabin in the valley.)

Some places where dropoffs can really lookgreat is when you’re trying to recreate diffuselyreflected illumination, such as the sunlightbouncing off the floor in a room. The sunwouldn’t have a dropoff, but the patch of sun-light, because it is effectively a local light source,definitely has a dropoff. Consider how bland thelight from this effect would be without it.

Judiciously placed fill lights with carefullytuned intensities and dropoffs can give CG im-ages a radiosity quality. As light bouncesaround in a scene, that light has a rich, variedquality that rises and falls in a complex layer-ing of overlapping dropoffs. Using lights thisway can offer an organic feel that CG imagesoften lack.

Material properties

Understanding how light behaves is one part ofdigital lighting, but the computer cinematogra-pher also needs to have an understanding ofhow surfaces react to the light. Not that wehave to become part-time physicists, but thisinvolves knowing a little about the science ofmaterials.

5 4 P I X E L C I N E M A T O G R A P H Y

There aren’t really any rules for lighting, onlyguidelines, but there are some pretty well de-fined rules for the physics of materials, and itpays to rely on them, or even just use them as astarting point, because you’re nowhere ifyou’re just guessing and flailing around.Knowing about how different materials operatecan make the difference between images thatare just OK and images tat are outstanding.

Components of Dialectric Materials




I Stu


The specular component isolated. Specularity here is theneutral-colored highlight of the polished lacquer on thesurface of the ball.

A pool ball makes a good example to explain materialcomponents.The diffuse component is the reflected energy ofthe flat pigment that gives the object its color.

Page 55: Pixel Cinematography

Specularity and DiffuseThere are two major components that are usedto describe materials. These are specularity, anddiffuse. The color, property, and proportion ofthese two components are used as a foundationto describe just about any material that exists.Unfortunately, many renderers treat this simplerelationship in a more convoluted way, so thebetter we understand these characteristics, themore effectively our materials and shaders canwork for us.

When light hits an object, the energy is reflectedas one of two components: the specular compo-nent (the shiny highlights) and the diffuse (thecolor of the object). The relationship of these twocomponents is what defines what kind of mate-rial the object is. These two kinds of energymake up the 100% of light reflecting off an ob-ject. If 95% of it is diffuse energy, then the re-maining 5% is specular energy. When thespecularity increases, the diffuse component

Diffuse and specular combined.

When these two elements are combined, the two reflectedcomponents are represented. This is true for any material, it’sthe proportion that differentiates each one on this basis.




P I X E L C I N E M A T O G R A P H Y 5 5

The specular reflectance of non-metals is very mild: note howfaint the reflections are. This is important to remember whencreating realistic-looking materials.

A common error is to set the reflectance characteristics of non-metals too high, which isn’t representative of such materials.




I Stu


Page 56: Pixel Cinematography

drops, and vice versa. A ping pong ball is con-sidered to be a very diffuse object, with verylittle specularity and lots of diffuse, and a mirroris thought of as having a very high specularity,and almost no diffuse.

Nonmetals, white highlights, and4% of the lightIf we can divide two major components of re-flected light, we can also further separatematerials into two main categories: metals andnonmetals.

First, let’s talk about nonmetals. These kinds ofmaterials are called dialectrics. They includeFormica tables, enameled automobile bodyparts, bic pens, telephones, pool balls, polishedwood, dull paper, enamels, glass, magazinecovers, polished fingernails, granite, marble,any other plastics. The list goes on.

Dialectrics have two important characteristics.One is that the highlights are always white(well, actually the color of the light), and thesecond is that they usually have a specularcomponent of only about 4% or less.

What this means is that most nonmetals usu-ally reflect no more than 4% of the light asspecular energy, and this energy is the color ofthe light source, not the diffuse color of the ob-ject. The rest of the reflected light is of the dif-fuse component. This may not seem right, butwhen you take a closer look at the worldaround you, it makes sense.

Reflections in a car’s bodywork, for example,are actually rather dim compared to the reflec-tions in the chrome. In addition, these reflec-tions aren’t colored, they’re neutral. If the re-flections on a red car look red, it’s because thereflected light from the paint is showingthrough from underneath, not because the re-flection itself has a tint to it. Tinted reflectionsare the province of metals, which are discussedin the next section.

5 6 P I X E L C I N E M A T O G R A P H Y

When you look in store windows, it’s easy to see increasedspecularity as the surface turns away from us. Here’s a mate-rial that effectively has the diffuse component eliminatedfrom it, and the dimness of the reflection in the window isjust a fraction (about 4%, coincidentally) of the intensity ofthe street behind you. It isn’t reflecting like a mirror at all.

As the window surface turns away from us, the reflectionsbecome much more pronounced, to the point where the glassbecomes opaque, and it does behave more like a mirror.

When people discover, with delight, that theycan reproduce reflections in computer-gener-ated images, they enthusiastically make themistake of pumping up the specularity so thereflections are more pronounced. This is oftenseen in student animations, including my own.

Shinier on the Edges

Page 57: Pixel Cinematography

Now a word about white highlights. It’s OK tobend the white highlight rule from time to time,especially with subtle materials such as skin, butusing white highlights is a solid starting pointfor nonmetals. Most of the highlight variation inthese kinds of materials are in it’s size and in-tensity, not the color.




The highlight colors of metals are the same as the diffusecolor, and metals tend to have a much higher specularity thandialectrics.

Since the reflections are such a prominent component of theway metals look, lighting metallic subjects has more to do withcontrolling the reflected environment than adjusting the lighton the subjects themselves.

If there were no surrounding environment to reflect in thesespheres, all we’d see is the specular highlight on a black field.





Another interesting thing about specularity isthat it increases as the surface turns away fromthe viewer. A good demonstration of this phe-nomena is to take a piece of paper (somethingthat seems to have very little shine, if any) andhold it out in front of you so it’s almost parallelto your line of sight. You might be a bit sur-prised to see shiny highlights and even somedull reflections in the surface.

You can see this effect clearly at night whiledriving: in the pavement are reflections of theheadlights of oncoming cars. Pavement and pa-per aren’t materials you normally associatewith having reflections, but there you go. Lookout across calm water, and you can only see re-flections. Look down into the water, and the re-flection is much fainter.

Where this effect gets really interesting is in con-junction with rimlighting, or kicker lights, orbacklighting, or whatever you want to call it. Be-cause these kinds of lights hit only the edges ofthe subject, and they come from behind the sub-ject, these circumstances combine with the abovementioned phenomena of increased specularityto create brilliant, dramatic highlights that sepa-rate the subject from the background.

Unfortunately, understanding this detail aboutedgelighting isn’t enough. You might be stuckwith a renderer that is simply missing this ca-pability, and many of them are missing it. Thisis one example of something that seems small,but when it’s not there, your eye tells youthere’s something wrong.

MetalsAs materials go, metals are a different animal,compared to dialectrics. The highlight color ofmetals is the same as the diffuse color, as op-posed to neutral, and the specularity is muchhigher than that of nonmetals. A mirror is es-sentially a highly polished metal with neutralcoloring, and would be considered an ex-tremely specular object. All metals could be

P I X E L C I N E M A T O G R A P H Y 5 7

Page 58: Pixel Cinematography

considered totally specular, and the amountthat the user specifies really has to do with howmuch oxidation, surface abrasion or dirt thereis on the surface. A subway car is almost mir-rorlike, but the dirt and coatings drop thespecularity down considerably.

Beyond these characteristics, making metalslook great is the same as making transparentmaterials look great: it’s not about the light thathits them, it’s what they reflect and refract thatmatters. It’s about the environment around theobject, not the object itself. This is why anyonewho knows about photographing jewelryspends most of their time up front buildingthese house-of-cards environments around thepiece, and spending most of the time lightingthat controlled environment.

It also doesn’t hurt to sometimes recreate theblurry reflections in the surface, if it’s right foryour purposes. This can be accomplished mosteasily by blurring the reflection map, or writingthe diffuse reflections calculations into the ren-dering software itself. The latter is a more accu-rate technique, but the former is a nice cheatthat renders quickly.

Limits of classic computer lighting modelsBefore we jump into practicing what we’vepreached, let’s cover a little more ground on aspecific element of material characteristics.

Because of the way computer lighting modelsevolved, we have to use a somewhat limitedway of creating realistic material qualities. Ev-ery renderer or raytracer uses some variationon a solution for creating specular highlights inmaterials. They assume that the light source isa point source, such as a 100 watt light bulb onthe other side of a room. If that’s your lightingsituation, you’ve got it made. But what ifyou’re trying to create, say, little characters thatlive 2 centimeters beneath that light bulb? Sud-denly it’s not a point source any more, it’s animmense, broad field of light that encompassesthe subject. How do we recreate this situationwith “traditional” computer lighting models?

5 8 P I X E L C I N E M A T O G R A P H Y

The answer isn’t so easy. The problem is that asthe size of the light source increases, the one-stop shopping solution for specular highlightsstarts to break down and become inappropriate.The classic highlight is only a quick approxima-tion of what’s happening. A highlight is a reflec-tion of the light, sometimes soft and broad,sometimes tiny and tight, but still a real reflec-tion. Unless the light source is on the small side,the distinction between the lights and the reflec-tions of objects becomes more vague.

Also keep in mind that the diffuse lighting maynot “wrap around” the object properly if thesoftware is always treating the light as a pointsource. The quality of light from large lightsources such as overcast skies are difficult, if notimpossible, to reproduce with a renderer thatconsiders all lights to be point light sources.

These limitations are found in almost everyrenderer software that uses traditional lightingmodels for CG imagery. There’s been somevery interesting research done recently to ad-dress these shortcomings, but they often in-volve very compute-intensive solutions. (Let’sface it, these are complex phenomena that aredifficult to recreate). But when someone has noR&D department, or little time, other solutionsmust be found.

At some point, you have to make a call as to howyou’ll solve this problem: the reflection of thelight source doesn’t fit with what’s happening.

If the object is reflective, and you can raytraceor create an environment map, then you’re inluck. Get rid of the specular highlight and re-flect an object or a map of the kind of lightsource you’d see in such a situation. If the sur-face is rougher, you might try blurring the re-flection map, which is a classic trick that worksfor most situations.

Other, more diffuse materials have no visiblereflections but still have some highlights, likeskin, or clothing. In these very soft, large lightsource situations, usually you can simply

Page 59: Pixel Cinematography

spread out the highlight. Consider eliminatinghighlights altogether and concentrating on get-ting a softer quality of diffuse light with mul-tiple soft shadow-casting sources.

We’ve discussed some major points with lightsand materials, and we’ve isolated differentcomponents in all materials. Some peoplemight be scratching their heads when they goto apply these ideas to whatever software theyuse: Specularity doesn’t do what is described inthe notes. Perhaps “diffuse” is not available asa control. Maybe it’s called “shininess”, andwhat’s that?

Nobody has to conform to anyone’s idea ofhow to name, control or relate material charac-teristics. For instance, in the Blue Sky renderer,a field called “Roughness” controls the size ofthe specular highlight, and “Specular” controlsthe intensity of the highlight. In the Softimageoriginal renderer, the slider that says “Specular

decay” controls the highlight size, and a tripleRGB slider controls both the color and thebrightness of the highlight.

Sometimes a control called “roughness” or“shininess” is the control for the specularity ofobjects. However it ends up, it’s up to the userto isolate and find out what knob really con-trols what. In the end, even cheap software canend up being pretty impressive if it’s reined inwith what you know.

Lighting with a plan

The previous sections should give us a betterunderstanding of what computers do to recre-ate light and materials. This section continuesto emphasize finding the needs of the lightingsituation, and making appropriate decisions inresponse.

With the standard computer graphicslighting model, lights are consideredpoint sources. If this matches your situ-ation, then highlights look believable.

As the size of the light increases, thesize of the highlight can be increased tomatch, but at some point, this specularlighting model becomes unable to ap-proximate what would happen, espe-cially for highly reflective materials.

One soluplace it wcase, theon it tha




I Stu


A limited solution for highlights

tion is a eliminate the point source highlight and re-ith a reflection of something more logical. In this

sphere raytraces a box with a bright, vignetted mapt mimics a very soft, square studio light.




P I X E L C I N E M A T O G R A P H Y 5 9

Page 60: Pixel Cinematography

Working with computers to create pictures pro-vides advantages that visual artists from thepast never had: perfect perspective rendering,scientific models of the way light behaves in ascene, complex texturing capabilities. It’s neatstuff, but it’s only a tool: it can’t truly refine theimage. That’s the job for you, the user.

Ollie Johnston, a great Disney animator, createdan indispensable list of animation notes. One ofmy favorites says, “Spend half your time plan-ning your scene and the other half animating.”It’s good advice, and it’s no different for light-ing. There’s a lot to be said for having a clearidea of what you want to do before you evenbegin to rough out the lights in a scene. Some-times, if you’re working for a director, theyhave this clear idea about what a scene shouldlook like. That’s usually a good thing, but youstill have to bring your artistry to fleshing outthe details. There are other times, though, whenno one is providing the vision except yourself.So where does this vision come from?

Some ideas about where to start have alreadybeen utilized in the examples for these notes. Inevery situation, there is an effort to isolate eachscene on it’s own terms, to find just what ex-actly is happening with the light. Not to shoe-horn the scene into some formula of how light-ing should be, but really think about how thelight and materials need to fit together to makethe image work for you.

Using ReferenceOne image integrates perfectly, another falls farshort. One scene really looks like what youwant, another just isn’t right for some reason.There are a lot of reasons why some images suc-ceed and why others don’t. One of the keys tosuccess is to use reference, to recognize what it isabout the lighting situation that makes it what itis, and to put this knowledge into practical use.

6 0 P I X E L C I N E M A T O G R A P H Y

One place to start is to figure out what you’vegot to start with, and look at reference for simi-lar lighting situations. Not to copy from, butto give answers about why certain things areoccurring.

What kind of lighting situation is this? Exterior,interior? It sounds obvious, but sometimes theanswer surprises you. What time of day is it? Isexterior light coming in the window? Whatcolor is the light? What is the possible range ofcolors for that light? Are these colors what youthink are right for the light, or have you lookedat the real thing? What is the size of the lightsource? Is a soft or hard light, and what can bedone to get this quality in the CG lighting? Isthe environment bright or dark? If so, howdoes this affect the overall ambient light is thescene?

To go even further, what is the mood and emo-tion of the scene? What kind of lighting is ac-ceptable in this situation? Is it open and light,or dark and mysterious? How well does yourscene get this visual idea across? What can bedone to reinforce this idea visually? Is there astyle or genre you’re working in that has loadsof reference material to offer? If so, get a hold ofsome and find out what it is that makes thisgenre what it is.

Lighting in the CG realm is in many ways likedrawing. When you draw from life, the draw-ing is often successful when you constantlycompare your work with the reality of whatyou see before you. Without the reference fromlife, you rely on your visual memory to signalwhether something feels right or wrong. Thestrength and clarity of your visual memory dic-tates how resonant the image is that you pullout of thin air. If your visual memory is faint orvague, just look around you at life again, andthere– you have solid reference again.

Page 61: Pixel Cinematography

Having reference is crucial. Without it, you’reguessing. With it, your dead ends seem to dis-appear. Just imagine how far you could getlighting an airport at dusk with no referencematerial. Sure, it might end up looking quitenice, but why miss out on the subtleties wedon’t even know about in our mind’s eye? Withplentiful images of the real thing, the optionsexplode, often leading in exciting directionsthat you hadn’t even conceived before.

One place where reference is indispensable is interms of ambient light. How much? Whatcolor? What qualities? It’s all there in the pic-tures. At Blue Sky, radiosity is in the productionsoftware, and it can provide a very interestingreference for ambient light. When ambientlights are turned off and radiosity is turned on,all the ambient light occurs naturally and cor-rectly, so it’s like taking a photograph of thescene. From there, “imitation” ambient lightscan be tuned so they’re more accurate.

Using reference is a way of truly understandingthe dynamics of a scene. Don’t hack the scenetogether. Think about it what’s happening therefirst. Again, let the renderer do the work foryou, and don’t cheat until you have to. (Or atleast do it right first, then find out what youcan get away with.) Use the live action model:create custom effects and specialized lights torefine the scene only after the basic lights havebeen roughed out. If the lighting setup gets toocomplex and numerous to keep track of, breakdown the components and find out what eachlight contributes.

Efficiency and simplicity

The majority of lighting situations can be recre-ated and look great with variations of only afew lights. Lighting is very much about quality,not quantity. We have infinite control overquantity of light. That’s the easy part. Addingmore and more lights into a scene is not neces-

sarily going to make it better. Too many lightscan often make a scene look flatter, run slowerand creates a rig that’s too complex to keeptrack of. A clean, simple lighting rig can usuallyprovide a unified, balanced look, no matterwhat the subject.

Perhaps the idea in CG lighting shouldn’t be,“With the computer, I can add as many lightsas I want, and independently control eachone!” The software is based on a physicalmodel of reality, and much work has been doneby the people who wrote it to insure that whathappens, happens for a reason, so let it happen.Let the renderer do the work for you. Don’t fillin every shadow with a separate light, let theshadows fall, and adjust the lights in general,or move the lights. Your first instinct should beto refine the existing lights, not to add anotherone.

When the rig does get complex, it’s importantto know which light does what, and an easyway to do this is to temporarily turn off all thelights except the one you’re interested in. Un-less you know what each light contributes in acomplicated rig, you often find yourself flailingand getting nowhere.

With a simplified lighting rig, it’s easier to un-derstand and change the degree to which eachlight contributes. Not only is it easier for you,but also for whoever else works on the scene.And it’ll render faster with fewer lights.

There’s a time for deciding if there’s a way toget lighting effects without lights. Flying over acity at night, for instance, is the kind of scenethat could use a light for each street light on ev-ery street, but you’d be a lot better off figuringout some alternative with mapping effects.

P I X E L C I N E M A T O G R A P H Y 6 1

Page 62: Pixel Cinematography

Image impact

Adding more and more lights to refine an im-age brings up another issue: overall image im-pact. It is lights and darks and large shapes thatcreate depth and impact in an image. It’s muchof what the eye responds to. If every shadowacross the image plane was filled and groomedand perfectly balanced, it ends up lookingpretty boring.

You many find yourself grooming the lights in ascene for a long time, thinking it’s going to begreat when you’re done with all the tweak work.The next day you look at it with a fresh eye,and it looks as bland as oatmeal. When youwork too closely on the details of an image, andignore its needs as a whole, the details cansometimes have a homogenizing effect. There’sa reason that painters stand back from theirwork. They’re looking at the forest, not the trees.

It helps to think of separating the image into aseries of planes, where one plane reads againstthe next by virtue of its general value. Complextextures such as hair and trees can becomejumbled and unclear when layered over oneanother, so consider separating with an overallvalue shift, or avoid an overlap. Higher con-trast elements tend to come forward, whereaslower contrast sections recede, which is some-thing to think about when you’re trying to di-rect push and pull elements compositionally.

Introductory photography classes teach stu-dents to utilize the entire dynamic range of themedium: to have a perfect black and a perfectwhite in the printed image, which helps givethe picture more impact. This should never bethought of as a rule, though. Don’t think thatevery image should be pushed this way: tohave more bold lights and darks. Some lightingsituations call for a subtle, low contrast look,which can have just as much powerful impactand readability as a high contrast image. Lowcontrast can have a kind of impact all its own.

6 2 P I X E L C I N E M A T O G R A P H Y


If CG lighting in general can be helped by us-ing reference, then lighting for integration intoa live-action plate depends on it. Anytime any-body is trying to trick the audience into think-ing that something artificial is really there, it’simportant to know what it would look like if itreally was there. The key is whether or not itmatches our own eye’s reference, in terms oflighting, texture, motion or any thing else. Theaudience will be able to tell easily whether ornot the digital element fits or not, even thoughthey may not be able to pinpoint exactly why.

Getting the infoWhen lighting for integration, the place to startis there on the live action set. Most of the infoyou need to succeed in integration is there. Thelocation, color, size, and quantity of lightsshould be carefully noted. The environmentsthat may reflect of refract in your objectsshould be photographed. Most importantly,you must get whoever’s in charge to let youshoot reference footage of something thatclosely matches the digital object you’re goingto integrate. If it’s a person, shoot a referenceperson. If it’s a magical object that flies, hold areal-life, non-magical version in front of thelens and roll some film. The point is, when youget real footage of your digital object, you havethe ultimate calibration tool. It’s probably not agood idea to rely heavily on exact light meterreadings because it can be too confining to tryto match the info. If it looks right, it is right,whether the numbers match perfectly or not.

If you decide that you want to improve upon thereality of the reference, by all means, that’s allpart of the fun of the CG business, but at leastwith some reference footage, there’s a place tostart and return to if things get out of hand.

Page 63: Pixel Cinematography

The task of lighting often involves making images clear to theviewer: how well does it read? First-year art student drawingsoften get lost in the details and lack overall impact. The stu-dent gets in close to the piece, painstakingly shading everydetail, step-by-step, in the drawing. But from a distance, it be-comes a complicated jumble of lights and darks with little orno impact.

Image impact

“If it reads as postage stamp, it’ll read as a billboard.”

When the image is “corrected” to separate the majorelements into clearer planes of light and dark, the image hasmore impact.

If we degrade the image by blurring it, the adjusted image stillreads as a person in the foreground, exterior, with buildingsand trees in the distance: it still has depth and space.

The original image becomes illegible more quickly, and doesn’thold up under less than perfect conditions.


C I N E M A T O G R A P H Y 6 3
Page 64: Pixel Cinematography

Texture: blur and grainBlur and grain are more than simple tricks.They can make or break an integration job, andthey can add an organic beauty and authorityto totally CG scenes. These characteristics fallinto the realm of the photographic image, sowhen you add them to smooth and sharp syn-thetic pictures, you get a instant shot in the armof “photorealistic” credibility.

For all the high resolution of 35mm film, it hasan inherent softness, as does the film to tapetransfer process.

In terms of image impact, this scene from a television com-mercial represented an especially difficult problem. A feelingof tremendous complexity had to be portrayed, and three char-acters had to be revealed and understood in the space of onlytwo seconds. With the help of local spotlights, each characterwas emphasized by falling in a pool of brightness. The otherelements were subdued, but not enough to diminish the feel-ing of complex machinery. Behind the camera is a box,mapped with a bright, blurry image of the same kind of envi-ronment. All the outward facing metal surfaces reflect themapped box, helping to define the outlines of the complicatedpaper path clearly.

To give a feeling of richness of light in the space, a few non-shadowing lights project images of warm shadows of varyingsharpness down through the mechanical parts, something thathelps the scene look photorealistic. This kind of solution wasarrived at only by looking at a reference mock-up of realprinter and copier parts in a box.




You have two seconds . . .

6 4 P I X E L C I N E M A T O G R A P H Y

Tricks and tips

These notes try to provide a lighting founda-tion that applies to all aspects of CG imagemaking, but there’s also the matter of just hav-ing the right specialized trick at the right time,Here are some tips that may come in handy.

TelevisionWhen it comes to TV, watch out.

Depending on who you talk to, you’ll find dif-ferent ideas about what’s allowed for televisionimages. Some people never use total white oftotal black, just like my painting instructor toldme. It’s usually ok to use the full dynamicrange and just keep an eye on the scopes to seeif you’ve gone over. When certain images arevery contrasty, such as white titles on black, theonce-acceptable ranges can get illegal forbroadcast because the analog signal starts toget dicey, but you should find out about it per-sonally from someone who’s more qualified toexplain it.

Temporal AliasingThis issue concerns overly sharp images onNTSC, that video standard of choice for televi-sions across America. Diagonal elements in animage can look smooth as silk in a still frame,but as soon as it moves, it can look like a buzzsaw is running along it. This is called temporalaliasing, and it’s a separate issue from ordinaryaliasing effects.

Because of the way video scans and interlacesthe visual information, you have to be very care-ful with the sharpness of moving edges, espe-cially diagonals, in NTSC video. Softening theimage helps diminish the buzzing, but motionblur is what really does the trick. Sometimes it’simpossible to get rid of without destroying theimage, and it’s all you can see, you have to justlive with it. One solution is to render the imageon fields, but field rendering tends to have anelectronic, extra-smooth look to it.

Page 65: Pixel Cinematography

RGB good. NTSC bad.Don’t light for television on an RGB monitor ei-ther. The color and contrast fidelity of RGB ismiles ahead of TV, so you’re just kidding your-self unless you refine the image on a properlycalibrated NTSC monitor. That’s where it’s go-ing to end up inevitably, so bite the bullet andcorrect for it on a television, not your computermonitor. Also, if it looks great on a old blackand white 12-inch Zenith, then it’ll probablylook great anywhere. (It a bit like hearing a

In any detail of aclean, unadulteratedcomputer image, itwill tend to be toosharp, and too cleanto integrate well intoa live-action plate.

Slight softening anda little added grainare a few simpletweaks that make anyCG image look morephotographic.




I Stu


Blur and Grain

great song on a transistor radio at the beach.The tune is so solid, it even sounds good on a2-inch speaker.)

One last thing about NTSC. Very saturatedblue-green colors are outside the color space oftelevision. It simply can’t display certain colors.Teal is one of these difficult colors that is usedin commercial products so often these days, butis just about impossible to get on a television.Teal is difficult to get on RGB, in print, andeven on some film stocks, so watch out for that.

OverexposureWhen the light from extremely bright objectsgoes into a camera, overexposure occurs. Thishappens in your eyes, as well as in a camera, sowhen you see it in printed or projected image,it’s a strong visual cue to your brain that says,“Golly gosh, that’s bright!”

One thing that happens is diffraction, wherethe light passes through a very small apertureand produces a fringe of light, sometimes ap-pearing as bands of light and dark. Anotherconsequence of overexposure is called halation,where the light is so bright that it enters thelayers in the film emulsion and scatters aroundin between the sandwich of the film back andthe halide crystals. This light spreads outwardfrom the local area, resulting in another kind offlaring, blooming effect. A similar effect also oc-curs in your retina, and on the CCD chips ofvideo cameras.

Unless CG artists add this effect, objects thatare supposed to look bright appear merely asflat, white shapes with hard edges.

Fortunately, there’s been interest in these phe-nomena, notably by Alias and more recentlyMental Ray, who can dynamically process orpost process the image to distributesuperbright pixels values into the surroundingareas. Stock lens flare effects in AdobePhotoshop, 3D Studio, Lightwave, and Alias to

P I X E L C I N E M A T O G R A P H Y 6 5

Page 66: Pixel Cinematography

name few, have their place as well, when usedappropriately. If your software doesn’t havethese capabilities, you might want to explorethe compositing method described next.

Diffusion filteringThis is an effect that’s only appropriate in lim-ited doses. It’s like candy, somehow. It’s no bigsecret, only a recreation of putting a diffusionfilter, or a stocking, or a little nose grease overthe lens to soften the image, and its beenaround for ages. Take the original image andblur it significantly, then add a percentage ofthe blurred image back into the original image.

Sometimes it pays to pull a high contrast ver-sion of the original image, then blur it, as thislimits the “flaring” effect to the (logically so)brightest areas of the image. It also might be agod idea to dim down the original image beforethe add operation so you don’t blow out the re-sult. Though it’s not a perfect replacement formore accurate replications of overexposed pixeldistribution, sometimes it’s a good cheat thatworks fine.

Color in shadows and lightThis is one thing that adds an incredible rich-ness to CG imagery. Color changes in complexways as surfaces turn away from light intoshadow. Depending on the kind of environ-ment, it can become more saturated. For in-stance, if your key light is warm, sometimes thefill light looks good if it’s really warm. Skintones benefit from this quality. There is asubtlety added to the image whenever differentlights have colors in them. There’s so much go-ing on in the real world with inter-reflectionand color temperature that computers onlyseem to scratch the surface right now. Even ifthe color difference is very slight, adding itmakes it more interesting.

6 6 P I X E L C I N E M A T O G R A P H Y


What seems most characteristic of computergenerated sets is that a computer simulation oflight and materials is only a rough approxima-tion of the rich phenomena that occurs allaround us in real life. And getting the approxi-mation to look good has a lot to do with havingan understanding of how light and shadows andsurfaces behave in reality, and using this knowl-edge base to create our worlds in the machines.

The computer models only get us halfwaythere. The artistry and craft of this new cinema-tography is recognizing places where the com-puter models fall short, and again, using whatwe know about the real world and referencematerial to guide us as we fill these spaces. Un-less we understand what we’re doing, and whywe’re doing it, the software can and will leadus into a kind of computery-looking medioc-rity. We can groom, tweak and tailor our diora-mas of computer sets until they have a fineness,a sharpness, a resonance: something that feelsright. Eventually, it comes down to eachindividual’s aesthetic about the image. There isno correct or incorrect lighting, but there are in-telligent ways of finding the images we searchfor when we start lighting on the computer.

Someday, the tools for computer lighting maybe so interactive, that we can easily use light it-self as the imaging tool, as in the Sorcerer’s Ap-prentice portion of Walt Disney’s “Fantasia”. Inthat sequence, the shadows on the walls aren’tjust the light and shadow of an image, they arethe subject. No matter how sophisticated thetools get, making the images succeed will stilldepend on us, and knowing how to best usethose tools to get what we want.

Page 67: Pixel Cinematography

BibliographyPainting With Light, by John Alton, Universityof California Press, 1995 (first published 1949).

The Illusion of Life: Disney Animation, by FrankThomas and Ollie Johnston, Hyperion, 1981.

Principles of Traditional Animation Applied to 3DComputer Animation, by John Lasseter, Pixar;SIGGRAPH course notes 1987.

Tricks to Animating Characters with a Computer,by John Lasseter, Pixar, SIGGRAPH coursenotes 1994.

Toward High Level of Control of Computer Gener-ated Animation, by Chris Wedge, Ohio StateUniversity SIGGRAPH course notes 1987.

Animation Tricks, by Chris Wedge, Blue Sky Pro-ductions, SIGGRAPH course notes 1994.

Envisioning Information, by Edward R. Tufte,Graphics Press, 1990.

Special Thanks:Tom Berryhill, Cindy Brolsma, SharonCalahan, Dave Carson, Alex Levinson, LutzMuller, Dennis Murren, Ellen Pasternak,Steven Poster, A.S.C, Nina Rappaport,Sam Richards, Scott Senften, Chris Wedge.

Carl Ludwig, for a deeper understandingof light.

Blue Sky Productions

Design and layout by Jeanne Markel

Myst is a registered trademark of Cyan, Inc.Copyright ©1993 Cyan, Inc. All rights reserved.Images used by permission

P I X E L C I N E M A T O G R A P H Y 6 7

Page 68: Pixel Cinematography

6 8 P I X E L C I N E M A T O G R A P H Y

Page 69: Pixel Cinematography


By Dave Carson

IntroductionMost of us probably remember the first time werendered an image using 3D computer graph-ics. Our initial reaction was probably some-thing like ‘Wow! This looks really neat!” fol-lowed immediately by a second reaction; “Iwonder how I can make this look more real?”Though reality is only one of the goals avail-able to the computer artist, it frequently is theone assigned to (or chosen by) us.

Attaining a realistic looking image is difficultunder any circumstances, but the most de-manding environment of all is combining acomputer-generated image with a photo-graphic background. Here, by comparison , anobject that seemed realistic against a computer-generated background now may look artificialand somehow out of place when combinedwith the real-world image.

Since the very beginning at Industrial Light &Magic, we have been in the business of puttingtogether photographic backgrounds and indi-vidual elements which have been created else-where. Originally, these elements were usuallyminiatures or artwork, though sometimes theymight be actors or real objects filmed outsidetheir ultimate environment. In later years how-ever, these elements have increasingly becomecomputer-generated imagery.

Whether a natural or computer graphic object,many of the tricks for lighting and combiningthese element in a photograph are the same. Be-low are listed the steps that we typically gothrough at ILM in the process of putting to-gether a photographed background plate andcomputer generated imagery into one final shot.

Obviously, lighting plays a large part in this pro-cess, but in order to judge the success or failureof the lighting scheme under way in the com-puter, you have to combine the images at handand to evaluate your progress. Therefore, I thinkit is also valuable to cover some of the other as-pects of image generation and compositing, inorder to give an overview of how they all play apart in lighting and compositing realistic com-puter-generated images.

Notice that I am presenting ten steps in this jour-ney from planning the image to the final paintedtouches. But rather than being seen as adding tothe complexity of creating a realistic image, eachof these steps can actually be viewed as a way tosimplify the process of creating a photorealisticimage since each can offer an opportunity tocompensate for some of the shortcomingspresent in our current rendering systems.

Also, note that at ILM we typically deal inmoving images, so most of the references herewill assume the CG images are being put into amoving filmed background. Of course, most ofthe principles here would equally apply whenthe end product is a static photograph.

P I X E L C I N E M A T O G R A P H Y 6 9

Page 70: Pixel Cinematography

I. Designing the shot:When a shot is to be combined from elementscreated at different times, it is especially impor-tant that some form of ‘blue print’ be createddefining the intended final image. Productiondesign and storyboards are most often used toaccomplish this. The production design conceptart is important in establishing the look andmood of the shot (or of an overall sequence),while storyboards should define camera action,shot duration, and be fairly faithful to perspec-tive and image content. Both are used to deter-mine what will be needed to generate the finalshot, and in planning what elements should berecorded during original photography andwhich elements are best added in the post-pro-duction process.

II. Filming the background image (the plate).Special considerations involved in shooting thebackground image include choosing the filmformat, learning to compose for the missing el-ements, and deciding which environmental ele-ments should be left out for addition later. Eachof these aspects vary on a shot-by-shot basis.

During background photography, all necessarydata required for recreating the environment inthe computer should be gathered. This data in-cludes recording the physical attributes of theset (or location) as well as gathering informa-tion that might help in recreating the lightingpresent. This would include noting the locationand nature of the various light sources, as wellas shooting some reference images which pro-vide visual information as to the nature of thelighting being used.

7 0 P I X E L C I N E M A T O G R A P H Y

III. Generating a computer graphics model ofthe background environment and camera.Using the data gathered on the set, and thebackground film itself, a computer model of thereal environment can be created. This virtualset should include scale models of all objectswhich might interact with the intended com-puter model, any objects which might have castshadows on them from the computer-generatedobjects, and any light sources which werepresent during photography.

Also, a camera has to be created which matchesthe position, focal length, and if pertinent, themotion of the taking camera. This re-creation ofthe camera motion during photography canusually be done using a combination of thedata gathered during background photographyand a digitized version of the filmed back-ground itself.

IV Building the computer model to look real.How much detail does it take to create a realis-tic computer model? In most cases, when com-pared to a photorealistic environment, it takesmore detail than is practical using today’s sys-tems. Therefore, much of the modeling processhas to involve creating surfaces which will laterhave texture and bump maps applied to them.

Of the portions of the details which will rely onmodeling, aspects of the real world whichshould be taken into account are asymmetry,rounded edges and other common worldly im-perfections. Special attention needs to be paidto these aspects of the modeling in particularbecause these are the types of things that donot come naturally in the computer environ-ment, but are seen everywhere in nature.

Page 71: Pixel Cinematography

V. Posing and animating the model in the cgworld.Once the ‘real’ world has been faithfully recre-ated, and the new computer models are built,they can be posed and animated per the re-quirements of the shot. Once this phase is com-pleted, the task of creating photorealistic light-ing and rendering can really begin.

VI. Lighting the model.Matching the placement, nature, and intensityof the key lights that were present during origi-nal photography is an obvious first step in thelighting of the computer model. In addition tothis however, many other aspects of the lighton the set or location have to be matched in thecomputer. These include bounce cards, bouncelighting in general, duplicating the effects ofcukes and flags, and recreating general atmo-spheric conditions. Obviously the accuracy andamount of data gathered at the time of initialphotography can make a big difference in thetime required to set up these parameters in thecomputer.

Some of these aspects of the on-set lightinghowever are still not handled very well withcurrent 3D rendering systems, and therefore arebest added by eye during later stages.

VII Rendering the computer graphic image.The rendering phase takes all the informationassigned to the computer model and the virtualworld it exists in and creates a pixel representa-tion of what that object would look like. Usu-ally, this is done through some variation of raytracing. The primary limitation of ray tracing asit currently exists is the exclusion of bouncelight and ambient randomness which is soprevalent in real world lighting. Though a flatambient level can be set, this is usually an un-successful substitute for the nature of ambi-ent light in the real world.

By Contrast, some of the newer approaches torendering which do take the effects of bouncelight, or radiosity, into account, currently lacksuch vital elements as reflectivity, glossiness,and motion blur. Overall atmospheric perspec-tive may also be lacking.

Whichever of these two approaches is used (orwhichever combination of the two is utilized),some allowance must be made for their indi-vidual or combined limitations.

It is also at this stage that the alpha channel orcompositing matte is created for the computerelement. This is a key factor in how success-fully the element can later be combined into thebackground image, and again its quality is de-pendent on how well it is handled by the ren-dering system being used.

Finally, for moving images or moving objects,motion blur should be calculated and renderedat this stage. Currently, the accuracy of motionblur generated by different systems varies in itsquality and accuracy.

VIII Compositing the image.More than just a combining of two elements,the compositing stage offers an opportunity toaffect the incoming element in many subtleways which can lend it additional believability.In particular, any adjustment to black levels,color aspects and edge characteristics whichmight initially fall into some of the other cat-egories mentioned here can sometimes betterbe addressed by manipulation during thecompositing stage. Also, the appearance ofshadows is usually determined by the way inwhich they are composited into the backgroundimage.

P I X E L C I N E M A T O G R A P H Y 7 1

Page 72: Pixel Cinematography

Finally, the image and its motion blur are com-bined based on the alpha channel generatedduring the rendering stage. This involves somepercentage of opacity and the quality of theblurred edges of a moving object are heavily af-fected by the way they are handled during thecompositing phase.

IX Adding overall effects to the shotMany times some atmospheric effect will coverthe entire shot, overlapping both the back-ground plate and the computer element to beadded. In these cases, a tradeoff has to be madeas to whether to face the challenge of recreatingthe effect over the CG element only, and riskcreating a visible blend line, or whether it’s bet-ter to add the overall effect in the post produc-tion process, and give up the nuances of realitywhich would be present if the atmospheric con-dition was generated on the set. The types of‘effects’ referred to here include elements suchas rain, smoke, fog, etc.

In addition, issues of creating matching filmgrain, image filtration, and post-process cameramoves also fall into this same category.

X Paint as the final step:Attempting to create a flawless scene whichcombines computer graphics and photographyis an iterative process with the goal (hopefully)coming closer with each iteration. At somepoint, it can become more feasible to simplypaint out the remaining imperfections (or addthe missing enhancements) using digital paint-ing techniques. In a moving scene, these addi-tions have to be very carefully applied or theycan call more attention to themselves than theproblem they set out to address. However,when used successfully, they can solve prob-lems which might have taken significantlymore time to eliminate in the 3D world.

7 2 P I X E L C I N E M A T O G R A P H Y

SummaryAs you can imagine, each of the steps touchedupon here could in themselves be the basis foran hour-long discussion. Hopefully howeverthis overview will give some insights as to howILM approaches the challenge of creating real-istic computer graphic images and combiningthem with photographic backgrounds, and hasgiven you some ideas on additional factors toconsider in approaching such work in yourown projects.

Page 73: Pixel Cinematography

P I X E L C I N E M A T O G R A P H Y 7 3