max phillips a5 hdri report

8/3/2019 Max Phillips A5 HDRI Report

1/14

High-Dynamic Range (HDR) imaging is a modern photographic techniquethat utilizes computer technology to achieve a dynamic range (i.e. a range of luminance) much closer to that actually experienced by human eyes thantraditional, Low-Dynamic Range (LDR) digital photography is able to achieve byitself. Singular images from typical consumer grade digital camera have amaximum contrast ratio of only about 100:1, whereas the human eye tops out atapproximately 16,000:1 about 80 times greater. The result is that details atvery high, or very low luminance that humans can usually discern with the eye,are effectively lost in the digital image. The camera depicts highly shadowedareas as simply solid black, and highlights over a certain luminance areoverexposed and appear plain white.

It is the goal of HDR photography to correct this shortcoming. The way thisis done is by using data fusing software to combine sequences of singular LDRphotos with varying exposures into a single image that exhibits much morerefined extremes, more similar to what humans actually.

Although the main use of HDR imaging is for generating print images forviewing, the technology can also be used in other applications, such as simplephotometry. HDR happens to be a convenient solution for a very old problemwith traditional photometric devices and, specifically luminance meters , in thatthey can only measure one point at a time, and are fundamentally ill-suited toaddress luminance over an area, such as a continuum, or gradient of luminance.On the other hand, a camera can observe a whole scene all at once, effortlesslyaccomplishing what is impossible for a luminance meter.

Using cameras for photometric measurement is a good fit because themechanism for image capture is fairly similar to the way an eye actually works. Acamera is basically an illuminance-meter that can precisely differentiate withrespect to angle and spectrum. It is exactly this precision that makes it possible

to infer (with the aid of computer algorithm) values of luminance.

Field Measurement of Luminance Using HDR Photography

Max Phillips ARCE 651 Fall, 2011


2/14

Luminance distribution can be applied to fluxleaving a light source, or converging oncamera. Luminance units apply to both.

cd m2

=

lx sr

Luminous Flux reflects from surroundingsurfaces toward the eye/camera. If a givensurface is at spherical coordinates ( , ) then

d off = d on

This Flux is at a certain density.

M fromsurface

= E onsurface

= E oncamera

=

d o

A and a certain intensity

L ( , ) = d

2on

d dA

The significance of the above expression isthat, just like the eye, in order to form animage, a camera has to differentiates fluxdensity with respect to incident angle (whichcontrols where surfaces appear in theimage).

Differentiating with respect to solid anglegives a luminance distribution as a functionof the spherical position coordinates.

This formulation is only theoretical and

would not work in practice because of the high degree of discontinuity insurface characteristics for a givenscene.

These two light paths,although opposite in direction,can be described using similarmathematics.

Light direction corresponds tolocation of source in sphericalcoordinates, ( , )


3/14

Camera:Cannon EOS 550D

CANNON EOS 550D (REBEL T2i)TYPE: Digital, Single-Lense reflexEXP.COMP:

Manual: 5 stops in 1/3 or 1/2 stopincrements

SHUTTER: 1/4000 to 1/60

Lens: Sigma 4.5mm F2.8 EX DC HSM Circular Fisheye

SIGMA 4.5mm F2.8 EXDC HSM CIRCULARFISHEYE

TYPE:PrimeLenseVIEW

ANGLE: 180APERTURE:

MAX F2.8MIN F22.0

ELEMENTS: 13

Equipment


4/14

Illuminance Meter: Minolta T-10


5/14

Color Meter: Minolta CL-200A


6/14

Luminance Meter: Minolta LS 100/110


7/14

Additional Equipment/Software

Tripod

Macbook(need OSX for Photosphere Software)HDR Fusing Software

Photosphere

Color Rendition Chart(Macbeth) ColorChecker


8/14

Photosphere

Only for Mac OSXAllows one to create and calibrate HDR images from input LDR images.

Tripod

NOTE: THIS IS NOT AN ACTUAL SERIES OF EXPOSURE-BRACKETED IMAGES. IT IS FOR ILLUSTRATIVE PURPOSESONLY.


9/14

Focal Length

Aperture Diameter

"

! !

!"#$% %&'()*$+&,)-,& ./$0&)&,

f-number:Speci es size of aperture

typical range is f2.8 - f22.0

Aperture - f-number

drawn by Max PhillipsLarger f-numbers correspond to smaller aperture, and less exposure

F-numbers are often called F-stops in the photography eld. This refers to the factthat a camera lense has speci ed apertures (typically six, from ~2.8 to 22.0).

Large Aperture; Small f-number

Small Aperture; Large f-number !"#$%"""""""""""""""""""!"&$#"""""""""""!"'$("""""""""!")$*"""!"*$'&""!"%$("""""


10/14

Shutter Speed

Fast shutter speed is great for fast

moving objects, but only whenthere is plenty of light. At night,slow shutter speed can be used,but it can create a blurred image.

SLOWFAST


11/14

Exposure Value (EV)How much light will be admitted to the lm by the combination of the lense aperture and the shutterspeed. Since exposure is proportional to shutter speed, but inversely proportional to f-number, lines of constant EV are de ned by,

which can be proportioned on the abscissa, and ordinate to create the following chart, with lines of constant exposure value at 45 degree diagonals.

"

!" # #$ ! %&'() ! )*+( " !" ! #$


12/14

HDR Imaging Procedure

The digital camera settings were controlled manually, meaning the aperture andshutter speeds where chosen deliberately by the photographers. For thisprocedure, the aperture was chosen initially based on the best picture focusquality. The EV was varied by changing the shutter speed only, while keeping theaperture constant. The shutter speed ranged the entire range of the camera:from 1/60 to 1/4000.

The scene chosen was the engineering courtyard, in the morning, with anovercast sky. This was important because, it allowed for the lighting conditions toremain more or less constant for the duration of the procedure.

Mounting the camera on a tripod, the procedure utilized DSLR Remote Prosoftware to control the camera settings directly from a laptop. This was importantbecause adjustment of the settings directly on the camera can result in camera

shake, and without a heavy-duty tripod, this will affect the alignment of theimage. Although image alignment algorithms exist; however, none are perfect,and the best images are made without the need for these.

After the camera had taken a series of LDR images at variable shutter speeds, itwas necessary to take traditional photometric measurements, as well asdimensional/geometric measurements of the scene in order to calibrate the finalHDR image. The distance of the camera from the wall was 5.81 meters. This wasdetermined using the Laser Distance Meter. Using the Minolta CL-200 color meterthe average CCT was found to be 5800. Next, the luminance was measured for aa checker mounted on a vertical surface and also on a horizontal surface. Usingthe LS-100 meter, the Luminance was measured six times for the vertical surface(I dont know why it was done six times. The typical number is three readings).The average of these was found to be 295.3 cd/m2. The rest of the photometricmeasurements were taken in three. The average of the three horizontalluminances was 243.9 cd/m2. Next, the illuminance on these surfaces wasmeasured (also in threes). On the vertical the average was 4488 lx, and on thehorizontal surface the average was 3981 lx.

The photometric values were recorded, and the LDR images were saved.Back in the lab, the series of images was opened in Photosphere. From this series

of images, we had to select a smaller group of photos to use for the HDR image.The criteria for this was the Radiance Synthetic Rendering System developed byGreg Ward. The RGBE value needed to fall within the range of greater than 20,and less than 200. This information was determined using PhotospheresHistogram function (View > Histogram), which shows the response curve of theselected image, as well as RGB and Luminance values.

Once the range of suitable photos were determined, we selected them, and thenwent to File, Make HDR.


13/14

If the photos were taken very perfectly, or if the user is very lucky, thenPhotosphere may be able to fuse the images into an HDR image in one try;however, if Photosphere is unable to do so, one must narrow the range bychoosing less photos (but at least three).

After Photosphere generates an HDR image, it is necessary to check the

luminance value for a consistent point in the scene. Once can zoom in on aspecific point in the scene (e.g. a grey square on the Macbeth chart) and use thePhotosphere Histogram to find the average luminance. Also, this is where thefield measurements are needed because often, Photospheres luminance valuewill be different than that measured with the luminance meter itself. It isimportant to use the actual luminance measured with the meter as the morereliable value and to calibrate according to it.

The Calibration Factor is

After the HDR image iscalibrated to the correctluminance values, one canuse Photosphere toanalyze these values forthe purpose of design. Thefalse color feature allowsone to view the image in away that it is easier tocomprehend the differentluminance values. This canbe helpful in determiningplaces that are notsufficiently illuminated, orpotential places of glare.

CF = L meter

L HDR


14/14

max phillips a5 hdri report

Documents