[Radiance-general] HDRI - Camera Response Curve

[Jack de Valpine]

Hi Greg,

Thanks for the follow-up on this. I think that I have gotten the white 
balance part correct. I always set it by hand (for example to the sun 
icon) so the auto white balance feature is effectively off. And I have 
used aperture priority on the camera however I have used auto 
bracketing, so maybe I am not getting enough samples? It sound like 
perhaps the best thing to do is set the aperture manually and vary the 
exposure time manually as well in order to generate your suggested ~10 
samples. Although doing everything manually involves a lot of 
unnecessary touch of the camera.

So as to scene, clear sunny sky conditions would be ideal I suppose with 
a range from shadowed features to direcly illuminated features... Sorry 
for this next... So how do you use the luminance measure from the 
reference card to further calibrate the response curve?

Once a response curve has been generated is there some way to check or 
validate it (if that makes any sense)?

In the book (the new one;-) you also indicate that the darkest exposure 
should have no RGB values greater than ~200 and the lightest no values 
less than ~20. I could certainly figure out a filter routine with pcomb, 
however there must be a simple way to do this with ImageMagick. The 
point being to run a quick preprocess check on the the bounding images. 
Any takers or suggestions for how to do this easily?

-Jack

Gregory J. Ward wrote:

> Hi Jack,
>
> I've just been playing around with this, myself.  To get a good  
> response curve, it is best to start with a scene that has a daylight  
> color balance, as this is the design point for most cameras, and some  
> models end up boosting the blue too much if you calibrate under  
> incandescent lighting.  For this reason, I prefer calibrating with  
> natural light, as opposed to a controlled condition with incandescent  
> or fluorescent lighting.  In any case, you should fixed a white  
> balance setting on the camera appropriate to the test lighting.   
> (Memory aid:  If the white balance isn't fixed, then it's broken.)
>
> Color charts are of limited use, except as a means to verify your  
> calibration.  The more images you take in a sequence, and the more  
> closely they are spaced, the less you rely on the recorded camera  
> response function.  This is because the overlap of the images serves  
> (together with the known shutter speeds) to give you an accurate  
> result, regardless.  It is more important to have an absolute  
> calibration value if you are after real numbers, and for this a  
> luminance measurement on a reference card of known reflectance is  
> invaluable.
>
> -Greg
>
>> From: Jack de Valpine <jedev at visarc.com>
>> Date: January 9, 2006 9:19:54 AM PST
>>
>> Hi All,
>>
>> Happy New Year first off.
>>
>> I am making a first pass through the "High Dynamic Range Imaging"  
>> book that Greg co-authored. I am wondering about methods for  
>> generating a good response curve for a given camera. The book gives  
>> a variety of tips and suggestions, however I am curious about good  
>> scenes.
>>
>> In a prior email footnote from Greg (a faq item relating to camera  
>> response curves and Photosphere), he suggests shooting a scene  
>> looking out a window during daytime with about 10 exposures. What  
>> about shooting a controlled scene such as a Macbeth Color Checker  or 
>> a Kodak gray scale chart (can't remember the name of this)  outdoors 
>> under daytime conditions or using some type of interior  fixed 
>> lighting setup?
>>
>> -Jack
>
>
> _______________________________________________
> Radiance-general mailing list
> Radiance-general at radiance-online.org
> http://www.radiance-online.org/mailman/listinfo/radiance-general
>
>

-- 
# Jack de Valpine
# president
#
# visarc incorporated
# http://www.visarc.com
#
# channeling technology for superior design and construction