[carpents]

It totally shouldn't, but setting a custom white balance when shooting infrared makes a big difference on RAW captures.

I've taken two shots, one with tungsten white balance and one with a custom white balance, then processed them identically in ACR. The tungsten white balance produced a noticeably redder result. With the custom white balance it was (by my very rough estimate) 500 degrees cooler. This makes a big difference since I was developing IR at the very low end of ACR's temperature scale, around 2000 degrees. More importantly it impacts the relative saturation of colors (I'm guessing here) so that it is hard/impossible to even get the two to match at different temperatures.

Below are two recent samples shot using a custom white balance. Much different than my previous IR attempts. (Anyone who can answer why it would matter for RAW captures, I'd like to know. It totally shouldn't make a difference.)



(Images are click-able for bigger size versions.)

Previous example, using tungsten white balance:

[SWEngineer]

Sean, I don't have any answer, but I'll guess some for the heck of it.

Any chance the WB setting influences light metering? I would expect IR shot color temperature to be quite sensitive to the amount of light energy reaching the sensor behind the RGB filters.

Maybe you could repeat this in a more controlled setting: Identical scene from a tripod and shot w/ manual mode. Make camera WB the only variable.

Regards.

-Mark

[carpents]

Sean, I don't have any answer, but I'll guess some for the heck of it.

Any chance the WB setting influences light metering? I would expect IR shot color temperature to be quite sensitive to the amount of light energy reaching the sensor behind the RGB filters.

Maybe you could repeat this in a more controlled setting: Identical scene from a tripod and shot w/ manual mode. Make camera WB the only variable.

Regards.

-Mark

Hey Mark!

My samples don't show it, but this is what I did to determine the difference. Same setting, same light, same exposure. WILDLY different RAW conversions.

Strange. RAW isn't raw after all?

[jfdavis58]

I believe this was discussed some time ago on DPReview. If I recall, some fellow compared RGB histograms for various WB settings and found that all channels had some shift-red most of all. There was no final conclusion; except that RAW isn't exactly what most assume it to be. There must, in fact, be some response curve used to convert the analog sensor output to a digital form. It seems that WB might have some influence on this curve.

Hey Mark!

My samples don't show it, but this is what I did to determine the difference. Same setting, same light, same exposure. WILDLY different RAW conversions.

Strange. RAW isn't raw after all?

[carpents]

I believe this was discussed some time ago on DPReview. If I recall, some fellow compared RGB histograms for various WB settings and found that all channels had some shift-red most of all. There was no final conclusion; except that RAW isn't exactly what most assume it to be. There must, in fact, be some response curve used to convert the analog sensor output to a digital form. It seems that WB might have some influence on this curve.

Well, this would certainly make sense as it is much more apparent when shooting IR, since the vast majority of the information is captured in the red channel. Thanks for the tip.

[jeffkrol]

Well, this would certainly make sense as it is much more apparent when shooting IR, since the vast majority of the information is captured in the red channel. Thanks for the tip.

I tried to re-open this topic at dpreview and was met by silence...
Nobody care to re-open this topic???: Pentax SLR Talk Forum: Digital Photography Review
and


Re: RAW vs White balance: Pentax SLR Talk Forum: Digital Photography Review
I couldn't quite make out what was the consensus but it did seem to be "model dependent". But it does seem that some RAW has the custom WB applied directly to the data, it while normal "WB" is just a tag in the file. But I wasn't sure

[shutterdrone]

Are you creating your CWB from a filtered shot (e.g. CWB -> then using an image taken with the filter on and making something white), or using an unfiltered shot and making something non-white white in the CWB? Just wondering what technique you were using for CWB.

!c

[breischl]

Are you creating your CWB from a filtered shot (e.g. CWB -> then using an image taken with the filter on and making something white)

Maybe I'm showing my total ignorance of IR photography here (and it is total), but is there a defined "white" in IR? Wouldn't the "color" of a gray card change based on how hot it is? Or do you just make sure the card is at ambient temperature?

Just curious...

[Cideway]

Maybe I'm showing my total ignorance of IR photography here (and it is total), but is there a defined "white" in IR? Wouldn't the "color" of a gray card change based on how hot it is? Or do you just make sure the card is at ambient temperature?

Just curious...

IR Photography, as we know it has nothing to do with heat, heat is the Far IR light, while what we capture with our digital is Near IR, which is more like our Visible red light but longer. The grey card in all likely hood wouldn't be useful at all for IR as chances are the pigments in it don't reflect IR neutrally.

[carpents]

I set the custom white balance for IR on the *ist D using a shoot-and-pray method. I believe I had the most success pointing the camera at a neutral car dashboard. Since then the *ist D has kept that custom setting and I use it, not having to re-calibrate for every scene.

[khardur]

carpents:
I tried telling you a while back that it made a difference. I'm glad you figured it out

I'm assuming it must tone down certain channels that are picking up the most light. Really, how would a white balance work? Try to get all 3 channels to register equally, which is why when you do a WB (for normal work) you shoot a gray card, or something equivalent.

So why wouldn't it try to tone down the channels some? I know it's still RAW, and in theory RAW should just record data, and only color balance as an afterthought so the post processing program has a clue what you were doing - but as noted by others it seems to affect the actual data recorded as well, which is why your RAW results using a custom white balance are different.

[carpents]

carpents:
I tried telling you a while back that it made a difference. I'm glad you figured it out

You'll remember that I didn't say you were wrong, but obviously I felt the need to verify!

PS - You should have been wrong, but were not.

[khardur]

I know, you didn't say I was wrong, you wanted to check it out for yourself. (as it should be - so many people take what certain others spout off as gospel without checking themselves...)

And you're right that the whole thing should be wrong.
Have a good one.

[jeffkrol]

carpents:
I tried telling you a while back that it made a difference. I'm glad you figured it out

I'm assuming it must tone down certain channels that are picking up the most light. Really, how would a white balance work? Try to get all 3 channels to register equally, which is why when you do a WB (for normal work) you shoot a gray card, or something equivalent.

On My D I can get the multipliers (which is ALL WB should be and if Pentax does it different, they are WRONG in what there doing) to 1:1:1 (well it's real close most of the time at 1.04;1;1ect)by using a cc40M filter and shooting a grey card. Any "normal " wb will not be close to 1:1:1.... Nikon has a table of all the coefficents it uses for WB settings and none approach 1:1:1. Nikon does have the ability to "input" the coefficents of your choice, thus it's pretty easy to get 1:1:1. If Pentax screws up the RAW by some modification of the data w/ WB then they are really messing things up. Tell them to stop

els some? I know it's still RAW, and in theory RAW should just record data, and only color balance as an afterthought so the post processing program has a clue what you were doing - but as noted by others it seems to affect the actual data recorded as well, which is why your RAW results using a custom white balance are different.

[pentkon52]

Are you creating your CWB from a filtered shot (e.g. CWB -> then using an image taken with the filter on and making something white), or using an unfiltered shot and making something non-white white in the CWB? Just wondering what technique you were using for CWB.

!c

I create my CWB with filter on, and a shot of brightly lit green grass. I then get the mildly magenta'd image, as opposed to the AWB image that is mostly red in tint.
Similar effect as the first when using a white card and no filter.

Dave