Custom White Balance Problem

[Gerard2]

I dont have this problem. Could it be camera or OS dependant ? I use windows and canon 20d.

PS: first post here. Also moved from the bibble forum, managed to keep the same name.

Gerard

[springm]

Could well be that it happens not with all cameras. That's why I have submitted the raw file.

Markus

[rawshoter]

I have the same problem with Canon EOS 5D and as well as with Nikon D5100 Raw-files on WinXP.
Clicking on a white wall causes always a strong green color cast. The images where all made under a bulb light, so the in camera wb is expectedly much to reddish/yellowish.
But I found out that you have to click on green areas if possible (Plants, Leafs etc..) to get a proper wb. Well, sort of workaround.
I already send my bug report yesterday.

Tobi

[Denis de Gannes]

I checked the original poster's raw file and get the same problem on my system. I have not noticed the problem with any of my own Panasonic G3 files. Could be related to the lighting under which the photo was taken.

[springm]

Denis, it most probably *is* related to the lighting. But Custom White Balance with a color picker should just resolve this situation. And the grey picker of the curves tool does not show this bug

Markus

[Denis de Gannes]

The image is also taken at a high ISO 1600 and there is quite a bit of noise. I believe this noise and the size/range of pixels selected by the picker is responsible for this. Well over a year ago I inquired what range of pixels Bibble used and if it was possible to make adjustment. I recall someone saying it was one pixel and not adjustable. By comparison the default in Lightroom is 5x5 pixels. Maybe this is the problem. Certainly needs to be addressed.

[springm]

Denis, a too small sample size could be an explanation. For the grey picker they seem to set a more reasonable picker size.

[afx]

Denis, a too small sample size could be an explanation. For the grey picker they seem to set a more reasonable picker size.

picker size is identical....

cheers
afx

[springm]

picker size is identical....

Then the white balance algorithm seems to have a problem...

Markus

[claudermilk]

If it's selection size, then they shrank what "Large" means. I had WB troubles years ago on Bibble 4 & 20D files, and moving to a large sample size fixed it; ASP seems to have defaulted to large, yet shows this problem. The files I've tried it on are high-ISO from a 7D (I'm also on Win7, so that isn't it). Right now for me it's not much of an issue as I set WB in-camera so rarely use this any more, but on the occasion I need it this bug will suck.

[Denis de Gannes]

I can also reproduce this green tint problem with one of my Panasonic G3 files. I have submitted a bug report on the required form.
Hopefully this problem will be addressed soon.

[grubernd]

there is one miconception that some "we hide what we do" companies introduced.
click White Balance does not have to result in mathematical neutral values.
that's what the curves are for.
WB just determines the way the green vs red vs blue values from the sensor are put in relation to each other.
depending on the algorithm this might lead to different results.

the more extreme the light is away from ideal values, the more differences you will see.

i get very varying results, depending on where i click in the image.
this tells me that the light was very bad (call the museum).

mind: there might also be something wrong with the G3 calibration,
and the WB algorithm probably can quite surely be improved to handle off-cases more gracefully.

[springm]

there is one miconception that some "we hide what we do" companies introduced.
click White Balance does not have to result in mathematical neutral values.
that's what the curves are for.
WB just determines the way the green vs red vs blue values from the sensor are put in relation to each other.
depending on the algorithm this might lead to different results.

OK - that is new for me. I thought that the grey point evaluates the luminance of that spot, and black and white do the same for highlights and shadows. Up to now, with my Sony A700, that method worked quite well. So most probably the calibration for the G3 needs some improvement.

Am I correct to assume that one needs to do both - first whitebalance, then pick grey?

Markus

[Denis de Gannes]

there is one miconception that some "we hide what we do" companies introduced.
click White Balance does not have to result in mathematical neutral values.
that's what the curves are for.
WB just determines the way the green vs red vs blue values from the sensor are put in relation to each other.
depending on the algorithm this might lead to different results.

the more extreme the light is away from ideal values, the more differences you will see.

i get very varying results, depending on where i click in the image.
this tells me that the light was very bad (call the museum).

mind: there might also be something wrong with the G3 calibration,
and the WB algorithm probably can quite surely be improved to handle off-cases more gracefully.

I am not an expert in this field but I have been using Raw conversion software since 2004 and I do not normally see problems like this. The Corel/Bibble team need to do some homework on this one. Attached is the Lightroom rendition with a white balance click.

[rawshoter]

The Corel/Bibble team need to do some homework on this one.

full ack.
The WB-Picker just don´t does for what it is intended for. It´s shifting the colors in a complete wrong way.
In particular for people, like myself, who are doing a lot available light shootings, is this a very annoying bug.