Call for white balance finetuning samples

[zorgtool]

Hi everyone.

Sorry to bump up this thread.

But just get my new camera and try to help opensource software, so trying to do this white balance by following White balance presets · darktable-org/darktable Wiki · GitHub and this thread.

But I’m still confuse…my camera is a Fuji, so the fine tune is like this image

So should I go from horizontal line (R: -9: B: 0 to R:9 B: 0) diagonal up (R:- 9, B :-9 to R: 9 B: 9) diagonal down (R:-9, B: 9 to R: 9 B: -9) or the whole square (200 shots…!!!)

[johnny-bit (Hubert Kowalski)]

Hi,

Unfortunately fuji does not produce usable finetuning info. Your only option is to just create presets without finetuning. If you did finetuning here’s the message you’d receive from extract_wb tool:

Warning: Fuji does not seem to produce any sensible data for finetuning! If all finetuned values are identical, use one with no finetuning (0)

@zorgtool

So should I go from horizontal line (R: -9: B: 0 to R:9 B: 0) diagonal up (R:- 9, B :-9 to R: 9 B: 9) diagonal down (R:-9, B: 9 to R: 9 B: -9) or the whole square (200 shots…!!!)

This is explained in White balance presets · darktable-org/darktable Wiki · GitHub article, but thanks to various manufacturers variably vary interface it’s not as straightforward as it should be…

So in case of fuji, if they ever decide to actually put finetuning info in exif data, the correct axis is: vertical horizontally centered.

As you can see o top in vertical it’s blue and bottom on vertical is warm-ish yellow-orange. That’s amber-blue axis.

Horizontal line is from green to weirdo pink in your example and it’s the one we don’t want (magenta-green)

[zorgtool]

@johnny-bit

Unfortunately fuji does not produce usable finetuning info. Your only option is to just create presets without finetuning. If you did finetuning here’s the message you’d receive from extract_wb tool:

Warning: Fuji does not seem to produce any sensible data for finetuning! If 

Well…it doesn´t I try with a bunch of files (event with the wrong selection of finetune, I was doing horizontal change) it produce a tar.gz with files .txt.

@johnny-bit

So in case of fuji, if they ever decide to actually put finetuning info in exif data, the correct axis is: vertical horizontally centered.

As you can see o top in vertical it’s blue and bottom on vertical is warm-ish yellow-orange. That’s amber-blue axis.

Horizontal line is from green to weirdo pink in your example and it’s the one we don’t want (magenta-green)

Are you sure ? Because the problem (maybe it’s because my english suck) but if I go vertically, center horizontally I will go from (R:0 : B: -9 to R:0 B:9) and for me R stand for «red» (close to Amber) and B stand for «blue».

So can you just confirm it’s really what you need (R:0 B:-9 to R:0 B:9) without any change on R.

[johnny-bit (Hubert Kowalski)]

@zorgtool

Well…it doesn´t I try with a bunch of files (event with the wrong selection of finetune, I was doing horizontal change) it produce a tar.gz with files .txt.

The tool producing .tar.gz files is extract_wb_from_images.sh, this just extracts exif data to txt files so you can send it as a package to darktable devs who’ll ran extract_wb.py on them… It’s as simple as possible and makes no detection of fuji fun times. It’s extract_wb.py that does the heavy lifting and complaining about fuji not having normal data

@zorgtool

Are you sure ? […]
So can you just confirm it’s really what you need (R:0 B:-9 to R:0 B:9) without any change on R.

Yeah, I’m pretty sure. I wrote the thing
And I can bet that there won’t be proper finetuning data in exif output from fuji files, but I’d be happy to be mistaken

[zorgtool]

Hi @johnny-bit .

So you’re right, fuji doesn’t change anything. I just try both fine tune and lazy. And yes in the fine tune no data.

I will try the lazy mode

Z