Problem with Lights Kelvin and VFB White Balance

[Ondra]

ok, please test everything you can think of, so there is no hotfix 2 ;)

[Dionysios.TS]

Now I understand why Ondra did that change in 1.5. I tried the hotfix and it works, yes, you can go back to the "old" 1.4 method and colors under 5.000K seem correct BUT...., highlights are so crap and weak while the 1.5 method alters the colors it gives superior highlight's compression.

[Nik]

Can you post examples?

[Juraj]

How is that possible ? If the only change is when WB is done.

There obviously must be some good way to do it because I can adjust WhiteBalance and even tonemap (with highlight parameter) in CameraRaw/Lightroom at same time. Although I don't know what's the order of precedence.
I am gonna try VFB+ too.

[Ondra]

crap, in that case it is time to remove the checkbox from devel/debug :D We definitely do not want users to use this as artistic control ;)

[Dionysios.TS]

Seriously guys!!! The highlights are crap in legacy 1.5 mode! 1.5 treats highlights better but it has the color problem under 5.000K.

Ondra, you confirm?

[Dionysios.TS]

This are very WIP image, don't judge me on this, take a look at the examples.

The legacy one has nice yellow tones but HC are weak and without contrast.
The 1.5 one has those bluish/cyan tones but HC are great.

[Nik]

I can't confirm that something wrong with highlights in Hotfix. My tests shows that it's nice

[Dionysios.TS]

You mean 1.4 look nice???

Wait a moment, get your image, use the original HC method, not filmic, you turn on the new option (Use legacy 1.5 Color balance) and your highlights are not washed out a bit+ more blueish?

[Nik]

There should be difference in renders. You can't compare it with "broken" 1.5.
Of course if I enable "legacy 1.5" all became blue, BUT if I set WB to 6500 neutral and look to highlights then there is no visible difference between legacy and hotfix.

I think this means that difference of your images is due to white balance a bit lower that 6500. Compare neutral image and there should be no difference between legacy and hotfix.

Maybe I'm wrong. We need a 6500k test

[Dionysios.TS]

I agree with you, while using 6.500K almost no differences at all. The problem appears specially under 5.000K

The images I posted are using 5.200 and you can start already see the differences.

[Ondra]

it seems that in this specific scene the change in behavior just randomly produced a result that you liked better, but there should be no systematic reason why highlights in one version are better than in the other one

[Juraj]

I will post some big comparison between VFB+ and Corona soon.

Corona does a lot bigger problem that kind of ties this together, it clamps. WB cannot use clamped data, it always needs dynamic range, hence why it only works on raw photography, but not jpegs.

[Dionysios.TS]

Simple and stupid teapot example.

Same VFB values used.
IES 3.000K
Whitepoint 3.000K

[Nik]

In my Teapot's apartment (page 2) all is ok too.