[Bf-vfx] Mango Samples with ACES via rrt_sRGB

[Guillermo Espertino (Gez)]

El 16/05/12 19:20, Brecht Van Lommel escribió:
> I can't say that applying a 2.2 gamma makes the images look clearly
> better to me. Of course it brings back detail in the dark areas but
> the colors also get washed out again. The difficulty with this stuff
> is, how to verify that the conversion we are doing is right, i.e.
> neutral?
Let me clarify: I applied gamma 2.2 just as a test. If I'm right the 
reference images were corrected using the factor needed to translate 
S-Log to sRGB instead of linear to sRGB. In that case the resulting 
image wouldn't be uncorrected linear displayed in sRGB, but something 
else, so gamma 2.2 (or actually linear to sRGB) wouldn't be the right 
way to fix it.
Of course I'm talking about the reference jpgs you and Troy uploaded. 
For EXRs it could be more or less the same. But I'm not sure, it's just 
a guess and I thint it's worth a try.

> Also, my understanding is that the F65 has three color temperature
> settings: 3200k, 4300k and 5500k. From the F65 manual I understand
> that these are automatically taken into account from the RAW file when
> doing the conversion and "neutralized". OCIO comes with presets for
> converting from 3200K and 5500K S-Log to ACES. Not sure if anything
> here could be causing issues.
I don't think color tempreture has anything to do with this apparent 
gamma mismatch, but it could be related to the higher red saturation in 
your sGamut samples. I did a couple of tests applying gamma 2.2 to your 
first tests and they actually looked noticeably warmer than the one with 
ACES.

We should ask the DP about his settings. I don't think it's a good idea 
to keep trying to guess what settings did he used (color temperature, 
hopefully a color chart, etc.)

Gez