Hey I've seen some time ago post about tokens and differences in gl/cpu/xpu. It supposed to be solved but on 20.0.535 it still doesnt work for me. I've exported test png srgb texture image from nuke with some colors (maybe still some metadata but it supposed to be solved?). Generally those tokens doesnt change anything on xpu, but in cpu its fine so maybe Im doing something wrong. No matter of file format, does those .rat conversions may mess up? Also what is the proper workflow when it comes to render setting and setting "tone maping" and OCIO rules, default is no tonemaping even if in viewport and OCIO settings I have ACES 1.0-SDR Video? kma_ocio_transform works just fine (when previosly set to vector 3 in image node) on both cpu and xpu just those tokens are messed up. I'm attaching screens and image used as color texture.

I don't know if anything changed in 535, but all three engines match for me in 531 with that image.

I don't understand your other question about workflow and tone mapping. Are you asking about the best way to save display referred images in non-hdr formats when the display contains a tone map? If you are going into compositing software such as nuke, then you would apply it there, but it's also possible using COPs. For batch conversion of exr to jpg, hoiiotool/oiiotool can be used on the command line.

jsmack
I don't know if anything changed in 535, but all three engines match for me in 531 with that image.

I don't understand your other question about workflow and tone mapping. Are you asking about the best way to save display referred images in non-hdr formats when the display contains a tone map? If you are going into compositing software such as nuke, then you would apply it there, but it's also possible using COPs. For batch conversion of exr to jpg, hoiiotool/oiiotool can be used on the command line.

Hmm and when it comes to xpu how are dealing with this .rat For example when name is "name.exr.rat" Is this correctly read in thix .exr extension pattern, or should i use .exr in name patter and .rat extension? And when I have conflict with name token and extension, for exmaple I have .exr in ACEScg but there is linear srgb token in name pattern, whats stronger? I will try on older version as well.
My second question was about saving in exr but in correct color spacec and without any tonemaping baked in image cuz im my quick test I had different results in viewprt and in mplay/render gallery. There was more red like in wrong aces-srgb conversion

EDIT: Those .rat extension tokken was issue. I've been testing and forgot about this but its strange how this respect .rat in tokens. When ther wasnt this token then its treating those images as their original extension. But even if I have in name just .png it was using .rat. However on cpu it wasnt the case so maybe its bug?

sniegockiszymon
Hmm and when it comes to xpu how are dealing with this .rat For example when name is "name.exr.rat" Is this correctly read in thix .exr extension pattern, or should i use .exr in name patter and .rat extension? And when I have conflict with name token and extension, for exmaple I have .exr in ACEScg but there is linear srgb token in name pattern, whats stronger? I will try on older version as well.
My second question was about saving in exr but in correct color spacec and without any tonemaping baked in image cuz im my quick test I had different results in viewprt and in mplay/render gallery. There was more red like in wrong aces-srgb conversion

I've just been ignoring the rat files that are generated, I think they contain metadata that should allow them to be read correctly. I don't have any rat patterns in my config.

sniegockiszymon
And when I have conflict with name token and extension, for exmaple I have .exr in ACEScg but there is linear srgb token in name pattern, whats stronger? I will try on older version as well.

the order in the list should determine the resolution. I don't remember which order is stronger though.

sniegockiszymon
My second question was about saving in exr but in correct color spacec and without any tonemaping baked in image cuz im my quick test I had different results in viewprt and in mplay/render gallery. There was more red like in wrong aces-srgb conversion

I've noticed the 3d viewport is a little different than with 2d viewers, there seems to be an issue with precision and small values.

jsmack
I've noticed the 3d viewport is a little different than with 2d viewers, there seems to be an issue with precision and small values.

Hmm but in 19.5 it wasnt the case. At least not so noticable (cuz I've not noticed xd) I would say its pretty significant essspecialy in redness, and this test wasn't event that saturated so I have to make more examples

sniegockiszymon

jsmack
I've noticed the 3d viewport is a little different than with 2d viewers, there seems to be an issue with precision and small values.

Hmm but in 19.5 it wasnt the case. At least not so noticable (cuz I've not noticed xd) I would say its pretty significant essspecialy in redness, and this test wasn't event that saturated so I have to make more examples

the issue I'm referring to was present in 19.5. It's not so much saturation is increased as the black point is effectively higher that it should be.