I'm not sure how to name this topic, so here we are... I noticed that renderings in houdini don't match what I see in resolve or any other viewer.

So, I've been using the sidefx labs minimal ocio config globally via environment variable the last couple of years (and still do due to max vray not supporting anything newer). It's ideal for max or blender as well since it cuts down the dropdown options mess dramatically. This also always matched the official aces configs and is identical to what I see in Fusion or Resolve (resolve seems to have its own idea of an aces config and ignores any set in the environment variable - so I consider this to be the "ground truth").

However, now with 20.5 (and probably 20) it doesn't look right in Houdini with the provided "houdini-config-v2.1.0_aces-v1.3_ocio-v2.3.ocio". I've set up houdini.env to override ocio just for houdini with a path to this new config due to reasons stated above and because I need the new config for the file rules.

Below you can find some screenshots with comparisons. I assume I just don't understand what "SDR video" means, but I would expect it do be the same thing as Output sRGB was in the standard configs. At least I can't really see a difference in the config suggesting that it's not meant to be the same.

Can anyone enlighten me as to what I'm doing wrong here?

what colorspace is exr set to under file rules? If it's set to Linear Rec709 it might explain the increase in saturation when loading the exr in an application that assumes that exr is ACEScg.

racoonart
Below you can find some screenshots with comparisons. I assume I just don't understand what "SDR video" means, but I would expect it do be the same thing as Output sRGB was in the standard configs. At least I can't really see a difference in the config suggesting that it's not meant to be the same.

Yes, it has the same purpose as the old Output sRGB--it applies the ODT-RRT component of the ACES color transform.

jsmack
what colorspace is exr set to under file rules? If it's set to Linear Rec709 it might explain the increase in saturation when loading the exr in an application that assumes that exr is ACEScg.

Bullseye! Thank you very much for the right idea there. Exr rule was indeed set to 709 indeed. I had not thought that the file rule also applies to the output as well, since my output color space is set up manually to acescg in the aov render settings.

It's generally rather odd though... I left the exr rule at lin709 because downloadable hdris are usually like that. As soon as I renamed my hdri with a color space tag and changed the rules to exr = acescg, all started to come together - but it now looks like what I was seeing in davinci and everywhere else, not what was displayed in houdini before. Which in my mind means that the saved render was acescg all along, but somehow houdini felt it had to display its renderbuffer differently ... which I don't quite understand why, it should not have anything to do with the file output in the first place. I'm not sure if that makes sense or what I get mixed up in my head now.
Either way, the houdini viewport/mplay now looks like everything else and that's what I wanted. Still, not sure if that's the ideal default for an aces config then.


edit: well... some aovs had manually overridden output color space, not the beauty itself, but it still shouldn't matter I think. If it mattered, the output should now look like houdini before, not houdini now like the output did.