Hi Alex,
I will repost what I have shared on the official Octane Discord server, which you have left after it, somehow, escalated.
There are misconceptions about both ACES and Octane. This page could help:
https://www.elsksa.me/scientia/cgi-offline-rendering/octane-render-acesAlexATX wrote:The intermediate color should always be Linear
Linear itself is not a color space. ACES 2065-1 (AP0) or AP1 are both linear "scene referred" "color spaces":
https://www.elsksa.me/scientia/digital-imaging/linear-in-digital-imageryIt is straight forward to match any R'G'B' based renderer. When it comes to a spectral renderer, it is different.
The subtle "color shift" is, as mentioned, likely due to the spectral light transportation model of Octane.
ACES is not matching everything to a pure absolute accuracy. Digital Cinema Cameras can also have subtle (hard to notice) differences. They all have their own spectral response, different sensor/bayer pattern/filtration, proprietary debayering, and so on... However, the specifically engineered IDTs using Camera manufacturers data does help a lot to reduce it.
Even when setting "Linear-sRGB" and "ACEScg" in the OCIO Octane preferences (instead of the default "recommended default settings"), it will not 1:1 match. In none of the cases it does, to a pure absolute 1:1 match. You could found out yourself via a
Diffuse Beauty (Filter) AOV in Octane. In Clarisse, I use the
PBR diffuse reflection k AOV. I do not remember for Arnold but the same AOV is available, without a doubt.
I hope it helped.