Hi C Jean!
Thank you for posting this question! I´m working on ACES implementation but in the perspective of a photographer. Thus I pretty much try to understand everything in the pipeline from shoot to post. We have come far I would say but the VFX part is still to conquer. I have tried our ACES pipeline in Nuke and with OCIO it works perfectly fine. It´s simple to understand and very straight forward. There are small differences in color between Resolve and Nuke but I at this stage consider it reasonable. Obviously Resolve should have the most correct representation as it is a grading software but I might believe that Nuke is actually doing the math more straight forward so I actually trust what I see in Nuke more. But as I said the differences are to me reasonable small. Now…I have more questions than answers to your post because I think you have dived in to the Fusion implementation of ACES much more than I have…
I currently have BRAW files trying to figure out how to work with these in Resolve Fusion, I figure that the standalone Fusion is more like Nuke and thus more simple in this aspect. For the Resolve Fusion, outside of RCM the clip is developed into a Log file in the Color page and remains a Log in the Fusion page with no linearization applied. With ACES color management though, ACEScct with no IDT or ODT applied, the clip is developed into a linear file and is also linear in Fusion which is correct. Applying a view LUT however makes the file look different than in the Edit page and for example “Linear to sRGB” looks pretty bad…
Changing the ODT to sRGB affects the image in the Edit page as the linear image is now transformed into sRGB colorspace. In Fusion, with no view LUT the image looks Linear which is correct and also implies to me that the linearization is done already in the color settings in Resolve. This makes me think that no OCIO conversion is necessary in Resolve Fusion. The problem for the compositor is not about the OCIO conversion but to find a good view LUT that represents the image seen in the Edit page and actually also in the Color page. This is actually how it´s done in Nuke.
What do you think about this conclusion? I mean that you probably don´t have to use a OCIO conversion to get the image right in the Fusion page because it is already linear from the color management in Resolve, but to find a good view LUT that represents the image in the Edit page.
I might be wrong however since the Resolve manual states that Fusion is not yet connected to RCM and may work as a standalone application within Resolve, which I don´t really like.
I will continue to try to figure this out and I would really hope that someone with good knowledge will dive in to this thread.
All the best!