Multipass PSD's beauty pass does not match Redshift Renderview
-
I'm trying to render a multipass PSD but the beauty pass is very washed out and looks nothing like the RenderView in C4D. I tried rendering with Color Management set to ACES 1.0 SDR-video and Display Space set to sRGB, as well as rendering with Color Management set to Legacy (sRGB linear workflow).
Also, when I render an OpenEXR of the same scene, while it looks correct in After Effects, when I use EXtractoR to view the beauty pass, the colors are much darker than they should be.
How do I render out a correct looking beauty pass?
-
Hi kariomart,
Color management is not something that is simple, click here - click there, and all problems are solved.
Since I write in a forum, here are some tiny notes:
ACES is a system designed for multicamera use on a film set and its complete workflow. If not used properly, trouble is around the corner. ACES relies on perfect exposure; it is, after all, a camera system for wide gamuts and dynamics. In that sense, there is no IDT for Rednerings, and good care needs to be taken for at least a perfect gray-card value 18%. If that is not a given, problems occur.
.ACES is designed to keep the data as is over the longest part, only when it goes into small gauts or low dynamics, tonemapping is applied, which many have touted as ACES Look [sic], this is only to get the data unclipped into that small space or dynamic.
.Third, the value of the pixels inside the file is determined by the color space it is supposed to work in. Suppose a file is coded in ACES 2065. In that case, it will not look right in ACEScg, and surely not in sRGB or REC 709, if not converted properly, or if the settings for the monitor are not set, as the data inside ACES are not changed to be viewed in the pipeline, the signal is taken (split and then brought into Monitor space, again, not touching the main signal. Scene referred.
.If the monitor profile is not set in the app you are viewing it, it might look wrong. If the file itself is not recognized as ACEScg (if it is ACEScg), then it most likely can look wrong.
If the color engine in the app is not set for the colorspace of the file, it might look wrong.
.The safest bet is to keep it all in one color space and float. (I would avoid anything like REC 709 or sRGB during the pipeline, as it is the smallest colorspace (Highest chance to clip color!), as it is an 8-bit/channel integer, which is not designed for anything higher than dynamic range.
I stop here, as I typically tell friends to read first about ACES, preferably a book about color science, to get on top of it.
What I tried to do here is to give a small glimpse about possible steps where things can go wrong, and there are more, and sadly, YouTube has a lot of videos with little knowledge in Color Science, I get the problem.
.So, my suggestion, open the file, do not change a thing, except where to render that image. Name it wisely.
https://projectfiles.maxon.net/Cineversity_Forum_Support/2025_PROJECTS_DRS/20250819_CV4_2025_drs_25_RSac_01.zip
Then open Photoshop, set its color engine to ACES for RGB, and open the file. Do not convert the Color Space, use the Edit> Assign Profile… ACEScg- Open the file that was used in the tex folder, and open it in the same way.
- If the 0.18, 1.0, and 1.7471 fields can be measured in that way, eye dropper, that is a good start.
- Copy and paste one image over the other. If you get an alert that the content will be concerted to…, you made a mistake, start over.
- When you have both files on top of each other, set the top one to Difference.
- It should be mostly black. There is a white spot in the image, above this should be a red dark line. I adjusted it that way to have a better indicator.
- If both images are interpreted wrongly, they will also show black, but the measurements will most likely show different values.
Perhaps it is something else. Then please report your findings back.
If you like book suggestions, please let me know. I promise to suggest something easier than Tooms', "Colour Reproduction in Electronic Imaging Systems: Photography, Television, Cinematography" (Which is one of my favorites.)
All the best
-
Hi Dr. Sassi,
Thanks very much for the help. Here is my result from putting my C4D render, with no changes to your project, over your provided exr, set to Difference:
https://drive.google.com/file/d/1PT-lbS-_xK_lrCYOgcslHZtNtAlcWvcn/view?usp=sharingWith just your provided exr, I am able to eyedrop different values of .18, 1.0, 1.7471
Maybe I should back up, because I'm struggling to understand color spaces in general. I only changed the Color Management because I was following a tutorial, otherwise I never touch anything relating to it. I have an image I want to render with Redshift that I set up several AOVs for. When I render it as a PSD at 32 bit, the beauty pass looks nothing like it should. When I render as an EXR, it looks correct, but in After Effects when I use EXtractoR, the beauty also looks incorrect.
What would be the most direct way of correcting the beauty pass's colors if I'm unconcerned about color space? If it requires a deep dive into understanding color spaces, I can do that and yes I'd appreciate your book suggestion. But if you know a way that didn't involve color spaces, that'd be even better.
Thanks,
Steve -
Hi Steve,
Thanks for using Google and the image. I had to experiment a little to reproduce it.
The color engine in After Effects has no idea what it get sometimes, you need to set up the Color Space for each project, and based on what Ae version you have. When that space is defined, Ae will give you the colors you expect.
.My suspicion is you missed a step.
https://projectfiles.maxon.net/Cineversity_Forum_Support/2025_CLIPS_DRS/20250820_Wrong_Colorspace_assigned.mp4
Pixels are code for a specific color space, including Gamma.
The ACES specification states that only ACES 2065 (AP-0) needs a colorspace tag. The ACEScg, which has a smaller space AP-1, does not have such a suggestion. ACEScg is supposed to be used internally, never for exchange, hence the idea that the artist knows what they are doing.
So, for years my request was not fulfilled by Adobe to make sure that it recognizes ACEScg, which leads, obviously, even today to problems.My best guess is that the image you opened first was left without specifying it as an ACEScg file, and the pasted rendering was converted.
Here is a little video, and I hope it illustrates this problem.
Many people (?) have Photoshop running on sRGB, based on sRGB or perhaps P3 screen, that might not show up easily. Another guess is mine is that calibrated screens are not the standard.
You might find in the video that I left my screen for this in Low Dynamic, to record it accordingly. We live in an era where more and more HDR is requested, and in the past 15 years, I have advocated working in linear as it was clear in 2012 with ACES' released that we would get there. Low Dynamic with sRGB will not hold the water when shown close to HDR, like most of the advertisements have that Low/ High sandwich contrast, surely if not now then soon.
I mention that, as not diving into Color Management is not an option any longer. I'm surely biased, as an Alumni of the "Colour Academy in Hollywood for professional Colorists", and countless other Color Gradings trainings for around 18 years by now. However, I remember how tough it was to get any information on where Gamma is stored when asking in my network in the '90s for help. It took forever, yes that was years before YouTube, and web was dial in. It has become more complex these days, but there is a lot of good material available to remedy that.
I hope that allows to reconsider, and dive in. You will feel how much better things go when this theme turns into muscle memory.
Enjoy
