Redshift post processing effects as a separate pass and Color problem

[proposal-responsibility]

Hello MAXON,
I have two questions:

1. Is it possible to get post processing effects like bloom and flare from Redshift on a separate pass so that I can later add them on the original image? And a second question.

2. When I add an image as a background image to my camera the image looks a bit off from the original, the highlights burn out, image get a bit darker and also blurrier when I render it out. The image has a -sRGB profile and C4D also uses -sRGB. Any hints to solve this? Another profile for instance?!

Thanks!
Björn

[Dr. Sassi]

Hi Björn.

The Post Processes are done after the image; if you need them, there is no direct option. Rendering them out and one without them using the Difference mode in, e.g., Photoshop might (!?) help, as the mixture of all post options is too large to have tested them all. It is better to render out clean images and apply anything later, especially with background swaps.

In a nutshell, it is to render with an alpha channel and apply the background image later, including all the little things to make it work, a light wrap, and all adjustments needed to make the 3D object sit in the image.

Besides light clips, colors can also clip. Typically, sRGB is the smallest color space in pretty much any production. The amount of problems can accumulate based on that alone. It is not simple to say what happens to your image, and I'm on the opposite end of the idea: "When it looks right it is right".

Images in the background, including many "hdri" wannabe images, might not be linear; some have an S-curve added for contrast, some are tone-mapped, etc. Also, it's not easy to say, besides not matching colorspaces or missing a calibrated monitor. Be very precise with Colorspaces; leaving that on Auto might be the first problem source.

Resolution is tricky, especially with Depth of Field (which pretty much any image has), as typically, the background image does not match the rendering; even if the resolution is the same, the sampling of the 3D model has more data. To have the least loss, at least a 1.5 high resolution is needed.

If possible do any compositing in Post, not with post effects. Too many steps are missing, and things can go wrong in many ways.

All the best

[proposal-responsibility]

Thank you for this long answer. Do you have by accident a tutorial on hand the explain the best practice to deal with color spaces in a Still production. I read a lot of the color workflow with ACES for video production but wehn it come to compositing in Photoshop I can't find anything. I wonder if Photoshop can handle ACES ?! Because looking at the Colospaces I only finde RGBs and CMYK or ProPhoto as the larges Color Space. Any hints?

[Dr. Sassi]

Hi proposal-responsibility,

Photoshop has OCIO now (The framework that holds ACES), but I haven't looked into it in depth. make sure to check which profile is needed and what is applied. ACEScg has no "flag in the file and ACES 2065 should have, but often that is ignored, the changes are significant, if not addressed. ACEScg is an internal format anyway (not suggested for any distribution), so one should know how to set it up.

The typical computer monitor, e.g., iMac, has DCI P3 color space, and we move with UHD and HDR in even larger color spaces and dynamics (REC 2020). Most of the ideas, as mentioned above, "When it looks right, it is right", are 100% obsolete with those newer formats while working on sRGB monitors.

The first step in Photoshop is to set the internal color engine to the Color Space one likes to work in, which should be larger than a small delivery color space. However, that often sets a problem to going from the larger to the smaller space (Gamut) later on, and it depends on what material you typically work on. There is no simple answer.

Often, the material is tone-mapped to fit into a smaller system; this can be done just by using a Curve in Photoshop and flattening the highlights, for example. Since many tone mapping ideas are floating around, this is not reversible if the initial effect applied is unknown.

In a nutshell, if the material has been tone-mapped, the layer compositing is typically much more limited, if not impossible. One thing is a given: Linear and float are the needed formats. Any gamma-based material creates more or less larger problems, and it is not something that I would suggest these days. Which typically brings up ACES. I have taken a few courses over the years and finished the Color Academy for Professional Colorists last year, which left me with too much data. Hence, I suggest that there should always be some books first to fill the gaps if there are some. There is no shortcut, and I have been asked for shortcuts more than I like to share.

Another main problem is often using Alpha and, here, not understanding or even ignoring Straight Alpha.
https://help.maxon.net/c4d/2025/en-us/Default.htm#html/DRENDERSETTINGS-RDATA_GROUP_SAVE.html#RDATA_STRAIGHTALPHA
This is critical to the Compositing, as any image adjustment is otherwise flawed. (Steves Compositing Book goes into it as well.)

There is a lot of nonsense regarding Color on YouTube, and it is often easy to spot when more "hacks" than solutions or clear explanations are given.

The needed knowledge has been made much more accessible in two books:
The Color Management Handbook for Visual Effects Artists 1st Edition
by Victor Perez

Color Science and Digital Imaging: An Essential Guide for Visual Effects Artists, Filmmakers, and Photographers 1st Edition,
by Steve Wright
His color book has videos attached.

For Compositing, the newest release is my current suggestion for building a Compositing background.
Digital Compositing for Film and Video: Production Workflows and Techniques 5th Edition,
by Steve Wright

(I'm biased here with Steve Wright, as I took my first training about Shake Compositing in 2004 and have since held contact with him, as you can see in my review entry under the Amazon text about the book.)

Compared with the typical Color Science books, these books are also well-researched but much easier and faster to read.

Since I can't reproduce all that knowledge in a forum, I encourage going through one book at least, as Color Management and Compositing based on it is crucial and a central theme for the now and the future. There is no way to get away with it anymore, like in the '90s.

I hope this helps a little bit, and I'm sorry that I can't bottle this theme up in a more compact way; even ACES is kind of simple for 2D, but in 3D, it creates way more challenges, and explaining those needs, the base I have suggested.

The question about ACES is whether you have many different sources to mix. If you follow the idea that it looks more filmic, then I might disappoint you here; it is a system that is supposed to not influence the material at all. People often claim to be filmic because the Tonemapped output for super small gamuts and dynamics tends to look that way; if you export for larger, more present-time formats, that effect wears more and more off. This leaves me with the question: why ACES?

I hope that helps a little bit

Enjoy