I have noticed, that simple switching between Filmic and AgX in Color Management tab breaks color hue.
True indigo hue is 243° because it looks the same in Blender and in Photoshop.
AgX shifts indigo hue to 232° [it of course also shifts brass hue from proper 52° to 46°]
What am I doing wrong here in Blender? Am I missing something?
[contrast, brightness and saturation do not matter here]
Blender 4.1.0
Filmic vs AgX small.blend (4.0 MB)
It’s a color transform. It transforms color. Wouldn’t be any good if it did nothing, would it?
What you are missing is that indigo is not a number, it’s a color. It doesn’t exist in the screen, it exists in our perception as well as in the context of the image. Also there is no “true indigo”, it’s what wee choose to call indigo - just Google “indigo” in image search for a good indication what is accepted as that color. If you do not like that, that’s perfectly fine, use color management that you like. AgX does this though. It’s supposed to. Do not expect something else.
filmic is also a color transform, and as demonstrated does not shift so aggressively in normal exposure ranges. i get the need to shift things to avoid the ‘notorious 6’ in blown out regions, and even that the colour transform problem is ‘hard’. but when you’ve got a colour picker showing a representation on the display that is noticeably different to the representation when run through the colour transform and no exposure settings can align them, there’s an issue imo.
I think I addressed that fully as well:
Nothing else to discuss, that is not discussed million times in another threads.
Your options are Filmic or in Blender 4.2, there is going to be “Khronos PBR Neutral”.
When you use one of those, you are getting predictable hues, contrary to AgX.
Use Standard, it’s the only way to get 1:1 results with Blender. And do yourself a favor and ignore all the forthcoming anti-Standard pedants that will shortly arrive on this post to scream hellfire and damnation in no civil terms from their soapboxes
The point here is simple:
My point is straight:
good hence neutral color transform shall NOT skew hue.
Hue is god.
Don’t say that too loud in this forum 
If you are looking for this, try an experimental version of Blender 4.2. Khronos PBR Neutral does pretty much what you are looking for as far as I can see.
i mean i definitely get MartinZ’s point that different transforms can have different strengths and weaknesses, and i’m generally able to make a call for which one is right for my needs (including standard, especially for npr stuff!)
and i know this has been litigated to death in other threads. i think if anything i bring up the hue shift issues because i do like the concept of what agx is trying to do, i just wish people got less angry and defensive when trying to bring up questions as to whether the highlight preservation qualities of it could still be achieved without the distracting colour issues that seem to crop up in the midtones.
BF needs to stop integrating opinionated color transforms. I am for standardization in the color business across DCCs. If you are going to include a color setting at least make sure it is relatable to other known stuff. My position has nothing to do with whether the transform is right or wrong, or better or worse. The color stuff is already sticky across pipelines and diverging from it just creates more “confusion”, unless BF can come up with a setup that can be the gold standard across DCCs and studios.
My recommendation is to use the “Standard” with EXR outputs pipeline if you need to take your renders/textures/videos etc to somewhere else like Davinci, Premiere, Photoshop, Substance, Katana, Renderman etc.
I was already here.
Now that I feel, as if it’s sort of adequate amount of screaming, I’ll go mourn the photorealisim and aesthetics soon to be lost, occasionally nervously shaking as I remember all the nightmares with radioactively oversaturated and overexposed highlights.
But, whatever works for the purpose I guess…
Taking simple hex/rgb values from the internet, references or our own photos - is very natural workflow. But not when hue gets screwed.
I have been told that using a color picker to select colors and expect the color transform to produce predictable results is not a valid workflow… (I am omitting the insults)
I wouldn’t try that line of argumentation in this forum.
It’s a valid workflow for lots of people. You’re not one of them, and that’s fine, but that doesn’t mean it’s not valid
It certainly is a valid workflow. The issues only begin when people want to use scene referred workflow and still do that. So that should probably be considered.
The industry has no need for X, therefore no one should have X (even if the standard way is outdated and has obvious issues, but it barely changes because it might break legacy workflows rooted in the 1990’s). Now there is the potential issue of matching camera footage, but the limitations of physical photography should not be imposed on digital imagery.
AgX is hardly opinionated because one key point is to have a perfect base for color grading (so things like abney, notorious six burnout, banding, and other things that are quite difficult to remove with compositing can be reintroduced if for some reason it is desired).
And then what?
Will said people (without getting angry) explain it again:
And then you insist the hex-values or whatever are a spoon and then the screaming and foul language starts and then the thread gets closed.
That is going to be a valuable use of everybody’s time.
greetings, Kologe
At least that fire is not over saturated yellow Filmic stuff.
…Ok, I’ll get my coat!
Don’t make me break out my SMPTE color bar test again!
That made my day (regardless of what information is correct).
