Apparently from when I tried to use it about a month ago, you have to pay for the full premium version of Da Vinci Resolve ($300) to be able to access to the nodes necessary to make “real” AgX color grading even possible since otherwise those color grading nodes are locked behind a paywall.
3 Likes
Just started to read and it wasn’t difficult to find out it was heavily discussed. Couldn’t find out yet why certain decisions were made though.
I think it is quite impressive for a bunch clueless and fragile victims to figure those sorts of problems out too.
You are generalising about everyone only thinking about your goals and your workflows. The statements you make are only true in your context and are definitely false in other contexts.
This is so random… Where does this assumption come from? It’s built on another assumption, that there is something that must be fixed in the first place, but why other software? Use Blender’s compositor if you want. Works fine(mostly, …OK, sometimes works fine… 
). Anyway, what rules are you making up for everyone here? I do loads of compositing for my purposes and I only us Blender’s compositor, everything is fine, no control lost, everything I want works for me… I know other people who are fine with it. Could it be better? Sure. Do I need to use other software? No. I could if I felt like it… I don’t.
It seems like you tried it, you did not like it, maybe it did not match your needs and now you go shouting the sky is falling… while other people just use it and get amazing results with it and are very excited about the benefits and improvements it brings or there are probably people who don’t need it and use something else. It’s just one option, you don’t like it, it doesn’t fit your needs, don’t use it. Don’t make assumptions about the whole community based on yourself. Especially not even knowing what it does and what it’s for, because it’s getting clear you don’t. Go find the information, watch the videos, read the articles, the documentation at least before telling everyone its a tragedy. That’s absurd.
1 Like
If you read Troy_s’s posts, the creator of AgX - posts in https://blenderartists.org/t/feedback-development-filmic-baby-step-to-a-v2/ he not only says that AgX in blender is not real AgX and not reccommended to use but he gives a way to use his “real version” in Da Vinci Resolve, but you have to pay for the full version to access those nodes.
1 Like
His statement is not wrong. By definition, the default that the software ships with is the default.
How long has it been since you were new to Blender, and dealt with the myriad of UX papercuts that Blender brings to the opening night party?
This is a very important note that is being almost entirely forgotten in this thread- Blender AgX is not AgX. It’s a custom flavor of AgX that has no correlation to AgX in other programs. As such, it’s yet another Filmic- a Blender-specific mapping that is not out-of-the-box compatible with other programs
4 Likes
Technically, it was Eary who did most of the work for the AgX transform, so this is not a case of Blender users receiving a ‘dumbed down’ version on purpose to punish freeloaders.
We still got the elimination of the notorious six, the far better handling of realistic light values, and a far better saturation curve, so it definitely works well enough for me in most cases concerning photorealism. Standard and Filmic are not going anywhere, and hopefully we can someday have round-trip live compositing for a render in progress so we can more easily do things like saturate highlights.
An emphasis on the old color transforms not going anywhere, I do not get why some appear to be acting offended by the mere existence of AgX in Blender.
I have no use for X, therefore NO ONE should have X.
3 Likes
I have nothing against AgX, nor do I think @DeepBlender does. The point is that it shouldn’t be a default
2 Likes
Do you think it shouldn’t be the default as well? What should be the default and why?
I don’t think we have a good default option for all users in Blender. 2D artists, NPR artists, and motion graphic artists should use Standard. Legacy artists should use Filmic. Users aiming for photorealism exclusively in Blender should use AgX. Users working with pipelines should edit the OCIO files to match their pipelines. I agree with Sterling, I don’t think there should be one default. It should be a choice on the splash screen
Just to emphasize- I like AgX. I don’t think it should be a default. I like Standard. I don’t think that should be a default either
2 Likes
About the red-orange shift.I have tryed to compensate the hue shift with a simple hue rotation.
Maybe someone want to check this out if this helps.It gives me the red-orange hue of around 10° back to near 0° what primary red has.
(hue 0.493)
Here AgX with the hue rotation vs AgX default
Here a colorbar exr with very high contrast Agx and hue rotation of 0.4972
with different contrast looks the shifts can be different
so far i have none look hue (0.4937) to very high contrast (0.4972) to get red to 0°
Default of hue in the node is 0.5 (no shift)
You can try to lower this value in your own renderings,and check with a colorpicker the correction.Red has a Primary hue rotation angle of 0°.
3 Likes
That makes the most sense I have read here as far.
1 Like
Thanks for pointing that out. I remember now having read it at one point and I forgot about it. It is a great opportunity to check out the original one.
I picked this one, because it is the newest one. If anyone knows whether there are “better” versions for OCIO, let me know.
I assume it has all the advantages or at least most of them which were praised for AgX. Now to my surprise in the low number regime, there is way less “hue shift”. There is some, but it is way less extreme then the AgX in Blender.
So, it might not impossible to have many advantages of AgX while being more intuitive in the sense that the picked colors are more similar in the final image.
I tried to visualize what I am talking about a little bit better (I fear it is more confusing than helpful, but I will take the risk anyways). I picked a bunch of sRGB colors, like (1, 0, 0), (0, 1, 0), (0, 0, 1), (1, 1, 0), … you get the idea, put them into an emissive shader, with varying intensities from 0 to some large number to see each of those colors from very dark to very bright. All the results were an sRGB color and I place all of them in a cube at their sRGB location.
The images show how this cube looks from (0, 0, 0) so the black coordinate (in the image center) towards the white coordinate at (1, 1, 1) which is hidden behind the black one.
Filmic looks like this:
Blender AgX looks like this:
As you can see the blue line (or sequence of points) drifted quite a bit to the left, while the red line drifted to the right.
This is the AgX from the repository mentioned in the previous post:
I haven’t tried it out, but just visually, the direction of the lines is a lot closer to Filmic and having way less of a shift.
2 Likes
Could you share the .blend file for the first two images? I would like to compare the two AgX variants.
The original ‘here be Dragons’ version from Troy_S’ github did not contain any hue shifting, but it also did not include a lot of research from later versions (as it was just a straightforward upgrade from Filmic). It also did not contain any notable looks other than Punchy (the contrast looks were added later by Eary).
Though I would not be surprised if the Resolve version contains some heavier calculations that simply cannot be done in Blender, as it is mainly a compositing app. and therefore somewhat less sensitive to performance (to note, there was concern the devs. would not like the performance reduction compared to filmic regardless of the quality).
Thanks for the information. Do you know if there is a place where the information about the changes is concisely bundled (aka not the Filmic V2 thread 
).
You just need to put the Hue/Saturation/Value node between the image and viewer in the compositing.Check use nodes in the compositing for activation.
I am interested in the color table (don’t know how it is called) which goes from -10 to 15 EV. I tried to recreate it for AgX to get a better understanding, but didn’t know how to do it. When I have constant intervals for the inputs, I am getting something very different. I essentially don’t know how to convert or apply EV to the input.
At Troys github you can download a ton of test images.The -10 to 15 is there as well.Its called Sweep_sRGB_Linear_half_zip.exr
1 Like