ACEScg Lightwave3D Fusion workflow

I’m new to ACEScg I have a workflow question

I render in Lightwave 3D which does not have built in ACES support or any type of OCIO native support the only thing I can do is render 16/32 bit linear EXR’s. I then take those files into Fusion 9 and do my comps in the traditional linear workflow.

I have downloaded the latest ACES config files ver 1.2

Inside Fusion I load my rendered EXR’s and then use a OCIO color node that I have pointed to my aces_1.2 config file

I then set the Source Space in the OCIO node to Utility-Linear sRGB and the Output Space to ACES-ACEScg

I then duplicate that OCIO node and within the duplicated OCIO node set the Source Space to ACES-ACEScg and the Output Space to Utility-LInear-sRGB

I then click the LUT button in the viewer and choose OCIO ColorSpace ViewLUT and under edit point the OCIO Config setting to the 1.2 config.ocio file and set Source Space to ACEScg and Output Space to Output-sRGB

When I drop a color corrector node between the two OCIO color space nodes (ACEScg out <----> ACEScg in) I cant really see a difference. I can put the original EXR sRGB render and the ACEScg workflow in side by side viewers each with a color corrector with the same settings and not see a difference.

Is this the correct set up and am I truly working in ACES, or am I limited because I can only render sRGB EXR’s from Lightwave 3D and working in a pseudo ACEScg color space/workflow?

1 Like

I’m replying to my own post.

I think I figured things out initially I had the second OCIO ColorSpace node Output Space set to Utility-Linear sRGB but it should have been set for Output-sRGB the same settings as the OCIO ColorSpace ViewLUT

I also realized I only needed to use the OCIO ColorSpace ViewLUT when viewing the first OCIO ColorSpace node since that’s the one that is outputting ACEScg

After loading both an HDRI image and a 32bit EXR CGI render with the setup and adjusting a color correction node between the OCIO ColorSpace nodes I could see the difference and the improvement.

This is all new to me but I’m 98% sure I’m doing things correctly with my workflow from LightWave now

Hello Kevin,

I am not familiar with Fusion but what you describe looks correct.

  • First OCIO Color Space to convert correctly to ACEScg.
  • Second OCIO Color Space to display with the ODT of your choice.

This setup is correct and even if it is missing some of the benefits of rendering in a wider gamut, you should get some good results. You could also convert your textures to ACEScg but without a proper display transform in Lightwave, it would not really be worth it.

All the best,
Chris

Hello Chris thanks for the reply

You’re correct until Lightwave has native support for ACES there is some limitations but now that I have it set up correctly in Fusion using the OCIO color space nodes and using 32 bit linear sRGB renders from LightWave I can still see enhanced benefits. What I noticed in Fusion is that when making color corrections etc. usually its best to input the values with the keyboard instead of the slider. The input values don’t have to be that large so the sliders can be over sensitive.

One thing I do notice is I sometimes lose detail and contrast in the white values on some renders any advice on how to deal with that?

Hello there,

a few answers :

Awesome ! I’m glad you’re seeing some benefits.

This might come from an incomplete OCIO integration in Fusion. Same thing was happening in Guerilla Render before developers take a real look at its implementation.

You are probably talking about the rolloff of highlights here which is part of the tonemapping. If you could provide me an example, it would be easier to help you. But general rule of thumb would be (if you were working with a simple sRGB display before ACES) :

  • Put more energy in your lights.
  • Legacy assets (made with a sRGB display) would probably have to be tweaked/improved.
  • If we compare with a plain sRGB display we should normally have much better details with an ACES ODT so I’m a bit surprised by your question.

Regards,
Chris

Hey Chris,

Here are 3 side by side screen shots of the render if you look at the white grass on the football field you can see once its in ACEScg some of the sun light color, detail and contrast is lost. One of the pics is the original ACEsg setup with no color corrections applied the other two are my attempt to bring the detail back with color corrections I also tried adjusting settings in the shadow ranges but that didn’t fix anything

Sorry for the late reply. Its is definitely the tone-mapping doing its job on the highlights here. If I may, displaying a linear image/render with an sRGB function is incorrect. My favorite quote from Thomas :

“Friends don’t let friends view scene-linear without an S-shaped viewing transform”.

I totally understand what you mean by loss of details and contrast but you should not compare to the left image because it is basically incorrect. I’ll quote Doug Walker here :

The way that the Linear Workflow is often described contains a crucial error. Usually the description implies that the viewing process simply inverts the gamma adjustment applied on the input side. However, this is wrong and will lead to results that are too low in contrast, have too light a mid-gray, and have clipped highlights, therefore requiring artists to compensate by biasing the lighting and material decisions in unnatural ways. In order to correct this, the viewing transform must account for the fact that the input is what color scientists call “scene-referred” whereas the image being viewed on the monitor is “display-referred”. This means that the viewing transform should not be a simple inverse gamma and instead needs to incorporate what is sometimes called a tone-mapping step.

So I don’t know what stage is your project. But you may have a few options here :

  • Go back to look development, check your albedo/spec values, your light’s exposure (if you have time) with the ACES ODT activated
  • Try to compensate in Nuke (not ideal at all) but it would be faster in the short run.

To sum it up, we are comparing two images : one with no tone-mapping (with clipped values that seem to give contrast) and one with tone-mapping (which looks flat but is in fact technically correct). To give you an idea, to reach a display-referred value of 1 with ACES, you would need a scene-referred value of around 16. You need to put a lot of energy in your scene to reach these values. When a project has been developed under certain display conditions (sRGB for example), it is difficult to just apply an ACES ODT (which is quite heavy on the look) in its last stage. If you send me an exr, I am happy to have a look.

Hope it helps,

Chris

1 Like

It is actually passed from hand to hand and can be tracked back to Jeremy Selan! Way of the Warrior!

Cheers,

Thomas

Awesome ! I didn’t know it was from Jeremy Selan and this makes me like this quote even more. For those who haven’t read it, a link to Cinematic Color never hurts. :wink: Can’t wait to check out v2 of this bible !