Octane Baking System ignores Opacity mask

[Rikk The Gaijin]

Alright, my head is about to explode today, after I've spent 5 hours just to try to export the goddamn Alembic with multiple UVs, and that's working now.
My next problem is that Octane Baking System ignores the Opacity mask!

Why do I want to bake the opacity mask, someone could ask. Here's why:

This is my model. It's hair made out of flat stripes, for a real-time character.

And this is how it looks with my Opacity Mask applied:

I want to bake the Ambient Occlusion to use in my game engine, my Opacity mask is on UV channel 1, and I want to bake my AO on UV channel 2.
Unfortunately, since Octane Baking Camera ignores the Opacity mask, the baked result is this:

And obviously that doesn't work at all since it's calculating the AO of the entire polygon, which has a much bigger occlusion area than the same polygon+opacity mask, so the result, when applied in Maya is this horrible mess:

Now, the big question for the Octane Devs is: can this be fixed?
Will it be possible for Octane to bake a texture with the correct Opacity mask?

[bepeg4d]

Hi Rikk,
let me make a couple of tests, I have an idea but it's better to test first
ciao beppe

[bepeg4d]

Hi Rikk,
since I cannot use more then one UV set in C4D it's a bit hard to do an exact test like your
anyway it seems to work correctly if the opacity map is in the same UV set
here is my simple test scene:

UVopacity_orbx_02.orbx.zip

(1.77 MiB) Downloaded 227 times

I think the best is if you could create a simple scene that shows the issue and share it via PM with mojave
ciao beppe

[mojave]

Hey guys,

I am not too sure why this is not working in your case, but there might be several options:

1. If you are rendering using the info channels kernel, make sure "Opacity threshold" is set to 1.0. Otherwise, all samples with opacity equal or higher to the specified value will be consider as opaque. I.e. if the value is 0.0, no opacity is considered at all. The same applies if you render AO as a render layer.
2. If you are using other kernels, make sure "Alpha channel" and "Keep environment" are checked.

I am attaching a project with a setup that shows how it should be setup so it works.

If none of these work, would you mind sharing your project for further testing?

Thank you.

[Rikk The Gaijin]

Thanks for the replies.
Enabling "Alpha channel" does indeed enable the Opacity mask, but it seems the AO calculation when baking still doesn't consider it.

Here's a gif to show the problem:



As you can see, there is no difference in the occlusion between the two. I am expecting to see a much lighter occlusion from the thin hair, but it still occluding the whole polygon.

[Rikk The Gaijin]

Ok, to help you debug it I made a test scene:

This simple scene has two planes, a larger one with a smaller one on top, both have two sets of UVs.

The plane on top has an Opacity mask with a shape on it

But when I bake it, you can see that the plane underneath doesn't receive the AO from the shape, but the AO from the whole plane above.

I've attached the scene so you can debug it.

[Rikk The Gaijin]

More testing:
Using the AO render pass seems to work! Well, kind of. As you can see from the image below, now the correct occlusion gets casted on the plane below, but why do I see an occlusion border around the shape? The polygon on top should be completely white.

[abstrax]

Ok, to help you debug it I made a test scene:

This simple scene has two planes, a larger one with a smaller one on top, both have two sets of UVs.

00.jpg

The plane on top has an Opacity mask with a shape on it

01b.jpg

But when I bake it, you can see that the plane underneath doesn't receive the AO from the shape, but the AO from the whole plane above.

02b.jpg

I've attached the scene so you can debug it.

What works and is much faster to render anyway is using the info channel kernel with channel "AO" and enabling the option "AO alpha shadows" instead of the direct lighting kernel. We will look into why it's not working with the baking camera.

[Rikk The Gaijin]

Yep, I wrote it just above your reply.

[abstrax]

Yep, I wrote it just above your reply.

Yes, indeed. Arturo probably has already a fix for the issue of the weird "self-shadowing" you can see in the info channel case. The original problem with direct lighting (or any other kernel, except the info channel kernel) seems to be a bit more tricky.

Thanks for the feedback anyway.