OpenImageDenoise

[epilectrolytics]

Initial support in BlendLuxCore is done. Some UI polishing might follow in the future.
OIDN is the new default denoiser, but you can switch back to BCD.

I think the "remove fireflies only" option in BCD that Dade has provided recently is very valuable.
Is it accessible in BlendLuxCore?

[B.Y.O.B.]

I think the "remove fireflies only" option in BCD that Dade has provided recently is very valuable.
Is it accessible in BlendLuxCore?

I think I didn't add it.

[Dade]

I wonder how good OIDN fares vs. BCD when objects are behind glass or seen through a mirror.

Why not render a "deep albedo" like I suggested earlier with this example (right part is the albedo image):

I can do it but would be useful ? The problem we is we don't really know what Oidn needs. For instance, glass is a pure specular material but it has ray split (Reflect Vs Transmit) so it will have quite some noise and may work better with current Albedo while mirror is pure specular with no split and no noise so it may work better with your method. We are going to need some test and input from Intel guys.

[epilectrolytics]

I can do it but would be useful ?

I guess it would be useful for plain surfaces with mirror or (arch)glass material.
Maybe as an option in the material node.
I think it is useless on curved surfaces and rough specular.

As you said, once BlendLuxCore is out we will do some tests with mirrors and panes.

[Dade]

I have tracked an usual cut&paste bug in my Oidn plugin code (B.Y.O.B., you were near to catch it) ... so we have now to redo all the tests, the normal buffer was wrong

[B.Y.O.B.]

Ah I see, the albedo was fed into the normal input
By the way, it seems to me like there is sort of a memory leak: For the first denoising, a bunch of memory is reserved, but it is not freed after denoising. But it is freed before the next denoising, so it doesn't claim more and more.

[Dade]

By the way, it seems to me like there is sort of a memory leak: For the first denoising, a bunch of memory is reserved, but it is not freed after denoising. But it is freed before the next denoising, so it doesn't claim more and more.

It may be a side effect of some internal Oidn behavior: Alpistinho is initializing the Oidn device and filter at the plugin allocation, we could do that for every apply() however it may be slower so I tend to agree with Alpistinho's decision.

[B.Y.O.B.]

We should test how long it takes to initialize.

[Dade]

Second try, hopeful fixed now:

[lacilaci]

Second try, hopeful fixed now:


COMP-3x3-fixed.png

Looking good!