Cycles Requests

Well again the title says [WIP] meaning work in progress, it means the person hasn’t finished developing the code yet. When patches are finished, often times they remove the [WIP] in title and add some relevant developers in that module as reviewers.

2 Likes

I can’t really remember where I read it, but I seem to remember POM for cycles was more or less rejected because the core devs preferred to optimize microdisplacements / true geometry. So at that point the author of that patch probably quit working on it.

In an ideal world the author of that patch would mark it as ‘abandoned’ if that is the case. But as lots of people do blender coding as a hobby you can’t really expect them to correctly tick all bureaucratic boxes.

It is a bit of a nuisance that because of the multiple communication channels (here, developer.blender.org, blender.chat mailing lists) not always all relevant communication ends up in a visible place.

Not ideal, but it is what it is.

4 Likes

Thankyou. I thought that in this case w-i-p was because of the need for a reviewer only. I understand better now.

1 Like

In that case, Cycles is in need of the devs. getting around and finally committing to bring microdisplacement out of being an experimental feature. We have been getting teased with the promise of microdisplacement as a core feature since the first attempt with the Big Buck Bunny project (only a few years after Blender became Open Source) and I do believe it is about time to officially make it happen.

To note, microdisplacement is usable in Cycles right now thanks to Mai’s work, but I am concerned it will go back to being broken because of ongoing work in things like the subdivision code (while being unable to make bug reports that will be seen as a priority). Recent work already caused it to no longer support creases for instance.

6 Likes

Don’t take my post for gospel though. I think I remember that I read something like that somewhere. But I could completely misremember. I can’t find it anywhere, so it was probably on the chat, and I can’t even remember who said what :smiley:

1 Like

Would that be Parallax Occlusion Mapping - #12 by brecht ?

2 Likes

Blender is always in need of devs, there has been a job opening for a rendering developer for the longest time.

If anyone is interested: Senior Blender Developer — blender.org

9 Likes

Hello, I hope this is the right place for this request/question :slight_smile:

Is there a plan to get AOVs through glass ?
And if it’s planed do you think it could be possible to get Z, vector or UV through glass ?

Actually this is not possible or I missed something. Currently we are using some trick to achieve something similar but could be nice to get everything directly in Cycle.

best regards.

2 Likes

There’s something “in the works™”, but don’t hold your breath:

https://developer.blender.org/D15972

2 Likes

Instant jaw drop !! :heart_eyes: :heart_eyes:

Could be amazing ! just amazing…

1 Like

Is it possible to put in a % for how much denoise occurs. Maybe I’m wrong but it looks like cycles blends the denoise layer with the original art so if this true is it possible to have a % value to dial in the amount of denoise during this blend? Some of our users don’t want 100% denoise, they want to leave a little bit of artifact.

4 Likes

Another very useful feature for Cycles would be some sort of “Spread angle” in AO node.
It would allow to get sharp masks with fewer nodes and less heavy for computation.
image

4 Likes

From my understanding Blender/Cycles doesn’t blend the denoised result with the non-denoise result. However, you can achieve this effect if you wanted by using a simple compositor setup:

(This assumes you’ve enabled denoising in the render settings)

We have a stand alone app that does not use blender, only cycles. But that does give me an idea to maybe set up some nodes to do this.

Suggestion to add a scatter group ID. One issue is when rendering eyeballs inside of a model you don’t want that included in the overall model when doing scatter or you’ll see odd dark lines inside the eye sockets etc. Eye balls are part of the same mesh but should probably have their own scatter group ID to distinguish them from the body scatter group. This could be added to the shader.

1 Like

Not sure if it is related to Cycles, but the ability for pointiness to detect open mesh edges (optionally, alternatively to curvature) can be useful.
Or something that can draw gradient ramp from open edges.

3 Likes

You should be able to do it with Geometry Nodes. First you need to detect non-manifold edges and then use Geometry Proximity node. Then store the result in an attribute which can be used in Shader Editor.

Right now the proximity node will return the absolute distance, not geodesic (=distance by walking on surface). Something to keep in mind because it might break on some particular meshes, or not work as well as they would expect.

How does one detect non-manifold edges though ? We can single out edges with <4 attached vertices but this is a little error-prone because it relies on the object having a grid-like topology. There’s no input for “border edges”, is there ?

Perhaps proximity node could see an upgrade in the future? Tissue addon has 3 distances: euclidean, geodesic and topological, so maybe these algorithms can be used in GN?

I was doing it with:

2 Likes

Ah good one, this is probably more reliable.

Geodesic distances are already there in edit mode (courtesy of Brecht), I assume they can be added to the proximity node with minimal effort. That would be nice. But I’m getting really off topic !

1 Like