Parralax occlusion Mapping in development?

i saw this discrete mark in my right click select subject about this
is this an error ? could someone confirm ? :thinking:

I’m don’t think this tells anything. Because you can edit the status yourself.

Well if YOU didn’t set the status to that then maybe one of the admins did. I’ve seen this been mentioned couple times and I do think it is on the course for Eevee eventually. (Just not really the priority immediately)
The default state seems to be blank, though, so unless you accidentally changed it when creating the post it is safe to assume it was mod who changed it.

well i created this topic before right click select was introduce with this “in developement” feature, so it must be an admin :thinking:

Maybe someone saw this and changed the flag. 3 years old and counting…

well if this is the case its heart breaking, because it look like this thread is dead

It is not in active development. However, I am really interested in implementing this. I can’t promise anything for now because we are in a bugfixing phase but that may be possible in the future.

It is not a hugely complex thing to add, but one issue is how the implementation would work. First idea would be to create a new node type for that but other issues arise:

  • How to treat derivatives (for mipmapping) in this case? (not doing it properly can lead to 2x2 pixel artifacts)
  • Could it just be a replacement for micro-displacement?
  • Needs support of proper shadowing.
  • Needs pixel depth offset for correct intersections with other geometry.

That sounds lovely!

What about actual tesselation, GPU tesselation? Despite probably being more expensive, it’s a more straightforward solution to shadowing and intersections. A lot of things things I do in Eevee are slightly more expensive than real time just so I can get that CG look, I’d be willing to take the increased rendertime of tesselation if it’s a nice stopgap solution that matches integrates with Cycles’ current displacement nodes.

I agree it would be nice but it won’t be as easy either. I think this is an even more far stretch goal.


Thanks for the reply !

im happy that some devs are interrested about this feature too. POM could be a huge performance boosted alternative compare to our only polygonal displacement solution that is driving polycounts crazy.

i didn’t saw that dev are working on this right now, i set back the statut of the RCS thread to “Open”, this must be an error

Sorry for bringing this up again. As a stopgap, it would be nice if we get simple displacement of the current geometry, no tesselation, no parallax, just displacement, and the user is left to subdividing the mesh. I would use the Displace modifier but I can’t use the nodes and match it with my material

1 Like

There’s a working build of it already
There’s a node or 2 you can search for and use it

It has self-shadowing working in Eevee even

No silhouette magic yet though.