Chaos Corona Forum
Chaos Corona for Cinema 4D => [C4D] Feature Requests => [C4D] Resolved Feature Requests => Topic started by: Eddoron on 2018-04-03, 18:39:17
-
I like to build many things using shaders instead of textures.
Corona and even the native C4D material doesn't support Blur/offset of many procedurals and/or inside the layer shader.
I haven't gone through all (yet), but bitmaps don't seem to be affected.
This can be quite annoying, especially with displacements when you want to keep some details using the screen space divisions.
A blur shader, especially if it had a masking slot, could enable the creation of more complex materials.
A shader or option in a channel could do wonders...if that's not too much to ask for.
-
I just remembered that, a while ago, Octane introduced a filtering option for their displacement node.
Filter type(Box, Gauss) and filter radius.
That'd do the job and it would be helpful to smoothen LDR maps.
-
It seems that a high sample weathering shader(64 iterations here but more would be needed)) setup can smoothen the disps but it's terribly slow.
examples:
1. normal 2 passes, 13sec
2. weathering shader, 2 passes, 42 sec
-
While we're at it, here some more ideas for the displacement. Might require some core work but nonetheless, I think they're worth considering.
-Per-material choice of screen-space or scene-unit-dsp +values.
-Scene unit (cm) displacement is not adaptive like the screen space displacement. Anything out of frustum could be decreased to save memory. The number of units should be changeable in a cascade box, opening when scene-unit-dsp is chosen.
-Auto bump/Auto normal (quite obvious), Parallax Mapping (with silhouettes (https://www.youtube.com/watch?v=gcAsJdo7dME)) would be cool .
-Something unusual: Focus picked area for higher divisions(a fall-off would be good)/ decreasing the subdivisions of areas out-of-focus. User-driven, of course. Together with auto-bump, this could work very well and also allow fancy animations.
-GPU tessellation: Possible with an SDS object in versions that support open-subdiv. The limit is usually 64 steps but often more than enough for certain things(Some applications allow much higher amounts, though I'm not exactly sure how they work).
So, let the GPU handle the subdivisions and be one step closer towards being a hybrid.