lacilaci wrote: ↑Fri Apr 03, 2020 1:42 pm
No, im talking pixels as screen space. Currently there is screen size parameter but its not in pixels. And it also still divides whole mesh so not adaptive
And I'm talking of Blender/Cycles/Corona/whatever, they will produce the same result. If you use 2x2 pixels displacement map, you will get some really BIG jag, no matter of what you use. And for any displacement map pixel resolution, you can zoom enough a still see the jags (because an image has a finite resolution).
As I wrote, you need a procedural or vectorial description as displacement map or you will see jags, first or later.
Im probably not making myself clear, sorry for that.
Im talking pixel size as in render resolution. So if rendering is a given resolution then mesh would be tesselated for example 1 pixel max edge length.
Instead of uniform subdivision it would be good if you could tesselate mesh Based on camera position and render resolution adaptively (setting in pixels) and then displace mesh... It seems pointless to have huge subd surface and then in a separate process simplify it. Its incredibly slow and still very taxing on ram overall
The question if, in lacilaci's example, the displacment texture resolution is high enough. It would be good to see an albedo render of this texture on the mesh to judge if pixels are visible (and would cause jagged edges in displacement, regardless the method of subdiv and displacement)
The current LuxCore shapes, which require to first subdivide the mesh very fine (using lots or RAM and CPU time), then displace, then simplify (again, using lots of RAM and CPU time), which is wasteful compared to doing adaptive subdivision and displacement, where the mesh is only subdivided in places that need it (i.e. where the displacement map has detail)
Im aware of the texture limitation, thats how i ended up making a 6k texture cause lower res was showing pixels on displacement. I can render 16k disp texture if i want that is not a problem.
Problem is half of my ram is gone and i have one tire in the scene and it also takes long to get that mesh ready by subdividing it.
Get me pixel size controls and adaptive tesselation so that i dont have 33 mega polys on my tires and leave me playing around with texture size to figure how big it needs to be for a decent close up or high res render.
