A tale of two triangles

[psychotron]

oh micropoly displacement.. finally some real bumps - nice work lord keep it up

[jeanphi]

Hi,

Thanks for the number, but I was asking more for comparison to the current subdivision scheme: ie how fast renders a triangle subdivided 50 times with the current subidivision code and displaced and how fast renders the same triangle dynamically subdivided and displaced. This is the really significant metric. A single triangle will obviously be faster to render than millions of triangles

Jeanphi

[Lord Crc]

Ah yes, however it's very hard to give a proper comparison as the loop subdivision shrinks the original mesh.

I don't see it as a replacement for loop subdivision though...

[psychotron]

on the side note.. are there still plans to implement catmull-clark subdivision into lux instead of loop subdivision?

[jeanphi]

I thought there was an option to preserve edges, so that meshes with discontinuities could be subdivided.

Jeanphi

[psychotron]

ot but isn't subdivision by loop algo different then with catmull-clark algorithm?
catmull-clark doesn't preserve edges either (as implemented in blender.. only simple subdiv preserve them)

edit: oh I misunderstud comment..

[Lord Crc]

Ah yes, totally forgot about that feature

Anyway, I'll do some more proper benchmarks when the bugs are out. Currently it has some issues with rays starting within the subdivision volume.

[jeanphi]

When you're done, it'll be able to fully replace the heightfield primitive. That's a nice fallout I think. And while you're at it, don't forget to implement proper NURBS and strands and ...

Jeanphi

[namekuseijin]

ow man! Can't wait! Blender's limit with subsurf is 6 -- of course, my memory usually limits that before. Thankfully, not for this!

[SATtva]

Blender's limit with subsurf is 6

Did you know you can add several subsurf modifiers to an object? ;)