Christoph Hormann wrote:

> If i see it right your sample is similar to the heightfield solution.  The
> obvious starting point would be to take the image map used for the texture and
> try to get a reasonable corresponding heightfield/isosurface map using filter
> functions.

Which exactly what I did here. The map was desaturated, and then I made a copy of it.
On
this one I painted the planks white, leaving only the gaps black. I gaussian-blurred
it
and then added this map to the previous map.

> heightfields should be a lot faster than isosurfaces in many cases, but of
> course they are not that flexible.

Heightfields are surely faster in the case of  flat objects. I rendered the previous
picture with a height field instead of an isosurface and it rendered 5 times faster
for
an identical output (no need to post it !).
Their main problem is that they are not tileable (unless you make copies of them and
stitch them together) while an isosurface can be of any size.

G.

Post a reply to this message