Tuesday, October 6, 2009

3D diplodocus techniques; uv mapping

A common issue when texturing 3D models is stretching and mismatched resolution of uv maps. As I just encountered this on my diplodocus and some interest has been expressed in 3D techniques, I thought I'd share here. looking at the uvs alone, both uv layouts might look fine, but applying a checkerboard reveals that the first is badly distributed (the surface distribution is not consistent with its 3D counterpart).
In this case, I solved the problem by stretching the layout about and by creating a new problem - a seam. Seams are bad, and I try to avoid them, but this guy's neck is just too long. I could have also tried to lay out the geometry long - then stretched the result into a square, but the legs were too blocky for an elegant layout in that direction.

As I'm always out to use every possible pixel, I want to pack my uvs fairly tightly... you don't necessarily have to be as anal as I am. This is already packed into the square uv coordinates required by my 3D software, I work with a rectangle format until I'm happy, then squish it into a square. I use proportions of the original rectangle for my image however... as it also gets projected onto the 1:1 uv coordinates during render, you have respective resolution.
Let me know if this type of making-of post is welcome or not... I feel that I might be crashing a party.

another w.i.p. view, here the untextured, unsculpted model: