Page 2 of 2

Re: Need help testing my model preprocessor.

PostPosted: Thu Jul 18, 2013 7:43 pm
by Epifire
Well this should be it. An old scene I did a bit more work on, all in Softimage '12 so here it is...

http://www.mediafire.com/?28f93h7sm2eyyh2

I tried to add in some hard edges around the base but the ambient occlusion map may be outdated for that. So if your looking for the hard edges you might want to leave that out.

Re: Need help testing my model preprocessor.

PostPosted: Thu Jul 18, 2013 11:28 pm
by stoopdapoop
uh...

Turns out that some modeling programs export the upper left corner of a uv map as (0,0) while some others consider the bottom left to be (0,0). Guess I didn't notice before because the textures I've been using didn't really match the models.

Image

[temporarily fixed in the most hacky way possible]
Image

Thanks a lot epi.

Just need some maya and blender and I'll be done harassing you guys :P

Re: Need help testing my model preprocessor.

PostPosted: Fri Jul 19, 2013 12:11 am
by Epifire
Yeah no problem man. Don't know why my AO bake ended up getting that darker section around the top of the pillar but just so you know I allowed that to stay in the bake. So the bottom version is totally correct just to confirm that.

Re: Need help testing my model preprocessor.

PostPosted: Fri Jul 19, 2013 9:56 pm
by Black_Stormy
I'd say that dark area is because of your smoothing groups. The platform at the top is part of the same contiguous mesh isn't it? You should change it to a different smoothing group or separate it from the pillar mesh entirely before you bake. It's something to do with vertex-shading I think.

Re: Need help testing my model preprocessor.

PostPosted: Sat Jul 20, 2013 12:26 am
by Epifire
I think I forgot to clean out my gradient color data from the tests I was running on it. That was actually the whole point of that pillar to use that and see how it performed. The Pillar is three pieces of floating geometry since it helps me avoid smoothing issues and makes selection easier. Smoothing in XSI is a sort of automatic method, with some name I can't remember.

You can tag hard edges in XSI but as long as verts converge producing an obtuse angle, it will not appear sharp by default. There is the exception of adjacent verts occupying the same space (from accidental duplication) which cause major shading issues. I've eventually overcome that issue with careful observation but it's an easy mistake to overlook.