Hi everybody.
Iam trying to bake the ambient occlusion and color from a model.
I have followed the steps in the odwiki by Jason Iversen but I a mstill getting a strange error.
I get a strangle, triangle like, pattern whe rendering using the -u option.
The model is perfectly rendered but when I try to bake something I get this strange patterns:


Somebody have sufferend similar problems?
I think that it is a geometry problem because I have used this option succesfully previously, but I can't guess what is the problem, even more, because in a regular mantra render (without -u) I get correct results.

Thanks.

Looks like either you have duplicate polygons or the bias on the occlusion needs to be increased. The bias depends on the size of your geometry so if you have very large geometry, like 100 units across, the bias might be .1 or even .5 even though the default is .001

Cheers,

Peter B

pbowmar
Looks like either you have duplicate polygons or the bias on the occlusion needs to be increased. The bias depends on the size of your geometry so if you have very large geometry, like 100 units across, the bias might be .1 or even .5 even though the default is .001

Cheers,

Peter B

Thanks Peter.
Yes it seems a bias problem, I haev scaled down the model by 0.03 to have a smaller one and be able t use a bias of 0.001, but this solutoin creates another problem, I need bigger bias to solve interpenetration problems but then I lose detail, I have mede several test with different biases and scales and everytime I solved the problem of the bias a lose almost all the detail.
Are there any other solution?

I have tried the -r option, much better but slower than using micropolygons. But there is an important limiatatiom: the -r option is not compatible with the -u option, then is impossible to bake using only raytracing. I am using H 8.1.704 Linux.
I have guess that many of the problems are in the interpenetrating areas where, several planes intersetcs, here the micropolygins pattern is more noticiable.
Another option to improve the results is to increase the shading quality, but to expensive.
This is the best I have get using an occlusion surface shader with bump mappint and a shading quality of 5:

Still having errors and patterns.
I think that Mantra bake option needs a lot of improvement. Hope for H9.
Finally I have to return to LightWave to generate the occlusion pass

The problem is due to overlaped primitives in the UV space.

Ah yeah, that makes sense. That's what it looked like but of course, it's a lot harder to see and detect in UV space.

I miss the Lightwave feature that would detect coincident polygons, though I doubt it worked in UV space (v 5.5 was the last version of LW that I used).

Cheers,

peter B

pbowmar
Ah yeah, that makes sense. That's what it looked like but of course, it's a lot harder to see and detect in UV space.

I miss the Lightwave feature that would detect coincident polygons, though I doubt it worked in UV space (v 5.5 was the last version of LW that I used).

Cheers,

peter B

Jeje finally I have made the occlusion pass in LightWave.
Much better than mantra doing this.