Saturday, January 5, 2013

Opaque Layer Issues, cont'd

So after a few tests, I think the edge might exist in the alpha image.  That, or for some reason the particles don't exist along that border due to a bug in the culling code.

To try and fix the transition between opaque and transparent particles, I've tried a couple methods.

1. I tried interpolating between the 2 alpha calculation methods (opaque falloff vs. semi-transparent) to see if I could create a smooth transition.  It ended up smoother, but still not right.

No Interpolation
Interpolation attempt

2. I also tried ignoring the opaque tag and just calculating the alpha based on distance, relying on the line-of-sight estimation to create opaque areas.  As you can see in the alpha image, for some reason I'm still getting dark rings around areas that should be opaque.  This becomes less noticeable when using a non-PDF set of particles.
density = 20, PDF
density = 10, PDF
density = 20, no PDF
density = 10, no PDF
This revealed a new issue.  The estimated density is 20, based off of the original density of the voxelization.  Combining that density with a PDF causes the particles to be too transparent.  I'll need to find a better way to calculate the density. 

No comments:

Post a Comment