Skip to main content
8 events
when toggle format what by license comment
Jul 20 at 11:11 comment added Enigmatisms One thing to note is that you should avoid self-intersection caused by insufficient floating point accuracy. For example, you have an intersection on a surface, but the calculation is not completely accurate and the ray origin is slightly below the surface, then you can have black dots (or stripes) for the rendering result. Generally, culling is not a good way to avoid abnormal samples.
Jul 20 at 11:09 comment added Enigmatisms Actually, the problem 'discarding intersections with dot product between the incident ray and the surface normal greater than or equal to zero' resembles that of this post. Since you will can barely come across invalid dot product, if the geometries are properly set.
Jul 20 at 11:06 comment added Enigmatisms @KotakaDanski Don't feel bad about not being clear. Edit the post any time you want and I will update mine, too.
Jul 20 at 10:39 comment added Kotaka Danski The problem for me is that without culling, the rendering is not accurate and I get some huge black spots in my image. If I keep the culling, there is almost no refraction.
Jul 20 at 9:43 vote accept Kotaka Danski
Jul 20 at 6:04 comment added Kotaka Danski The surface and vertex normals I calculate in runtime.
Jul 20 at 6:01 comment added Kotaka Danski Thanks for the extensive information. To be honest, I find it kind of complicated, since although I have all these different kinds of materials and rays, I haven't dug down into things such as BSDF. This is something that I should have mentioned and it's entirely my fault for failing to do so. To put it simply, the role of the diffuse non-reflective material is indeed to absorb light. As for the backface culling, I really implied it in the ray space context only, i.e. discarding intersections with dot product between the incident ray and the surface normal greater than or equal to zero.
Jul 20 at 2:39 history answered Enigmatisms CC BY-SA 4.0