ompf2.com

On page 214, I don't understand the sentence "When working with these error intervals, it’s important to remember that because (1 ± em) terms represent intervals, canceling them incorrect:" Seems fine to me. Although the numerator and denominator contain the same 'term', they don't cancel because t...

Unfortunately you cannot easily use some trivial extension of ray differentials for bidirectional level of detail. As an example, imagine some object A close to a light source. The required tessellation resolution of object A for light rays passing by should not (only) depend on their screen space f...

Also, does anyone know how the updated metropolis sampling, which requires no warm-up, works? http://www.hungrycat.hu/MetropolisSampler.html I just looked at the code. For the most part it is just an implementation of Kelemen style metropolis with MIS but a few things stood out: 1. Instead of estim...

Does a more "difficult" path have a higher probability using MLT, while an "easy, but dark" path has a higher probability in standard path tracing? Is it possible to run both algorithms over the full path space and somehow weight the results in a way has lower error than a hard partitioning? Yes, t...

Hm. I wonder if there's some nutty MIS way to automatically combine "regular" path tracing and MLT... Well, in a way that is what Keleman is doing by mixes large step mutations with small step mutations using MIS. Basically it mixes MLT and PT using MIS. MLT is best at sampling high contribution pa...

...high quality samples ... On another note, whether MLT really produces higher quality samples is debatable. Sure, the sampling density is proportional to the target function, perfect importance sampling and all that, but it comes at a price. When MLT gets stuck in a peak of the target function fo...

That way we get high quality samples AND equal amount of samples per pixel. Has anyone tried this before? You might want to look at the paper "Arbitrary Importance Functions for Metropolis Light Transport" by Hoberock. The idea is to boost the importance of dark pixels and reduce the importance of ...

ad 3) The MIS used is derived for connecting to all vertices of a single "companion" path. When you connect to a subset of vertices from a superset of paths, it is possible that the optimal MIS is different. I haven't looked into it too much, but Dietger's thesis/techreport (cannot find either righ...

I don't think adding materials/shaders will slow it down much. Think again :D Unfortunately the times of 10% shading/ 90% traversal are over (if this was ever true to begin with). Traversal performance scales quite well with increased geometric complexity while shading performance does not scale al...

Did you take a look at https://mediatech.aalto.fi/~timo/HPG2009/index.html ? Looking at their stats the raw ray throughput of diffuse rays is probably more like 2x/2.5x slower than raw primary ray throughput on Kepler. This is the performance of a trace-only kernel, not including any shading, but it...