Cutting edge global illumination acceleration:
Bruce Walter, Sebastian Fernandez, Adam Arbree, Kavita Bala, Mike Donikian and Donald Greenberg. (2005). Lightcuts: A Scalable Approach to Illumination. Proc. ACM SIGGRAPH, 1098-1107.
Milos Hasan, Fabio Pellacini, Kavita Bala. (2007). Matrix Row-Column Sampling for the Many Lights Problem. Proc. ACM SIGGRAPH 2007.
1 comments:
Lightcuts:
I like this paper. Not just because it has good results, but mostly because it is one of those simple, elegant solutions that are seemingly rare these days. I'm actually quite surprised something like this wasn't thought of long ago, it seems to be very intuitive.
Matrix Row-Column Sampling:
The results in this paper are obviously impressive. In Fig. 4 they point out artifacts but I think these would not be noticeable unless someone was scanning the image looking for errors. Especially considering the time reduction, these subtle artifacts can be lived with.
I'm a bit leery of the low rank assumption from 3.6. Presenting one figure (fig. 3) based on a reduced matrix of one of the models hardly shows that they are correct. All they've shown is that their intuition might be correct and that their algorithm works for a small set of models/environments. Is this true in general?
Post a Comment