Question about the way you store visibility

[fzy28]

Thanks for your good work!

Previously I read your paper and remember you use a SH to store the visibility, which right now instead you use something like a tabular to store each ray direction as well as its visibility results, and this increases the storage and may generate bad results when envmap composited by a powerful point light source.

I wonder why you choose to do this, and if you use SH to store, will the results quality decrease very much?

Thanks!