[mrshoe]

My opinion is probably biased because Henrik taught me most of what I know about raytracing, but I think photon mapping is a great algorithm. I'm pleased to announce that I just finished a D implementation of photon mapping and it is now part of Sol's SVN repository. It's probably very buggy, but initial rudimentary tests indicate that the map at least stores photons.

I'm excited to see the increases in photorealism that photon mapping will provide for Sol.

A few initial ideas on changes to the scene file spec are:

* Number of photons to fire (specified per light)
* Number of photons to use in the irradiance estimate
* Whether to use the map for direct lighting or only indirect lighting
* Number of rays in the final gather (if/when I do one... slooooow)

Any other ideas?

[h3r3tic]

Sounds good. You might also include:

* maximum photon search distance
* whether to use a disc-based estimate
* disc height for the disc-based estimate
* percent of photons for which irradiance precalculation should be done ( see http://www.seanet.com/~myandper/jgt99.pdf )

BTW, I've successfully compiled and ran Sol on WinXP So far so good.
Keep up the good work !

/Tom

[mrshoe]

Thanks for those great suggestions.

The maximum search distance I will definitely include in the first version with photon mapping. The disc sample area and irradiance caching are both things that I plan on implementing in future versions (I've never implemented either of those techniques before, although I've always wanted to do so).

I'm glad to hear that Sol compiles and runs on WinXP. Would you be willing to submit your makefiles?

-mrshoe

[h3r3tic]