I don't know enough about radiosity, so I can't really make an intelligent comment, but here it is anyway:

1 - I really like the look & feel - is that a BSP? Reminds me of a counter-strike map.

2 - Looks way too bright in the upper two images.

i like your article, very good explanations with nice pictures, i like nice pictures!

that's a pretty nice overview on your page, and it sounds like a good way to do diffuse gi :)

personally i find the amount of coding and testing required for radiosity (especially the fancy variants) a little bit daunting, and in that regard will prefer to use monte carlo methods.

really good work though, it's cool to see people exploring research gi methods!


[extremely minor stuff: consider using hdr exposure/tone mapping on your images, and the word "algorithm" instead of "theory" when appropriate.]

Nice work Sebastiano,
I've been toying with digging into Radiosity/GI etc for awhile but haven't made the leap to vc++ d3d vertex/pixel shaders etc, and am wanting some high quality output.
I'd like to quickly mention though, if I read your paper correctly:

The scene above doesn't do a particularly good job showcasing the strengths of face cluster radiosity. The scene above, as mentioned, was from (some low poly count place) and isn't (or doesn't seem to be) particuarly intense in # of triangles. Cluster radiosity seems like (again, if I followed his pictures and description correctly) something that will apply very well to very dense scenes (say a complex indoor/outdoor scene with on average each triangle taking only several pixels onscreen).

Although the pictures are the first thing you see, (and they're great, btw, the only problem being they're kinda washed out) if you're interested in the difference between this and other GI stuff out there, I would encourage you to check out his paper (with pictures! not too scary) and look into face cluster radiosity.

-Michael

are you perhaps michael *granz*?

Your written english is fine. You might want to know that some could find your screenname offensive, though, if they don't understand the context.

Hey Sebastiano (Maciej here)! I see you finally released some screenshots, good work, keep 'em coming. I also wait for new articles.

Hi Maciej, thank you but I think I won't work on it for a while. Before studing new features I want to make some experments on the new pixel and vertex shaders.

Haha, nope, Sorry. I am Michael *Goeke*. A nobody in the computer industry - nonexistant on the web (since my free hosting switched over to "give us money" hosting.) A "vetern" of Flipcode, Gamedev, and OpenGL (since '99), I know my load of...rubbish. (Always interested in this stuff but the most I ever did was a simple prototype of (basically) realtime model based "radiosity" using some processed cubemaps and vertex shaders. Neat, but not plausable in complex scenes without some serious reworking.)

Anyway, I try to keep up on the latest while balancing many other, more important things in life.

-Michael

Oh ok, I looked for the means of nig, but for me it's only an acronym. I've used this nick since I was twelve-thirteen years old :).