it's a bit long, compared to the paper actually. but that's because you have to read several other papers, to even understand the original paper.
-
-
এই থ্রেডটি দেখান
-
and this time I actually put some effort into the summary, and made my own illustrations. It puts a more personal touch to it, which I like.
এই থ্রেডটি দেখান
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
-
-
ধন্যবাদ। আপনার সময়রেখাকে আরো ভালো করে তুলতে টুইটার এটিকে ব্যবহার করবে। পূর্বাবস্থায়পূর্বাবস্থায়
-
-
-
Hmm.. couldn't you also make a CSM the "other way" too? I.e. output closest depth (for fully covered pixels). Then you can find both "definitely in shadow" and "definitely not in shadow" using those two CSMs, and run IZB only on edge pixels.
-
IMO, If you can figure out how to save work that way, it'd be a big deal. Oddly enough, it's surprisingly difficult to early out fully-lit pixels without making assumptions re geometry size relative to your light-space texel size. I spent a lot of time thinking on this. :-/
-
It seems like it would be exactly like the CSM but "the other way around", right? So for pixels that are 100% covered by triangle, set depth to closest (other pixels I guess should be cleared to near plane). That way you'll only pass the shadow map test if you're definitely lit
-
If I'm following, part of the problem is a fully-lit pixel lies *on* the triangle. It's neither beyond a conservative max nor in front of a conservative min. Using anything more than a tiny epsilon offset for precision issues leads to missing contact shadows (or the opposite).
-
oh yeah.. doh. you're right.. the triangle would end up shadowing itself if you made it conservative.
-
Maybe rescue the idea is to use a triangle ID instead... Write closest triangle ID in buffer (uint32). Only when triangle fully covers pixel still. When lighting, if triangle ID is current tri, then must be the closest triangle in pixel and thus fully lit.
-
Actually that wouldn't work either I think.. You couldn't make it conservative so you'd need "full res". Bah.
-
It's one of those times in my career where "obvious" appears to be wrong. To me, the obvious idea is that fully-lit would be easier to trivial reject. My experience is fully-shadowed is much easier.
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
লোড হতে বেশ কিছুক্ষণ সময় নিচ্ছে।
টুইটার তার ক্ষমতার বাইরে চলে গেছে বা কোনো সাময়িক সমস্যার সম্মুখীন হয়েছে আবার চেষ্টা করুন বা আরও তথ্যের জন্য টুইটারের স্থিতি দেখুন।

