And that’s not me calling anyone a doof for not knowing. Us graphics coders are not the best at sharing knowledge in a friendly way outside our subfield.
Show this thread
-
Yeah ... T junctions are straight-up incorrect. They will cause actual visible rendering artifacts (which get even worse under e.g. global illumination). Compared to that, a triangle fan is no big deal. Rasterizers used to have problems with thin triangles like 20 years ago,
but my understanding is that this has been solid now for a long time. Even if you are doing your own clipping or something, the kind of “thin triangle” that would cause problems would be much thinner than these.
-
-
Yeah the downside with fans (or thin tris in general) is that you end up with overshading because of tile shading granularity. But maybe this is less of an issue now?
-
This is still a significant efficiency problem. You can't shade smaller than a 2x2 block, and packing blocks from multiple triangles together in a single wavefront has a bunch of extra inefficiency.
-
Round triangles are best triangles.
-
It would be interesting to have a fan primitive where three of the vertices are designated as the ones that cause shading (negative indices?). All the other verts just affect rasterisation. Then you could combine all the fragments together in PS waves.
-
If you don't have vertex-interpolated values that you want to match up ... which most games would. It's just not worth the complexity.
-
I said "interesting" not "practical" :-)
End of conversation
New conversation -
-
-
Microsoft has a whole suite of torture tests like this for triangle slivers, and you're required to pass to get WHQL certification. That's been in place for well over a decade, so people absolutely get it right these days.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.