On 2002.04.04 09:44 Sergey V. Udaltsov wrote: > > But the OpenGL spec says that the fog color is calculated on a _pixel_ > > basis and not on a _vertex_ basis. Indeed the result is different, > > especially in long polygons that span from the front way to the back. > Does 4 do pixel-based fog? >
yep. > > Mach64 is able to do the fog properly, i.e., on a pixel basis, but > _not_ > Why? I know - it's only ATI who can answer this question...:) It's because they are quite similar operations so they use the same chip logic. In fact you have a bit to choose wether you want alpha or fog. It's was design option. > > when alpha blending since it uses the path on chip. So the problem is > only > > what to do when both fog and alpha blending are enabled. > Are there many apps using this effects together? > Don't know, but a transparent window in a foggy level is not a situation very hard to happen... > > The solution of using these depending of the contents of a env var is a > > compromise so that gamers achieve a better gameplay sacrifying a little > > > the visual quality and the OpenGL conformance. > Actually, end users in 80% (or 99%?) do not specially care about > conformance. The visual quality really matters. > > > There are other situations as this one. Leif checked on Unreal and > there > > is one (also when alpha blending) that happens and according with his > > experiments reverting to software leads to a severe performance hit. > It was predictable, wasn't it? And any predictions about _visual_ > difference between these two methods? Will users see the difference > easily? Say, if you get 10* speedup with 5% worse quality (I do not > really know how to measure it though:) - almost nobody will really use > SW mode. > In this case the visual difference can be very big... > ... > > Cheers, > > Sergey > José Fonseca _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel