| Pemalite said:
|
Nvidia claims otherwise ... The HD 6000 series did have better texture filtering but there's no such thing as "correct" anisotropic filtering since the whole concept is arbitrary. Nvidia may have handled the transitions better but angle independence is the only one to be able to fix cases like that to this ...
(I hope you didn't listen to guys like this ...)
|
Pemalite said: I'm still not talking about mip-maps.
|
Judging from the description, you were ... Changing the texture quality for a surface on the fly implies that it is "mipmapping" in effect.
Being interoperable makes no difference as long as the compression algorithms support a common surface format. If your wondering about their similarities their both based off of S3TC and 3DC+ only builds upon it. There are no practical differences in terms of conserving memory usage (In fact it's worse when your using an INT8 texture which tons of games use.), the biggest change is increase in accuracy which translates to higher quality and Microsoft THEMSELVES say so ...
|
Pemalite said: Unfortunatly it takes years for a standard to become ratified, also doesn't help that the market has shifted where the focus is on power consumption rather than performance. (Aka. LDDR3.) |
The big reason why we're focusing on power consumption rather than performance is for the purpose of exascale and there are better options such as stacked DRAM ...
