NV Kepler

[MaxG3D]

Nvidia, upon the release of Kepler, is going to show off the new antialiasing method TXAA, which is suppose to be better than 16xMSAA at 1x and even better than 16xMSAA on 2x. Now, is it just me, or the antialiasing on the 8xMSAA is fake?

Anyway, what You think of it? What TXAA might stand for? Are You going to buy new Kepler GPU?

[ENBSeries]

It's one more fake. No one method of pseudo antialiasing is capable to be better than multisample antialiasing, because they work differently and applying to different areas. If you have thin lines of geometry (not the alpha test), then all fake methods like fxaa, smaa and anything else can't antialias it properly. They are good only as faster alternative to multisample antialiasing and where multisampled antialiasing not work (alpha test on textures, like fence f.e.). But of source ideal is supersampling antialiasing, nothing else better than it.

[MaxG3D]

@ENBSeries
Well, since it's probably a temporal antialiasing, it use 2xMSAA to find the edges, so I think that it acctualy might be as good as 16xMSAA. Here's the link to the news, if You are interested: http://www.techpowerup.com/162504/NVIDI ... ailed.html

[MaxG3D]

http://timothylottes.blogspot.com/2012/ ... -info.html