Alternative AA methods and their comparison with traditional MSAA*

FXAA in my experience does wonders for foliage and shadows with some noticeable blur. However it's weakness besides the blur appears to be long diagonal lines.
 
kind of out of nowhere but anyone know if theres a way to process FXAA on picture files? It can help to speed up 3d rendering time by skipping anti aliasing step and use FXAA to process it.
 
kind of out of nowhere but anyone know if theres a way to process FXAA on picture files? It can help to speed up 3d rendering time by skipping anti aliasing step and use FXAA to process it.

Or it can speed up also some camcorder work… like when you have to deal this old CCD video make… and some time don't have time to do it via usual tools. Like to test.
 
ya... so is there one, I have been using the MLAA one for a little, sort of work but you cant save the image after you process and I had to do a print screen and crop it some time its a few pixel off.
 
FWIW, there was a poster on another post-process AA method at HPG last weekend. The PDF is online at the HPG website.
 
Hey something new to read while at work. MLAA and the like could get very interesting if they are able to adequately handle the subpixel objects as well.
 
ya... so is there one, I have been using the MLAA one for a little, sort of work but you cant save the image after you process and I had to do a print screen and crop it some time its a few pixel off.

are you trying a mlaa'd pic by displaying it then printscreening it, then saving the printscreen ?

I could be wrong here but does mlaa show up in screenshots ?
 
The inject FXAA mod does wonders in Shift 2, which normally suffers from awful aliasing even when in game AA is set to high. It has no perceptible FPS loss (using FRAPS) at 1080p on my machine. Truly remarkable.

Here is the latest version - beta 9, rehosted on a friendlier host. I had trouble grabbing it off the original one.
 
If this console gen hadn't gone on for so long, there may not have been any impetus for focusing on this particular post-process. On PC, you just upgrade.
 
I think that if new generation would start just after GOW 3, we would still see quite many research about post-process AA as now, but transition and competition would be much slower.

What is more interesting for me is a fact that this dll hack was written by a random guy, instead of Timothy Lottes or Jorge Jimenez.
 
What is more interesting for me is a fact that this dll hack was written by a random guy, instead of Timothy Lottes or Jorge Jimenez.
Injecting code (DLL) to executables requires different skill set than creating graphics filters. Lottes and Jimenes are (academic) researchers, not hackers.

I much rather see them focusing all their effort in algorithm development (a field they are really good in), rather than trying to develop some kind of half hearted injection hack for a single platform (PC/DirectX). Post AA support should be include by the game developers. Injection hacks simply do post AA for the final image, and that blurs things not intended by the game developer (UI for example). Integrating the AA filter to game engine also often results in better performance (FXAA v3 for example requires luminance in alpha channel. Injected filters cannot assume this, and must use a slower version of the filter).
 
I still think that it would helped them in research.

And of course those post process AA should be implemented directly in rendering process, but we know they are rare are [lets hope only for now, upgrades UE 3 SKD is step in right direction] and only few if any developer would consider to patch old games with them, even Crytek hasnt patched in new versions of FXAA with DX 11 upgrade and we know that they are already using early version.

But You can be right that they havent got skills for that..
 
Back
Top