of course not, since the vid itself is already in a downscaled format.

which can be really slow for longer pp-chains.

let us know when it succeeds

That is funny :-) if I remember the specs right, the N64 had a hardwired antialiasing processing included, because in that era the output was smaller than TV resolution (either PAL or NTSC). They compared the dark-/brightness distribution in the direct neighbourhood around each pixel - and smeared then the pixel along the most prominent staircase-line.

Go for it!

I Would be very happy to see it in action into A8.

I found a demo with sourcecode, some of them might be of interest to you.

Links to other sites of interest:
http://www.vizerie3d.net/
http://igm.univ-mlv.fr/~biri/mlaa-gpu/index.php
http://www.gamedev.net/community/forums/topic.asp?topic_id=580517
http://www.adobe.com/cfusion/exchange/index.cfm?event=extensionDetail&extid=2230024



Thank you (maybe) all, for the positive reaction!

@heelX, and somewhat off topic....I was honestly going to bring up that exact point, as I thought I was one of the only people out there who drooled over the old N64 tech....at the time it was really very advanced, using mip maps, anti aliasing etc....it was interesting to see you brought that up....kinda made my day...thanks.

Okay, in principle I was finished two weeks ago but I'm incredibly busy these weeks. Here is what I got so far:

45°-angles seem to be smoothed quite well.

high contrast edges are smoothed better than low contrast ones; also, non-45° edges are somewhat ragged. also note the weird artifacts.

again, weird artifacts, but it does look smoother.

The good news is, it works! It is extremely cheap on resources (render time some miliseconds on a GF 8600m GT) and independent of scene complexity.
I have to work on the artifacts, though, and I'm still curious on other implementations. It all depends on the edge detection algorithm you use, propagation and composition is quite straightforward.

Joey.

Hi Joey!
Awesome news and the screens looks very impressive for me, any chance to see it in action dynamically?
I'll pm you about edge detection algorithms that I found on the web, but You've reached the limit of your Private Topics.

Hey rojart,

thanks for your comment. I emptied my inbox. What I'm really interested in is this approach:
http://www.iryokufx.com/mlaa/
the results are way better than mine. But I find no technical paper describing their technique.
You'll get a demo of it when it's working properly.

Joey.

edit: the shaders are in GLSL, so you'd have to convert them anyway.

That is maybe because it will appear in the "GPU Pro 2" book which is apparently not on sale yet.

Ah, good to know. Well, if the book contains more than only that I'll buy it; GPU Gems 3 was a good investment, too. Does "GPU Pro" has to do something with the ShaderX-series?

Afaik, it is a different series at a different publishing company, but with Engel as author/coauthor/whatever. It seems his name is like a golden seal to a shader programming book Both books cover obviously the same field (GPU programming), but don't relate to each other. I'm afraid, the "first" GPU Pro Book didn't received well recommendations on Amazon, anyone here owning that book?