More pages: 1 2 3 4 5
6 7 8 9 10 11
Yet another AA technique - SDAA
Sunday, August 28, 2011 | Permalink
I have just posted another demo of yet another anti-aliasing technique.
I call it Second-Depth Anti-Aliasing (SDAA) since it requires the second depths in the scene, i.e. the depths of the backfaces. Quality-wise it rivals that of GPAA/GBAA and is easier to integrate into an existing engine.
[ 10 comments
| Last comment by www.mrgoldengoose.com (2019-09-28 12:06:18)
GBAA demo updated
Monday, August 22, 2011 | Permalink
I have updated the GBAA demo with a few fixes to performance and image quality. The most glaring oversight I did in the previous version was that I made a copy of the backbuffer for the last resolve pass to read. I tend to copy an old demo and start from that rather than setting up everything from scratch. In this case I copied the GPAA demo and it didn't pass my mind that the backbuffer copy that was necessary for GPAA is not necessary when you have a final fullscreen pass like in GBAA. Instead you can just render to a temporary render target and read that while writing to the backbuffer in the end. This buffer copy was nearly half the cost of the resolve process, so eliminating this is significant win. The other thing I did was to only clear the buffers that need it, i.e. the depth buffer and geometry buffer. I could have removed the geometry buffer clear too, but felt it was good for educational purposes to leave it there, plus I would have had to change the skybox pass to accommodate the change, so it wouldn't necessarily be a win in all cases. For image quality I changed the unused component to be 0.5 instead of 0. This is to avoid the case where the distance computation landed exactly on zero, which would be interpreted as "no edge here" rather than edge cutting through pixel center. This resulted in some rare single pixel errors, especially for lower precision formats like RG8S. This demo is still using RG16F though because I saw nearly no performance gain to go down to RG8S.
[ 0 comments
Geometry Buffer Anti-Aliasing
Monday, July 18, 2011 | Permalink
There is a new demo up demonstrating another anti-aliasing technique called GBAA (Geometry Buffer Anti-Aliasing). Like GPAA it uses actual geometry information, but stores it in a fullscreen buffer. This should scale better with dense geometry and additionally allows alpha-tested edges to be anti-aliased.
[ 9 comments
| Last comment by Yours3!f (2012-03-04 22:17:17)
Monday, May 30, 2011 | Permalink
So it's been a little while since my last post. The main reason for that is that I'm in the middle of moving to a new apartment and trying to sell the one I'm living in now, and for most of the time my main computer has been packed into a box. I'll be stuck on my laptop for a while yet, but everything will be back to normal soon.
So what's new?
1) I'll be talking at Siggraph in the Filtering Approaches for Real-Time Anti-Aliasing
course. I'll be covering a technique called Geometry Buffer Anti-Aliasing, a technique I have a working prototype of on my main computer, but haven't had the chance to release yet because my computer is in a box, but once that's resolved I will release a demo of it.
2) I'll be covering about the same topic in GPU Pro 3
3) My wife's belly is getting larger by the day and we're both getting increasingly excited about becoming parents.
4) After signing up to Twitter I was taken by surprise by the amount of people who cared enough to follow my tiny bits of wisdom and mini-rants. The first days my mailbox to spammed to death by "[xyz] is now following you on twitter" email, which was interesting. Now it's more of a steady trickle, which is interesting to watch and find out what kind of people are following me. If you didn't sign up yet, you can find me here
[ 7 comments
| Last comment by drp (2011-08-23 02:22:27)
Announcing the next generation
Friday, April 29, 2011 | Permalink
This is the current version:
Final release is expected at September 21.
[ 20 comments
| Last comment by Jaillum Malord (2011-06-27 00:53:38)
Monday, March 14, 2011 | Permalink
I've added an illustration for the GPAA technique with a little more explanation in the demo description.
The wide line is the geometric edge. The arrows show how the neighbor pixel is chosen. The dashed lines shows the value of line equation for that pixel, which is used to compute the coverage value.
[ 0 comments
Geometric Post-process Anti-Aliasing
Saturday, March 12, 2011 | Permalink
I have a new demo up demonstrating a post-process antialiasing technique. Antialiasing algorithms generally come with four letter acronyms, like MSAA, SSAA, CSAA, MLAA, SRAA etc., so in that tradition I think GPAA (Geometric Post-process Anti-Aliasing) would be a decent name for this one. It works by using the actual geometric information in the scene to figure out how to blend pixels with the neighbors.
Here's a comparison before and after applying GPAA:
[ 21 comments
| Last comment by mkhaled (2011-06-07 16:09:38)
I'm now on Twitter
Wednesday, March 2, 2011 | Permalink
As with all kinds of "social" stuff, I'm late to jump onto the bandwagon. I transformed my personal homepage to a blog long after blogging turned cool.
I joined Facebook long after everyone else did. Seemed like a waste of time to me. Eventually I kind of realized the value of it, it keeps you connected to a lot of people that you otherwise would not keep in touch with. Although it's true it can easily become a time sink. You just need to learn to filter out the noise.
So this Twitter thing came along. For a long while I thought it was pretty pointless. Well, recently I've been reading some people's twitter regularly and kind of realized the value. It's a great way to share thoughts, links and short insights to other people that might be interested. Things you have no time to write a lengthy blog post about, or don't really make sense in blog form. Of course, it's a matter of following the right people. So I finally signed up. And I even tweeted. I have no idea how frequently I will tweet, but I'm sure I will have something of interest once in while.
An unfortunate consequence of being late is that the name "Humus" tends to be taken. So I will be known as "_Humus_" on twitter. Back in 1998 or so when I picked up this nickname it was rather unique. It was never taken if I signed up for something. Guess I didn't compete with as many people back then as with the global mainstream services of today, besides probably being more of an early adopter back then.
[ 2 comments
| Last comment by Dwarden (2011-04-05 15:59:43)
More pages: 1 2 3 4 5
6 7 8 9 10 11