08.16.03




h o m e : n e w s : r e v i e w s : d o w n l o a d s : a r t i c l e s : d i s c u s s i o n : l i n k s : a b o u t : a d s








"NV35 Specifications"

Intel Canterwood Chipset reviews

Radeon 9600 Pro reviews

First Mobile 9200 and Geforce Go 5200 benchmarks

Geforce FX 5800 Reviews

ATI announces AIW 9800 Pro

NVIDIA to show a First Look at E3

OpenGL Meeting Notes

Ignore the left column links

Radeon 9800 Pro review from Guru3d




<< archive
news >>




How many GeForce boards do you own? Choose below.

More than 10!

5-10

1-5

None








home : articles : A look at Full Scene Anti-Aliasing : page 2




Joe Glass
April 2, 2000

Why do we need it and what's involved?
FSAA takes a huge amount of pixel fill rate away from the chipset. The reason for this is that in order to do FSAA, the chipset has to render more pixels to a frame than normal. There are several techniques in doing this. One is to do super sampling. This method involves rendering a frame at twice the screen resolution and then downsample the frame (reduce in size) by 50%. This is very simple to do but not the most efficient. To do 800x600, it requires the memory and pixel fill rates equivalent to 1600x1200 resolution. This method has been intriduced by Videologic some time ago in their PowerVR chipset. You can simulate this in a graphics program such as Paint Shop Pro or Adobe Photoshop. Take a screenshot of your favorite game in action. Import it into your graphics program and downsample the image 50%. The end result should give you an idea of how FSAA looks.

There are better and more efficient ways of doing FSAA. The 3dfx method works with their T-buffer technology. It renders every object in a rendered frame with a slight offset. Again, this takes a huge amount of pixel fill rate but does not require the same memory resources as the previous super-sampling method. This is all provided by 3dfx's T-buffer technology.

How does it look?
For super-sampling, try my method mentioned above. For an example of 3dfx's method, try the T-Bluffer demo on Demonews. The T-Bluffer demo uses OpenGL and demonstrates nearly all of 3dfx's T-buffer effects. The difference here is that the T-Bluffer demo is software based and not accelerated in hardware like the VSA-100.

For those that can remember, NVIDIA tried FSAA with the RIVA 128. Unfortunately, the RIVA 128 had dreadfully slow fill rate thus FSAA was dropped. Thankfully, NVIDIA's CEO Jen Hsun Huang hinted that FSAA would make a come back this year and it should debut officially in the NV15. For those that can't wait, if you have a GeForce 256 or RIVA TNT2, you can try NVIDIA's leaked 5.13 Detonator drivers. these drivers give a feel as to what FSAA will do for NVIDIA chipsets.

I have been able to take some screenshots using the 5.13 drivers. I'm not going to provide benchmarks since this has been discussed in length elsewhere on the web, and we all know that there is a huge hit in performance in the 5.13 beta drivers.





<< previous page next page >>





a Filnet solution



Copyright 2000 pcrave. All rights reserved. Privacy Statement. Terms of Use.
Site Design by Filnet, Inc.