NVIDIA's UltraShadow in Doom3

[SteveHill]

So the "chart" is just a little blurb of color? I was expecting a huge table filled with numbers or something, thanks for showing/explaining it to me.

Well, it is a (1D) table filled with numbers , indexed by the texture coordinate's x component.

Well, here it's in hex if you so prefer:

It probably makes most sense as a graph, but that's an exercise for dw.

 

[SteveHill]

Man, it's a damned good thing nVidia moved away from Register Combiners. Wow, they're hard to read.

Tell me about it . I went through with a combination of the spec and the GL Extensions book (nice RC diagrams, but watch out for one or two minor errors in that chapter).

There are also some RC diagrams on the web: http://developer.nvidia.com/attach/6747

Now I just have to figure out what it does.

When I looked at it a few days ago, the specular function seemed to be this (after removing redundant clamps) :

MAD_SAT R1, specular, scaleTwo, subOne;
MUL     R1, R1, R1;
MAD_SAT R1, R1, scaleTwo, subOne;

It doesn't match the LUT exactly though -- whereas the current patch is pretty much spot on -- so it's quite possible that I've misread the spec and/or calls. OTOH, I suppose it's also possible that the r200 path is slightly different and the LUT is based on that.

 

[Matt]

I thought I'd point out that an unnamed NVIDIA employee posted in our comments about UltraShadow II and Doom 3:

http://3dgpu.com/archives/2004/08/20/how-ultrashadow-ii-affects-doom-3/

The benefit from “r_useDepthBoundsTest 1†will be greatest when you turn up multisampling and increase your screen resolution and turn down (or simply leave alone) things that affect shading speed (aniso, texture detail).

This is because “r_useDepthBoundsTest 1†improves the stenciled shadow volume rendering performance portion of Doom3’s rendering workload (but can’t help the shading portion).

You can SEE the benefit of “r_useDepthBoundsTest 1†by also enabling “r_showShadows 1†(that renders the shadow volumes in wireframe to visualize the shadow volumes that are otherwise rendered invisibly). If you toggle betweeen “r_useDepthBoundsTest†on an off, you’ll see less wireframe shadow stuff is rendered. This is indicative of the kind of pixel update savings going on when “r_useDepthBoundsTest†is enabled.

The wireframe shadow volume pixels you no longer see being rendered when “r_useDepthBoundsTest†is on are for pixels that are NOT going to be in shadow for a particular occluder. UltraShadow is effectively saving this pixel update work.

 

[KimB]

Hmm, just tested it, and I can definitely see the difference, but it seems like it's mostly with self shadowing.

 

[anaqer]

WinRAR pwns all those apps. 8)

Nothing beats TC wrt ease of handling compressed files though.

 

[radar1200gs]

Winrar should be used by most computer enthus in the know, as has much better compression than winzip.

Then why aren't they using WinACE? And really besides 'warez' I rarely see items rarred. I would say roughly less than 10% of things I download are not zipped.

More things I download are tgz compressed than rarred...

Thats because compression, like everything else in the computer world follows the "lowest common denominator rules" rule.

Most average computer users don't even know rar and ace exist. They do know about zip though, so thats what they use. People who make downloads available do so in .zip format because it ensures the widest possible audience of potential users can use it.

What would be nice is if there were to be a "zip V2.0" format that still uses the .zip extension but also recognises ace and rar compression formats internally. That way the noobs keep their .zip extension and the rest of us get smaller downloads.

 

[Nappe1]

naah, forget ZIP, RAR and ACE. True users do it with ARC. (maybe accidentally sometimes also with ZOO.)


In fact, rar and ace are getting more popular every day, but mostly this is just for PCs. There's quite bunch of older packers that are still used outside the Windows world. lzw (or lha) is a standard for amigas now and Commodore users use ZIP64. (which is not compatible with original ZIP, though shares lots of ideas algorithmically.) and gzip/tar are still most used in *nix world.

any mac users? what's the most used packing format in apple world nowadays?

 

[phenix]

any mac users? what's the most used packing format in apple world nowadays?

.SIT

 

[KimB]

This is perhaps the best compression utility currently available for Windows:
http://www.7-zip.com/

 

[Frank]

Nah, UltimateZip is by far the best.

 

[KimB]

Skins? They support skins on friggin' compression utility? WTF????

 

[radar1200gs]

The actual compression codec you choose to use is an academic choice frankly. The only important bit is the .zip extension on the filename since that's what the great unwashed masses know/recognise as compression.

It's just like movie files really. So long as you have the appropriate codecs correctly installed you can give any movie file any extension you want (foo.wmv renamed to foo.mpg for instance) and it will play back anyway. The extension is really rather meaningless when all is said and done and only serves to entrench outdated standards.

 

[Reverend]

Looks like Ultrashadow is dead as far as this thread is concerned. Locked.