Nvidia cards able to run AA in HL2 after all?

[Ruined]

http://www.halflife2.net/forums/showthread.php?s=&threadid=3071

Some snippets:
"Since people seem to be hyperventilating over the anti-aliasing issue, I thought I'd update everyone.

1) How bad is the problem?

With current multi-sample implementations of anti-aliasing, you may sample texels outside of the polygon boundary, which may result in sampling light maps from other polygons.

This has always been a problem. This is a problem with Quake 1, Quake 2, Quake 3, Daikatana, Sin, Elite Force, Half-Life, Counter-Strike on the X-Box, or any game that uses packed lightmaps with multi-sample anti-aliasing.

You would see these artifacts on polygon boundaries where the wrong lightmap is being sampled. It will look like a bright or dark line on the edge of a polygon.

Gary McTaggart brought this up in an email because he is being pretty hardcore about graphics quality right now. This is not a new problem. If you've run a game that uses lightmaps with anti-aliasing turned on, then you've been seeing these artifacts the whole time...

2) What are potential solutions?

*Support Centroid Sampling
*Use Pixel Shaders to Clamp Texture Coordinates

Centroid sampling doesn't have the problem that center sampling does in multi-sample antil-aliasing. ATI has supported this form of anti-aliasing for the 9000 series. The tricky part is enabling this when DirectX doesn't easily expose this.

There's a different trick you can use with hardware, such as NVIDIA's, that doesn't support centroid sampling. Basically you trade off some pixel shader bandwidth to clamp the texture coordinates so that you don't sample texels outside of that polygon's lightmap sub-rect.

Between these two approaches, multi-sample anti-aliasing artifacts should be a non-issue for any DX9-level hardware running Pixel Shader 2.0."

[gokickrocks]

well the ps2.0 performance aint so great on the nv3x series...imagine a level using pixel shaders and having AA on the nv3x...

[Ruined]

Originally posted by gokickrocks
well the ps2.0 performance aint so great on the nv3x series...imagine a level using pixel shaders and having AA on the nv3x...

From tonight's Nvidia chat the ps2 performance apparently will be improved in next major driver rev - also seems that ps2 on nvidia does work well when properly optimized, and it looks like hl2 will be big on shader optimizations. Don't think there is any reason to worry.

[Deathlike2]

You are missing a key point to all of this...

The NVidia alternative at least gives the MSAA AA a chance in HL2, however, understanding the current performance of the shaders in say 3DMark03 and other benchmarks to test it... it's going to be more a performance strain on the GFFX series... it'll be more of a negative for NVidia than anything...

[StealthHawk]

Originally posted by Ruined
From tonight's Nvidia chat the ps2 performance apparently will be improved in next major driver rev - also seems that ps2 on nvidia does work well when properly optimized, and it looks like hl2 will be big on shader optimizations. Don't think there is any reason to worry.

How long have they been saying that PS2.0 performance would be fixed? NV35 claimed double the shader performance of NV30, which is untrue.

How long are people going to believe that shader performance will be *fixed.* It clearly cannot be. Every time performance looks like it's fixed across several applications, we find that NVIDIA has just rewritten all the shader programs to be less stressful. The fact of the matter is that it is a hardware issue and cannot, nor will it ever, be fixed.

That being said, HL2 will run fine, because NVIDIA will recode all the shader programs, hopefully without loss of IQ. But there is no way in hell that they will ever run standard PS2.0 shaders at full/decent speed. I don't know how many times this needs to be proven(or by how many people).

[Hanners]

Regardless of the performance issues, I think it's good news that at least some way of getting AA working on nVidia cards (well, GeForceFX cards at least) has been discovered. We'll just have to wait and see what the performance penalty is.

[dohcmark8]

You guys need to understand.........

nVidia is not the devil, they know that HL2 will be a big game, and they will not let us nV users down, they will do everything in their power to make sure AA works properly.


Funny thing is, if HL2 was a TWIMTBP game, and AA was broke on ATi cards, people would be yelling, calling nVidia the devil etc..... But no, ATi's sponsoring the game so it must be allright for AA to be broke on nVidia cards. :eek: Discuss.....................

[Hanners]

Originally posted by dohcmark8
You guys need to understand.........

nVidia is not the devil, they know that HL2 will be a big game, and they will not let us nV users down, they will do everything in their power to make sure AA works properly.


Funny thing is, if HL2 was a TWIMTBP game, and AA was broke on ATi cards, people would be yelling, calling nVidia the devil etc..... But no, ATi's sponsoring the game so it must be allright for AA to be broke on nVidia cards. :eek: Discuss.....................

The thing is, there's a difference between getting AA working, and getting it working well. The big worry is that if nVidia need to do a lot more shader processing to get AA working, then it's going to cripple the performance of their already weak pixel shaders.

As for your other point, you're absolutely right - People need to stop blaming the issue on the fact that ATi is sponsoring the game, just like people need to stop blaming ATi issues on the fact that a game has a 'Way It's Meant To Be Played' logo in it.

[The Baron]

3dfx is widely considered to have the best looking AA implementation ever (with Voodoo5).

Gary McTaggart, the same guy who brought up the AA problems, worked for 3dfx.

Coincidence? ;)

[reever2]

the only way Nvidia has been "fixing" Ps2.0 speed is putting replacement shader code in their drivers, i would expect nvidias drivers to gain some more mb of shader code once Hl2 and the AA fix is out

[euan]

It's alright for AA to be broke in HL2 when it's ati sponsored, because it is a known hardware problem with all DX9 cards.

It's not OK for a feature to be broken on ATI in a TWIMTBP game becuase if it has been coded, or had the developers assisted by Nvidia Dev Rel to use or implement methods that will be incompatible with the competitors product. IE using patented propriatory OpenGL extentions (instead if ARB, or ATI's own version). Or even checking obscure DirectX caps (supported features), that are not supported on the competitor's hardware, prevent the application from running, then never actually requiring or using those features in the game. Or even taking shortcuts in passing commands through DirectX that isn't to the DX spec, but Nvidias drivers handle, then seeing it fail on a driver that assumes all coding is to spec. (IE dirty code).

[NickSpolec]

Not a very big deal, but this can't be good for the GFFX series.

Shader bandwidth will have to be used to get AA working in Half Life 2 on Nvidia cards, when the game uses heavy shaders as it is.

D'oh.

Looks like a lot of people will have to live without AA (and I am sure most people are prepared for this anyway).

[Ruined]

Originally posted by The Baron
3dfx is widely considered to have the best looking AA implementation ever (with Voodoo5).

Gary McTaggart, the same guy who brought up the AA problems, worked for 3dfx.

Coincidence? ;)

Is it? I heard many who complained about it being problematic with text. I bought exclusively 3dfx from Voodoo1 through Voodoo3 and thought the T-Buffer was the coolest thing since sliced bread, but never saw it in action due to not owning a V4/V5... The real question is, if its the real deal, why hasn't Nvidia implemented it yet?

[AngelGraves13]

someone should have asked them about the T-Buffer.

[Ruined]

Originally posted by 99 to Life
someone should have asked them about the T-Buffer.

Even if they did Nvidia wouldn't talk about future products, which the T-Buffer would fall into.

[dohcmark8]

People, dont worry about ps2.0 performance, have you noticed that it is slowly increasing with every driver release, by the time Det 50s rolla round, I think that shader performance will be a non-issue for nVidia cards.

And, since HL2 will be heavily optimized for ATi hardware, Valve is just asking for nVidia to optimize the crap out of it, then there will be a big controversy, as to why nV optimized, nV will optimize because they are being forced to by Valve, and any reviewer who doesnt make a tidbit about why nV is optimizing should get their ass kicked. That is ofcourse if nV does infact optimize.

[StealthHawk]

Originally posted by dohcmark8
People, dont worry about ps2.0 performance, have you noticed that it is slowly increasing with every driver release, by the time Det 50s rolla round, I think that shader performance will be a non-issue for nVidia cards.

No it's not. Where do you get that information?

[StealthHawk]

Originally posted by dohcmark8
Funny thing is, if HL2 was a TWIMTBP game, and AA was broke on ATi cards, people would be yelling, calling nVidia the devil etc..... But no, ATi's sponsoring the game so it must be allright for AA to be broke on nVidia cards. :eek: Discuss.....................

Right now FSAA doesn't work on any card in HL2. Where's the conspiracy?

[aapo]

Originally posted by dohcmark8
People, dont worry about ps2.0 performance, have you noticed that it is slowly increasing with every driver release, by the time Det 50s rolla round, I think that shader performance will be a non-issue for nVidia cards.

Are you sure that the ps2.0 performance has been increasing? If I'm not mistaken, all ps 2.0 demos out there are dog slow on NV3X, unless fragment programs are spesifically detected and replaced with hand-optimized versions of shaders. The real, pure ps2.0 performance in NV3X is not good at all.

[StealthHawk]

Originally posted by aapo
If I'm not mistaken, all ps 2.0 demos out there are dog slow on NV3X, unless fragment programs are spesifically detected and replaced with hand-optimized versions of shaders.

Correct.

[prodikal]

This may be a very stupid querstion for some of you smart folks but what kind of impact does PS 2 performance make. Or will make in future games. Like is it worth it to invest in a card with really good PS 2 performance if you have a NV35 card? Or trade my nv 35 in for 9800 Pro 256 or the rumored NV 38?

[The Baron]

Well... in theory, yes.

In practice, nobody knows. It's not a bad idea to have fast PS2 support, but then again, by the time it's widespread, we might have NV50 and R600.

[StealthHawk]

Originally posted by prodikal
This may be a very stupid querstion for some of you smart folks but what kind of impact does PS 2 performance make. Or will make in future games. Like is it worth it to invest in a card with really good PS 2 performance if you have a NV35 card? Or trade my nv 35 in for 9800 Pro 256 or the rumored NV 38?

I would definitely wait until the new architectures roll out, ie R420 or NV40. If NV40 still has the PS2 problems that plague NV3x I would look very hard and carefully at R420.

Unlike with DX8, it seems DX9 will catch on much faster.

[Nutty]

Right now FSAA doesn't work on any card in HL2. Where's the conspiracy?

*sigh*

Yes it does!

The only problem, is polygon edges will sample texels slighty outside of texture coordinates. The actual edge aliasing will still look like its anti-aliased.

And like they said, the problem is already in all Q3 engine games, and how many ppl noticed it in that? Probably very very very few, so chances are it isn't that noticable in HL2, even though they claim the artifact is amplified a bit by their extreme lighting.

This thing is blown out of all proportion.

Unless ATI can force centroid sampling on as well, they'll have to sacrifice pixel shader power to clamp the texture coordinates like nvidia too. I have a feeling that centroid sampling wont fix it anyway.

[Hanners]

Originally posted by Nutty
I have a feeling that centroid sampling wont fix it anyway.

Just curious - What makes you say that?