Talking nasty with Tim Sweeney

[DarN]

GamingGroove has a nice interview with Tim Sweeney:

Tim: I see anisotropic filtering as clearly worthwhile and uncontroversial in all gaming, since it's inexpensive on average, and it improves the worst-case blurring of stretched textures significantly.

But I remain somewhat of a skeptic on the various forms of multisample antialiasing. These techniques are costly, often cause visual anomalies when enabled on an application without its knowledge. And in next-generation games, it's not clear that it is worthwhile to focus so much hardware on solving the edge-aliasing problem when the interior-aliasing problem is at least as significant. Once you move to next-generation per-pixel lighting models, the lighting equations (such as exponential specular lighting) are nonlinear, such that texture mipmap filtering doesn't avoid aliasing as is the case with linear diffuse lighting.

LINKY

 

[Geo]

Yikes! Life without AA would be pretty ugly. Somebody fix it, quick!

Here's an interesting back-down. .

In an interview in 1999 (http://archive.gamespy.com/legacy/interviews/sweeney.shtm), I predicted that CPU's and 3D accelerators were on a collision course that would be apparent by 2007.

But that was before programmable shaders, high-level shading languages, and floating-point pixel processing existed! So, I don't think many people would take that prediction seriously today. But from time to time, developers do need to evaluate the question of whether to implement a given algorithm on either a CPU or GPU. Because as GPU's increase in generality, they are more capable of doing things beyond ordinary texture mapping, while CPU's have unparalleled performance for random-access, branch-intensive operations.

 

[Dave Baumann]

Vince would love that one!

 

[digitalwanderer]

I NEED MY AA!!!!!

 

[Reverend]

This is actually a very good short interview.

 

[Miksu]

I found the bit about programming languages really interesting. Aye, a good interview.

 

[MasterBaiter]

You guys need to upgrade your moniters.

I noticed that Tim mentioned Zbrush, but I'd be curious to know what other software they are using. I would assume Maya... Painter was mentioned many moons ago as well (I wonder if they still use this)...

 

[Tim]

You guys need to upgrade your moniters.

I noticed that Tim mentioned Zbrush, but I'd be curious to know what other software they are using. I would assume Maya....

They might use Maya, but 3ds Max has a bigger markedshare in the gaming industry and has better workflow for polygon modeling and is general more gaming oriented. Maya has an edge in regard to animation, and improved a lot when it comes to gaming oriented features.

The development tools are availeble for both Maya and Max.

 

[Reverend]

I found the bit about programming languages really interesting. Aye, a good interview.

All the questions were good for me, save for the AF/AA one.

Tim always tells me he wants the return of super-sampling

 

[nelg]

Vince would love that one!

PM it to him

 

[PeterAce]

More on these 'new types' of aliasing :

as we get more sophisticated with
what we're doing inside surfaces we've got whole new classes of aliasing that are coming into it which are
"in-surface aliasing" based on the actual texture calculations. So what happens in games that have normal maps
on there is the calculations where you may have a specular highlight that happens at the interpolated point between
one sample and another where one facet may be pointing up and one facet may be pointing off to the right, and
depending on where the viewer and the eye is, some combination either at those points or in-between them, may have
a really bright specular highlight, and Doom doesn't suffer from it too badly because the specular highlights tend
to be very broad, but as you tighten it up it does get to be more of a problem where slight movements cause the
bilinear interpolation (or trilinear interpolation) on the surface to generate normals that either approach or move
away from the exact specular highlight, and that will cause little shimmery speckles to happen on the surface as
things go in and out of the exact highlight point on the reflection vector. So this is something that I'm still
working on various techniques to combat on there. The primary direction that I'm looking at is to go ahead and
analyze the actual surface normals along with the specularity factor and basically broaden the specular highlight
as more geometry is pushed into whatever may be covered by the filter kernel on there, and that seems pretty
promising on there, and it works nicely. One minor drawback is that it does wind up having to tie together the
specularity maps with the normal maps where you wouldn't have the freedom to take a single surface and flip a
different normal map onto it without having a matching specularity map on it, so they become kind of multiple
channels of a more complex data structure on there. That also takes away the ability to scale and rotate them
independently because it again looks like just a deep multichannel texture.

http://www.gamedev.net/community/forums/topic.asp?topic_id=266373

 

[Humus]

I don't get the reasoning. Just because edge aliasing isn't the only kind of aliasing it makes multisampling useless?
I would argue the other way around. Internal aliasing is a good argument for multisampling and against supersampling. Supersampling can't solve internal aliasing in the general case, so it's no solution. The only general solution is that the shader makes any antialiasing as needed for its particular situation. Multisampling can then take care of the edges cheaply and transparently.

 

[Reverend]

I don't get the reasoning. Just because edge aliasing isn't the only kind of aliasing it makes multisampling useless?

Tim said multisampling is useless?

 

[GraphixViolence]

But I remain somewhat of a skeptic on the various forms of multisample antialiasing. These techniques are costly, often cause visual anomalies when enabled on an application without its knowledge.

Seems to me there's a simple and straightforward solution to this: support multisampling natively in the game. Accept the fact that it's a popular and useful feature, and just make sure the game handles it properly rather than constantly relying on IHV drivers to solve any issues.

If by "costly" he means performance-wise, well, faster hardware is always coming out, and there's always the option to turn MSAA off if it's too slow on a particular system. If he meant costly transistor-wise - does he expect IHVs to actually drop this widely used feature from their future processors? Unless display resolutions get a lot higher than 1600x1200 or displays sizes start shrinking, this doesn't seem to make much sense either.

The reason practically all modern hardware accelerates MSAA is because it's a straightforward technique that can improve image quality in almost any 3D application. If someone can come up with a similarly global solution for interior aliasing issues, I'm sure we will see that get accelerated as well. Until then, it's going to be up to developers to come up with solutions for their particular engine.

 

[Reverend]

There is only one reason why MSAA is available on all the current hardware and SSAA is not.

 

[CosmoKramer]

Yikes! Life without AA would be pretty ugly. Somebody fix it, quick!

Easy, just move on from the ugly (hello, alpha texures anyone?) speedhack that is called multisampling and bring back rotated grid supersampling.

 

[Pete]

If Tim thinks MSAA is costly, I don't think SSAA would be any more attractive from that POV.

The GF6 series still offers (partial) SSAA, BTW, and it takes a correspondingly huge performance hit--even with 500MHz DDR.

 

[Killer-Kris]

The bit that caught my attention was the discussion about modeling and all that...

Can anyone enlighten me on why nurbs or other higher order surfaces aren't an option? Even if it's for just the normal map and then tesselated down to a relatively low resolution triangle mesh it would seem like a huge time saver.

 

[CosmoKramer]

Yes of course SS is costly, primarily in fill-rate (including shader fill-rate) but it gets the job done while reducing all types of aliasing mentioned in this thread.

If we also could get back proper anisotripic filtering (hey, wasn't competition supposed to be good for the consumers ) things would be on the right track.

 

[Farid]

I don't get the reasoning. Just because edge aliasing isn't the only kind of aliasing it makes multisampling useless?

Sweeney is not saying MSAA is useless, he's just implying that the "transistor budget" allocated to MSAA in modern V/GPUs should be used for resolving what he calls interior-aliasing problem, since, for him, MSAA is not that important anyway.