id Software's next game... info?

[Razor1]

For the first correction, that's what I meant. For the second, I meant modelors and not programmers. For the fourth, what I'm saying is that Carmack himself woudn't be too interested in advancing the techniques used in various games to make their HDR look good, at least as far as I know....

Don't know about the first but the second, modellers are very used to normals, with out them, even on highly complex objects lighting wouldn't look right. They are needed even for offline rendering. Smooth groups takes care of this in 3dsmax and Maya and modellers can view vertex normals in either program to fix smoothgroups quickly to get the desired results.

 

[nintenho]

Don't know about the first but the second, modellers are very used to normals, with out them, even on highly complex objects lighting wouldn't look right. They are needed even for offline rendering. Smooth groups takes care of this in 3dsmax and Maya and modellers can view vertex normals in either program to fix smoothgroups quickly to get the desired results.

Why do they downgrade from assets that are normally mapped when the majority of games don't have normal mapping? Isn't it usually done the other way around where they make a one million polygon model and downgrade it to a five thousand polygon normally mapped model?

 

[Razor1]

models of this nature are never downgraded, the high poly is almost always built off the low poly.

I'm not sure I understood your question completely though. Most normal map makers can also "bake" textures which will also auto create textures for use without normal maps.

 

[nintenho]

models of this nature are never downgraded, the high poly is almost always built off the low poly.

I'm not sure I understood your question completely though. Most normal map makers can also "bake" textures which will also auto create textures for use without normal maps.

How does a program know how to up the number of polygons? What I was saying is that a super high poly cg model with raytracing and no "faked" detail such as bump mapping is used and then put through a program that makes the model change as little as possible VISUALLY with less polygons. One example is Fight Night 3 on XBOX360 where the original models for the fighters had three million polygons but it was much lower in the actual game although it was normally mapped (If you looked really close at the screenshots released that were at 4x the ingame resolution, you could even see some chin jaggies on one of the fighters)

 

[Razor1]

I think you are confused in the process of how the meshes are made.


The low poly model is made in the process of making the high poly model. Pretty much it is the base for the high poly. Then the normal maps, heightmaps, lightmaps are derieved from the high poly and applied to the low poly poly model and the low poly model is used in real time, The high poly model is not used for anything else other then creation of textures. All bump deatils come from the normal map, or heightmap depending on the shader being used.

 

[nintenho]

I think you are confused in the process of how the meshes are made.


The low poly model is made in the process of making the high poly model. Pretty much it is the base for the high poly. Then the normal maps, heightmaps, lightmaps are derieved from the high poly and applied to the low poly poly model and the low poly model is used in real time, The high poly model is not used for anything else other then creation of textures. All bump deatils come from the normal map, or heightmap depending on the shader being used.

'kay. thanks. was bored.

 

[Sxotty]

i think his post is about the real guys who dont work for ID now. The crew for the early games. The best left or where kicked out. And no JC cant make a good game, he can code for a good game, it takes the other guys to make the good game. Hence lack of any real creative stuff in D3.

Yes cause heaven knows they all went on to show how incredibly creative they were after they left.

 

[nintenho]

Yes cause heaven knows they all went on to show how incredibly creative they were after they left.

I don't think anybody came into this thread wanting to find out too much about any gameplay ideas that ID is working on. /sarcasm

 

[Sxotty]

So what did Romero, and Tom Hall do after they left? I am not saying they are bad, but nothing really spectacular in that direction either, so perhaps it was more the chemistry, or just being in the right place at the right time.

 

[Razor1]

Well I think I would have to agree with you Sxotty each member of the original Id team, did thier job extremely well, but didn't do another persons job equally well, and this is where they failed when they started on thier own.

 

[Fodder]

They're developing Wolfenstien themselves, right?

No. Raven are doing the SP ???, Nerve are doing the MP grin.

And Kevin Cloud is still there and so is Tim Willits (worked on Quake). Tom Hall and John Romero are the ones JR is probably thinking about. Where are they now?

Don't forget guys like American McGee, Sandy Peterson, Paul Steed, Graeme Devine and even Adrian Carmack. After Ion Storm Hall and Romero spent a while on mobile games at MonkeyStone and then had a stint at Midway, but are now working on separate, unannounced MMOs.

 

[rwolf]

I would like to see ID push the limits again on creativity instead of doing another better qualitiy rehash of the same old thing. Nobody can deny the impact they have had on gaming however it time again for another creativity leap.

 

[Reverend]

First of all, thanks to all who stayed on-topic instead of veering off into talking about whether id games have good or bad gameplay. I'm sure it wasn't easy to refrain from doing so.

You can find some hints in his Quakecon 2004 keynote,

I forgot all about that, thanks.

We now know he's using shadow buffers but he told me the following al;most a year ago :

We are using shadow buffers, but stencil is still used for volume selection tasks as an optimization.

Buffers are faster for low angle outdoor lighting that generates huge sheets of stencil overdraw, but for general indoor lighting it is significantly slower, at least at the resolutions and filtering levels I am using.

I am unaware of any (new) shadow buffer algorithms/optimizations done with the current batch of 3D hardware that may help with the problem John specified. I also don't know if he mentioned this in the context of implementing soft shadows (although I'm pretty sure it will be in). So my questions for this particular post are :

1) Does any of the current/latest hardware help address the shadow buffer performance problem John mentioned?
2) Can anyone suggest possible ways that John could be tackling this problem?

 

[nintenho]

Wait, I just remembered that that game Prey uses the Doom 3 engine. Who's the developer?

 

[Reverend]

Wait, I just remembered that that game Prey uses the Doom 3 engine. Who's the developer?

I hope this has something to do with my topic.

 

[Fodder]

Who's the developer?

Human Head, for 3DRealms. Nothing is known about the story/setting of id's next game beyond it being an all-new franchise (Castle Quakendoom!), so you can stop grasping at those poor straws.

 

[nintenho]

I hope this has something to do with my topic.

I thought so...........I'm gonna shut up now.

 

[Farid]

MegaTextureâ„¢ streams data from the HDD, right?
Could it be applied to an optical disc?

Also, why MegaTextureâ„¢ should allow remarkably detailed environments, because from the Quake Wars Screenshot, it seems to be the classical 256x256 and 512x512 color and normal map. It can allow varied environments, I guess, if the artists can produce a lot of different textures, that is.

Also, I hope that JC forgot about volume shadows, for anything that is close to the camera.
The thing is just plain terrible, hard edge, or semi hard edge (The pseudo soft effect implemented in recent games that use volume based shadows) should go for anything that requires details.

Wait, I just remembered that that game Prey uses the Doom 3 engine. Who's the developer?

Human Head.

 

[Richard]

No. Raven are doing the SP ???, Nerve are doing the MP grin.
Don't forget guys like American McGee, Sandy Peterson, Paul Steed, Graeme Devine and even Adrian Carmack.

All of those, except for Graeme and Adrian, weren't game designers (though Sandy's excellent levels almost assuredly qualify as game design by themselves). But Graeme only worked on Quake3+ and Adrian Carmack was still there when D3 shipped. So Tom and John Romero are the only designers from that period that are missing.

After Ion Storm Hall and Romero spent a while on mobile games at MonkeyStone and then had a stint at Midway, but are now working on separate, unannounced MMOs.

Rethorical question. I.e. they never had successes as they did in id Software.

I am unaware of any (new) shadow buffer algorithms/optimizations done with the current batch of 3D hardware that may help with the problem John specified. I also don't know if he mentioned this in the context of implementing soft shadows (although I'm pretty sure it will be in). So my questions for this particular post are :

1) Does any of the current/latest hardware help address the shadow buffer performance problem John mentioned?
2) Can anyone suggest possible ways that John could be tackling this problem?

He isn't very specific about the actual performance problem (as usual). I assume he means cubemap-based shadow buffers which he was bitching about that no card natively supported. This was a while ago so perhaps the GF7/RX1xxx do now.
As for solutions, well, he could always go the Oblivion route and use spotlights instead of pointlights (and give up on dramatic shadows).

MegaTexture™ streams data from the HDD, right?
Could it be applied to an optical disc?

From what is known so far, yes. Whether it could be used in optical media depends on the frame load transfers. The information we have is that at any given time the megatexture only uses 8mb of video memory. Assuming a fairly clever streaming management that shouldn't tax the drive's speed. Well, theoretically. Read speeds for DVDs range from 1.5mb/sec to around 20mb/sec depending on the drive. If the game requires a 4xDVD drive it should work, however, the streaming manager has to take into account the seek speed which is very much worse than a HDD.

Also, why MegaTexture™ should allow remarkably detailed environments, because from the Quake Wars Screenshot, it seems to be the classical 256x256 and 512x512 color and normal map. It can allow varied environments, I guess, if the artists can produce a lot of different textures, that is.

There is only one texture in a megatexture.

The thing is just plain terrible, hard edge, or semi hard edge (The pseudo soft effect implemented in recent games that use volume based shadows) should go for anything that requires details.

Shadowmaps are not a whole lot better. Take a look at BF2 or NFS: Most Wanted and the shadow edge aliasing is plain horrible when "close to the camera"; and worse, it's perspective-variant. JC talked about using 16x randomly rotated samples so that should aleviate this but it won't be hardware accelerated (see our 3dmark '06 interview).

 

[Farid]

There is only one texture in a megatexture.

From the artist perspective or, at least, from the engine persepective, you mean? ?
Because in the facts, it's just a lot of tiled 512x512 (or whatever the decided granularity is) textures streamed when needed.

Shadowmaps are not a whole lot better. Take a look at BF2 or NFS: Most Wanted and the shadow edge aliasing is plain horrible when "close to the camera"; and worse, it's perspective-variant.

You can alleviate the texture aliasing by smoothing the Shadow Map silhouettes.
There's a lot of papers on the subject.

JC talked about using 16x randomly rotated samples so that should aleviate this but it won't be hardware accelerated (see our 3dmark '06 interview).

That's one solution, yes.
Actually, I have seen solutions on PS2/Xbox engines, that therefore used low resolution Shadow Maps, that run quite well and looked good enough. I'm sure that with a higher resolution SMs and better techniques, it can alleviate most of the visible aliasing.

Anyway, I would take "too blurry" soft edge SMs over any Volumetric Shadows.