Why RE4's lighting may be the GCN's Best

[Squeak]

Talking about environment mapping, both standard and cube, wouldn't PS2 be the best equipped of the three for that, due to its supposedly excellent render to texture capabilities?

Also about (real) bumpmapping, GS obviously doesn’t have any way of doing dot products, that would have to be done on EE. Where the xbox and Gamecube has the calculating power split between the CPU and GPU, PS2 has all of the programmable power on the EE, then why not do the calculation of bumpmaps in one of the vectorunits?
All you would need, would be part of the poly-mesh, a normal map and a microprogram to do the stuff (maybe as an added bonus, the finished bumpmap could be send through the IPU, and then converted to a 4 bit CLUT).
The only thing I can't imagine being done efficiently on PS2, would be EMBM due to the reflectionmap having to be sent to the EE and back.

 

[StefanS]

Talking about environment mapping, both standard and cube, wouldn't PS2 be the best equipped of the three for that, due to its supposedly excellent render to texture capabilities?

Actually Xbox would be the best equipped for Cube Environment since it is implemented in hardware...

 

[Squeak]

It might be down to ignorance, but I can't really see how cube environment mapping would benefit that much from being hardwired, after all it's "just" rendering to texture from 6 (or most likely 5) viewpoints. Good buffer-bandwidth would seem to be a greater advantage.

 

[Li Mu Bai]

Talking about environment mapping, both standard and cube, wouldn't PS2 be the best equipped of the three for that, due to its supposedly excellent render to texture capabilities?

Also about (real) bumpmapping, GS obviously doesn’t have any way of doing dot products, that would have to be done on EE. Where the xbox and Gamecube has the calculating power split between the CPU and GPU, PS2 has all of the programmable power on the EE, then why not do the calculation of bumpmaps in one of the vectorunits?
All you would need, would be part of the poly-mesh, a normal map and a microprogram to do the stuff (maybe as an added bonus, the finished bumpmap could be send through the IPU, and then converted to a 4 bit CLUT).
The only thing I can't imagine being done efficiently on PS2, would be EMBM due to the reflectionmap having to be sent to the EE and back.

The PS2 has unlimited resources now? Phil & Marconelly both said it (DOT3) can be performed in 5 passes. Even at a fillrate which only processes one texture per pass at - 1.2 G/Sec (2.4 with only particle & framebuffer effects) 2 textures at 600m/sec & the numbers depreciate relative to the no.# of textures applied. Well you can do the math, think of the framerate & overall system resource drain it would be all gaming visual aspects considered. Though your theoretical approach differs.......

 

[[maven]]

It might be down to ignorance, but I can't really see how cube environment mapping would benefit that much from being hardwired, after all it's "just" rendering to texture from 6 (or most likely 5) viewpoints. Good buffer-bandwidth would seem to be a greater advantage.

AFAIK it's more the reads from the cube map that need hardware support (and I don't remember the 'Cube supporting those).

 

[cybamerc]

Cube supports Cubic Environment Mapping.

 

[StefanS]

Cube supports Cubic Environment Mapping.

Nope, it's only possible via Gekko; it's not implemented on Flipper.

Fog, Subpixel Anti-aliasing, 8 Hardware Lights, Alpha Blending, Virtual Texture Design, Multi-texturing, Bump Mapping, Environment Mapping, MIP Mapping, Bilinear Filtering, Trilinear Filtering, Ansitropic Filtering, Real-time Hardware Texture Decompression (S3TC)
Real-time Decompression of Display List, HW 3-line Deflickering filter

http://gear.ign.com/articles/306/306374p1.html

Image Processing Function: Fog, Subpixel Anti-aliasing, 8 Hardware Lights, Alpha Blending, Virtual Texture Design, Multi-texturing, Bump Mapping, Environment Mapping, MIP Mapping, Bilinear Filtering, Trilinear Filtering, Ansitropic Filtering, and Real-time Hardware Texture Decompression (S3TC).

Other: Real-time Decompression of Display List, Hardware Motion Compensation Capability, and HW 3-line Deflickering filter.

http://www.segatech.com/gamecube/overview/index.html

 

[Teasy]

Image Processing Function: Fog, Subpixel Anti-aliasing, 8 Hardware Lights, Alpha Blending, Virtual Texture Design, Multi-texturing, Bump Mapping, Environment Mapping, MIP Mapping, Bilinear Filtering, Trilinear Filtering, Ansitropic Filtering, and Real-time Hardware Texture Decompression (S3TC).

Not that I'm saying GC supports Cube Environment Mapping or anything (I don't know). But those specs do list Bump Mapping and then list, seperately, Environment Mapping. Of course Bump Mapping could mean Dot3 and Environment Mapping could mean EMBM. But on the other hand Bump Mapping could cover both Dot 3 and EMBM and Environment Mapping could mean Cube Environment Mapping. Either way the spec list is unclear really.

 

[Qroach]

Cube environment mapping and regular environment mapping are two very different implementations if I recall corectly. For example, Cube environment mapping requires you to render the scene from 6 different angles to have an accurate scene reflection. Regular environment mapping just uses a simple texture map and can't actually give you a real reflection of the surroundings. Regular evironment mapping has been in 3D hardware for years. It's something PS2 is also capable of.

 

[Teasy]

Yeah they are both very different techniques. I just meant that the phrase "Environment Mapping" could mean either Cube Environment Mapping or Environment Mapped Bump Mapping.

 

[StefanS]

Yeah they are both very different techniques. I just meant that the phrase "Environment Mapping" could mean either Cube Environment Mapping or Environment Mapped Bump Mapping.

I remember Korval (working on GCN at Treyarch) saying that there was a way to use CEM via Gekko, but not hardwired on Flipper...

 

[Squeak]

There is also spherical environment mapping, where the map is rendered distorted, as seen through a fisheye lens. Although not a good as the cube variant, it is perfectly acceptable for some purposes and is much cheaper (only requires one render to texture operation).

Talking about environment mapping, both standard and cube, wouldn't PS2 be the best equipped of the three for that, due to its supposedly excellent render to texture capabilities?

Also about (real) bumpmapping, GS obviously doesn’t have any way of doing dot products, that would have to be done on EE. Where the xbox and Gamecube has the calculating power split between the CPU and GPU, PS2 has all of the programmable power on the EE, then why not do the calculation of bumpmaps in one of the vectorunits?
All you would need, would be part of the poly-mesh, a normal map and a microprogram to do the stuff (maybe as an added bonus, the finished bumpmap could be send through the IPU, and then converted to a 4 bit CLUT).
The only thing I can't imagine being done efficiently on PS2, would be EMBM due to the reflectionmap having to be sent to the EE and back.

The PS2 has unlimited resources now? Phil & Marconelly both said it (DOT3) can be performed in 5 passes. Even at a fillrate which only processes one texture per pass at - 1.2 G/Sec (2.4 with only particle & framebuffer effects) 2 textures at 600m/sec & the numbers depreciate relative to the no.# of textures applied. Well you can do the math, think of the framerate & overall system resource drain it would be all gaming visual aspects considered. Though your theoretical approach differs.......

I wasn’t implying that "the PS2 has unlimited resources" Just suggesting that it might be feasible to do all the "passes" on the EE, instead of wasting tons of fillrate.
When calculating bumpmaps you deal with light and normal vectors, what calculating device better suited for that than a vector processor?

 

[Li Mu Bai]

The Cube does indeed support Environment Mapping, & it's not EMBM. "A mapping method that is quite similar to specular mapping is environment mapping. The surroundings of an object are reflecting from its surface. However, this view (consisting of six texture maps, one in each direction) needs to be converted into a spherical environment map. This generated map is used to lookup pixels during runtime. This map needs to be regenerated as soon as the camera orientation is changing." Does it matter if either Flipper or Gekko does it? As long as it is accomplishable.

http://www.gamasutra.com/features/20021002/sauer_pfv.htm

 

[Li Mu Bai]

Marconelly wrote: Splinter Cell had a mixture of dynamic vertex lights and projected light maps, and also HDR effects, but it's shadows implementation was all over the place.

Dynamic vertex lights are accomplishable, as are projected light maps on the GC. (ED used them) But what of HDR effects? ICO & GT3 use versions, or possibly hacks of it because they are done in software & not on a hardware "shader" format. (although still looking very much impressive regardless) HL2 as well as SC used pixel shaders to produce the effect, the TEV can produce the same effects ala RS3. So was Ubi wrong initially when they said only the Box could reproduce such effects, or did system fanboys spin this statement? (as the GC got the PS2 port primarily)

Again, what say you?

 

[Fafalada]

FAIK it's more the reads from the cube map that need hardware support (and I don't remember the 'Cube supporting those).

You don't actually "need" a hardwired implementation, you can simulate it with multiple regular texture lookups in software - however that does make it slower.
True though, GCN doesn't have it in hardware.

 

[StefanS]

Does it matter if either Flipper or Gekko does it? As long as it is accomplishable.

In fact, it does. First you have to send via Bus thus it eats bandwith, secondly it is slower...

 

[Li Mu Bai]

Does it matter if either Flipper or Gekko does it? As long as it is accomplishable.

In fact, it does. First you have to send via Bus thus it eats bandwith, secondly it is slower...

I meant the effect in itself hupfinsgack. Putting bw & speed issues aside, it can be implemented in-game. That is its relevance, it is not that system taxing as to where it cannot be done.

 

[StefanS]

I meant the effect in itself hupfinsgack. Putting bw & speed issues aside, it can be implemented in-game. That is its relevance, it is not that system taxing as to where it cannot be done.

Well, nobody doubted that it could be done, since people confirmed it to be in Wave Race and Korval said there was a way. The question was:

Talking about environment mapping, both standard and cube, wouldn't PS2 be the best equipped of the three for that, due to its supposedly excellent render to texture capabilities?

At least for CEM we know that GCN won't be the fastest. Agree?

 

[Fafalada]

Well, nobody doubted that it could be done, since people confirmed it to be in Wave Race and Korval said there was a way.

Except that what LiMuBai quoted is not CEM, it's spherical, so if that's what was used in WR, it isn't it

 

[Li Mu Bai]

Well, nobody doubted that it could be done, since people confirmed it to be in Wave Race and Korval said there was a way.

Except that what LiMuBai quoted is not CEM, it's spherical, so if that's what was used in WR, it isn't it

Fafalada, it was Sauer's assertion that true reflection/enviroment mapping is not possible. "It’s not feasible with current generation hardware to implement this correctly. This is because a ray tracing approach would be required; which can’t be done in consumer hardware."

Ray Tracing is a global illumination based rendering method. It traces rays of light from the eye back through the image plane into the scene. Then the rays are tested against all objects in the scene to determine if they intersect any objects. Ray tracing handles shadows, multiple specular reflections, and texture mapping. This is next-gen, not current generation capability. It would all have to be spherical.