Star Fox programmer: PSP even more powerful than PS2

I

Izzy

Banned
Taken from the Tokyopia forums.



Q: So you are saying even that color bleeding is realtime? How come the sperical harmonics are *that* much faster method than anything else (or is it something that PSP has that speeds things up?), and why aren't they used in games already, then?


Dylan Cuthbert: It can be used in current games but the PSP is simply faster at doing this kind of stuff than the PS2. The colour bleeding is all tied in with the spherical harmonic calculations which is a fairly recent (and public) technique for combining all the lights in a sphere into one set of coefficients at each point. The colour bleeding changes in realtime as the light source moves/changes.
Click to expand...
 
At this point its not going to be too surprising to see the PSP having better graphics than the PS2 considering it's doing in its first gen what its taken the PS2 years to get up to (stuff like GT4).
 
Izzy said:
It's still authentic. Concentrate on the subject matter.
Click to expand...
Damn me and my ability to be able to comment on something related to the topic at hand.

Misleading topic thread anyways. It's already known that the PSP is more powerful in certain aspects than the PS2 thanks to improved software resources and documentation along with some great hardware enhacements. Nothing to see here.
 
Perhaps, but I've yet to see it posted in this forum.
Click to expand...
I have posted it couple of weeks after E3. Actually, I think I was the one to ask him that quoted question on Tokyopia.

He was the guy who programmed the ducky demo on PS2, and also the ducky and 'Harmonic City' demos for PSP (which were probably the two best looking things on PSP at E3, alongside that bumpmapped planet surface demo which was not his work)

psp32.jpg


Dylan certainly knows what he's talking about. The project he is developing at Q Games is phenomenal.
Click to expand...
Wait a sec, he's not the one who's making 'Mercury' on the PSP... So what is he working on? I'd like a PM if you can say anything.
 
And now for something unrelated:

QUOTE (dylan @ Jun 16 2004, 12:15 AM)
Giles and I flew "upper class" a few times to and fro from Japan when we were making Starfox (back when Nintendo was a little richer) and it was amazing. They had a lounge area with leather seats that you could go and "hang out" in (on the plane, not in the airport!). On the table were Godiva-like chocolates and a basket of fresh fruit. Definitely the way to fly...
Click to expand...


and something related:

Yes, I think developers are going to find the PSP fun to develop for, that's something that is missing from PS2 development unfortunately. (or at least, it was missing back when I was doing PS2 stuff 3 years ago
Click to expand...

Yes, light and soft shadows were being calculated on the fly using spherical harmonics, hence the name "Harmonic City". The PSP is easily capable of that kind of stuff. Given a bit more time with the hardware I could probably optimize the code to run at 3-4 times the speed it was running for E3.

The duck was pretty much the same as the original PS2 duck demo, except *it too* was being lit with spherical harmonics, along with the refraction etc. With a bit more time too I'll get reflections working.

There's no reason why either of these techniques can't be used in a "real" game. The mars demo or the Luga demo weren't ours but were made by a couple of other scei engineers. I think it was specular bump-mapping.
Click to expand...
 
Marconelly said:
So what is he working on?
Click to expand...

I'm gonna assume they're talking about the game that relates to:

egg8.gif
and
egg3.gif
, which I'm super jealous that they've laid eyes on.

Dylan, Kenkichi & q-games in general rock it; really looking forward to seeing what's to come.
 
Looks like Spherical Harmonics Lighting is the new toy for programmers these days. I know it's heavily used in the Conker Remake too. Quite a nifty way to light a scene though.
 
More:

They had final hardware for a few weeks in Japan - that's how we got our demos done. Actually our demos are the only examples running at the show that used at least part of the full potential of the final hardware.

All the other demos were either streamed from the emulator into movie sequences or running on the real hardware but not using the full potential. There are several features of the hardware that haven't fully detailed to developers yet, two of which are very important:- how to use vram, and the specs of the vector co-processor (which is *very* powerful).

It's a shame you didn't get to see the city, it looked pretty damn good on the PSP screen. Hopefully you didn't watch it on those tv displays they had - the tvs only had a 30hz update and looked really bad because they were "de-interlacing" the psp output even though it isnt interlaced in the first place. :-(

I think the TGS is going to have some pretty nifty PSP software on display.
Click to expand...



Did you check out the two psp demos we made, the duck of course, and the city? The psp hardware is a nice bit of kit, and I hate to say it because I like Nintendo but 3d-wise it probably blows the DS out of the water (the DS doesn't have any texture filtering by all accounts), also using a touch-screen seems fun at first but it is kind of exclusive to using the gamepad so might get tiring/and/or/tedious after a while.

By the time the Tokyo game show comes around and developers have had actual access to the final hardware I think we'll see some pretty amazing stuff on display... and of course Nintendo doesn't deem TGS worthy enough to show up so Sony'll have a free uncontested showing
Click to expand...
 
I would just like to point out (no matter how futile it may be)...
Spherical Harmonics Irradiance(which is more correct name of the approach) is NOT the technique used in that PSP demo.
SHI is a nice method to represent many distant lights cheaply, and involves little or no precomputation, and afaik it's already been used in a couple of shipped games. It doesn't look significantly different from just using lots of regular "ingame" lights, but it's a lot cheaper - it's also Not a way to generate those fancy global illumination effects.

The technique from PSP demo is Precomputed Radiance Transfers - it's a relatively recent approach that simulates effects of global illumination by precomputing radiance transfers and encoding them using basis functions - most commonly, the basis functions for the task chosen are Spherical Harmonics, but they aren't the only ones you can use.

Based on what I've seen of Conker so far, and knowing the limitations of PRTs (which are many), I am relatively convinced they are using SHI in their game.


Anyway, sorry for the rant, but I just get irked over it being called "Spherical Harmonics Lighting"...
 
Fafalada said:
...most commonly, the basis functions for the task chosen are Spherical Harmonics...
Click to expand...
So are you seeing that members of the general dev community refer to PRT as SH because of this or simply because "Spherical Harmonics" sounds sexier?
 
SolidSnakex said:
At this point its not going to be too surprising to see the PSP having better graphics than the PS2 considering it's doing in its first gen what its taken the PS2 years to get up to (stuff like GT4).
Click to expand...

Assuming GT4 was PSP in game.
 
Fafa : any paper about "Precomputed Radiance Transfers" because google doesn't like it

The SCEA paper speaks about "basis functions" and "Orthogonal polynomials"
 
The Armored Core screens raised my eye brow, but reading all of this (which is all news to me) has me excited to see what this thing can really do...
 
Kaching said:
So are you seeing that members of the general dev community refer to PRT as SH because of this or simply because "Spherical Harmonics" sounds sexier?
Click to expand...
Well I guess it depends who you talk to - the people I know call it PRT. The original paper called it PRT... etc.
It's not a big deal, but Spherical Harmonics is a name of a mathematical term, not an algorythm. It's almost as bad as calling Phong shading - Dot product shading, because it involves evaluating a few dot products (there's probably better analogies, but it's pretty late here :p).

Milhouse,
I just googled "Precomputed Radiance Transfer" and got a bunch of relevant hits, including at least one SigGraph paper.
Maybe it's the last "s" that it didn't like in your search... :p

The SCEA paper speaks about "basis functions" and "Orthogonal polynomials"
Click to expand...
Spherical Harmonics belong to family of basis functions. It's possible to implement PRT shading using different basis functions.
 
DaCocoBrova said:
The Armored Core screens raised my eye brow, but reading all of this (which is all news to me) has me excited to see what this thing can really do...
Click to expand...

It was all news to me too - that's why I decided to post it here. With texture
compression, programmable HOS, AA and pixel and vertex shaders, PSP sure sounds intriguing.
 
^^ whoa, whoa - where did you get all that? Some of those are features in the PSP, but I don't know about AA and pixel shaders...
 
kaching said:
^^ whoa, whoa - where did you get all that? Some of those are features in the PSP, but I don't know about AA and pixel shaders...
Click to expand...

From Sony's E3 PSP slides. It's an old news my friend.
 
Kaiser responds: In practice, those paper specs mean nothing. It's really up to developers to decide how "good" the games will look. But from a very general standpoint (and a little of my own opinion), I'd say you could expect graphics somewhere along the lines of PS2 quality, though the PSP does have some features that the PS2 doesn't. For example, the PSP supports pixel shading. This is pretty big news. Pixel shaders allow each pixel on screen to be shaded individually. While it sounds kinda silly (maybe), there are a lot of cool things you can do with pixel shaders. Take the Xbox as an example. It also has pixel shaders, and that's how all those cool volumetric grass/hair effects are done. Neat huh?
Click to expand...

Because it is a portable, the size works in Sony's favor. A more impressive image can be drawn using less processing power. Because of this, games may even appear to look as good, if not better than PS2 games.

So is it out of the question to have better-than-PS2 graphics? Scaled down, not really.
Click to expand...




Link for Kaching
 
Fafalada said:
Milhouse,
I just googled "Precomputed Radiance Transfer" and got a bunch of relevant hits, including at least one SigGraph paper.
Maybe it's the last "s" that it didn't like in your search... :p
Click to expand...

:p
 
Err..okay, thanks, Izzy. I think I'm going to refrain from putting full belief in what IGN's mailbag guy is saying about PSP visual capabilities. Can we get confirmation from devs that there's built in pixel shader and AA support? I don't think these things were explicitly stated in Sony's E3 slides either.
 
Marconelly said:
It most likely wasn't, but regardless of that, some of the PSP demos demosntrated effects that I've never seen being used on the PS2 (not even in the demos)
Click to expand...

The Armored Core game is one of the most impressive so far. You really can't tell the difference between it and the PS2 game.
 
Can we get confirmation from devs that there's built in pixel shader and AA support?
Click to expand...
People sign NDAs to avoid having to confirm/deny things like that you know :p
Anyway, I can tell you that AA or not - the IQ will be a fair step beyond most PS2 games simply because of better implementation of mipmap LOD calc.
And if that will be a big improvement in any game, it will be GT4... :p
 
Fafalada, so if I design a new bike that does not crumble in pieces when I ride it, like the old one did, I can say that I have a better implementation of the concept of a bike ;) ?

You have soft words for LOD calculation on the GS, but you are also a nice guy who probably does not get that hot-tempered too easily, so that explains it :).
 
Fafalada said:
People sign NDAs to avoid having to confirm/deny things like that you know :p
Click to expand...

Yes, so that some SCEA guy can put in his GDC 2003 presentation slides that the GPU rendering pipeline is DirectX 7 level (pre-Pixel Shaders which appeared really with DirectX 8) and some unnamed SCEA guy can post in an OpenGL programmers forum a question on how to do various effects "considering an unreleased platform which has a fixed rendering pipeline" classifiable between OpenGL 1.1 and OpenGL 1.3.
 
Panajev said:
Yes, so that some SCEA guy can put in his GDC 2003 presentation slides that the GPU rendering pipeline is DirectX 7 level (pre-Pixel Shaders which appeared really with DirectX 8)
Click to expand...
How anal do we want to be with DX7 classification though?
DX7 specifies EMBM - a blend mode that is a "dependant texture read with a matrix transform"(at least 2xdotproduct). Possible speed-issues aside, that's all the math operations you need to do pixelshading - or at least, implement the core material shader of something like Doom3.
And if you have lots of texture-stages, and your dependant read is single cycle operation, you will get those shaders running comparably fast to your DX8.1 class hw (GF3...), as evidenced by NGC :p

If you want to proove yes/no pixelshading you'll have to get better evidence then that. ;)
 
Marconelly said:
It most likely wasn't, but regardless of that, some of the PSP demos demosntrated effects that I've never seen being used on the PS2 (not even in the demos)
Click to expand...

I did not reject the fact that PSP will/would do such things, even in its first gen. This is just the GT4 argument which is flawed.
 
Even if i don't understand half of what i read (due to my very poor math knowledge and my lack of expertise) this paper http://www.research.scea.com/gdc2003/spherical-harmonic-lighting.pdf is really interesting.

It's like a kind of cheap radiosity ( and of those talk of cubemap and radiosity where vaguely similar to an article i read a few year ago about a new unproved way to do radiosity http://www.gamedev.net/reference/programming/features/rtradiosity/ & http://www.gamedev.net/reference/programming/features/rtradiosity2/ )

Great read
 
Fafalada said:
People sign NDAs to avoid having to confirm/deny things like that you know :p
Anyway, I can tell you that AA or not - the IQ will be a fair step beyond most PS2 games simply because of better implementation of mipmap LOD calc.
And if that will be a big improvement in any game, it will be GT4... :p
Click to expand...
Why the would there be NDAs on those aspects of the hardware at this point, though? Is it still a question of whether such things will make the final hardware build?
 
kaching said:
Why the would there be NDAs on those aspects of the hardware at this point, though? Is it still a question of whether such things will make the final hardware build?
Click to expand...

I think he's already answered that question:


DX7 specifies EMBM - a blend mode that is a "dependant texture read with a matrix transform"(at least 2xdotproduct). Possible speed-issues aside, that's all the math operations you need to do pixelshading - or at least, implement the core material shader of something like Doom3.
And if you have lots of texture-stages, and your dependant read is single cycle operation, you will get those shaders running comparably fast to your DX8.1 class hw (GF3...)
Click to expand...
 
How anal do we want to be with DX7 classification though?
DX7 specifies EMBM - a blend mode that is a "dependant texture read with a matrix transform"(at least 2xdotproduct).
Click to expand...
Hmm yeah, let's not forget that DX7 offered a very solid, fixed function bump-mapping support. DX7 is basically the level Gamecube's hardware is at. I also wonder how close the developer's PSP emulator is to an actual hardware. I'm suspecting it might be missing some features. Demos made on the actual hardware devkit clearly used functions that none of the emulator-based games had implemented.
 
Marconelly said:
Hmm yeah, let's not forget that DX7 offered a very solid, fixed function bump-mapping support. DX7 is basically the level Gamecube's hardware is at. I also wonder how close the developer's PSP emulator is to an actual hardware. I'm suspecting it might be missing some features. Demos made on the actual hardware devkit clearly used functions that none of the emulator-based games had implemented.
Click to expand...

Yes, it is.

There are several features of the hardware that haven't fully detailed to developers yet, two of which are very important:- how to use vram, and the specs of the vector co-processor (which is *very* powerful).
Click to expand...
 
kaching said:
So are you seeing that members of the general dev community refer to PRT as SH because of this or simply because "Spherical Harmonics" sounds sexier?
Click to expand...
PRT is the first widespread use of Spherical Harmonics in gaming. SH is a pretty general concept, not restricted to lighting, but for the moment, confusion is inevitable.
 
You must log in or register to reply here.

Similar threads

  • Locked
The PS5 Pro is more powerful than my PC
15
2K
poppabk replied
Here's a list of some modern technology equal to/more powerful than the 8th gen gaming systems. (measured by TFLOPs)
6
136
kevboard replied
Saturn Was "More Powerful Than PlayStation" Claims Argonaut Founder
11 12 13 14 15
734
38K
Wolzard replied
Sony PS5 era profits exceed $13 billion, more than the profits from PS1-PS2-PS3-PS4 put together
3 4 5 6 7
304
3K
nial replied
Speculation: Will the Switch 2 be more powerful than Ally X ?
2
91
7K
kevboard replied
Top Bottom