Sony @ GDC: Phil Harrison's Keynote

[London Geezer]

Your racer in Waverace did not spawn and create waves in its wake, etc. that affected other players, you didn't affect the environment..you couldn't splash up waves over other players etc.(though I don't know even if it did if that would be the "same thing" as what's being done in Motorstorm, same concept of player affecting the environment, but not sure if it'd be technically as demanding/sophisticated).

As i said, of course Waverace had a primitive way to do things compared to Motorstorm, but that's due to technical differences that are obvious.
I'm quite sure that either Waverace or some other water-scooter racing game had racers that did create waves, affecting the race of whoever was behind him. I could be wrong but i don't think that's very important anyway...
The principle is exaclty the same which is why i posted that before, let's not mistake Motorstorm for some amazing new idea that's never been done before. It just takes an idea that's been done before and adapts it to its situation.

 

[StefanS]

Your racer in Waverace did not spawn and create waves in its wake, etc. that affected other players, you didn't affect the environment..you couldn't splash up waves over other players etc.(though I don't know even if it did if that would be the "same thing" as what's being done in Motorstorm, same concept of player affecting the environment, but not sure if it'd be technically as demanding/sophisticated).

As a matter of fact, in Waverave GCN your wave did affect other player, and you were affected by it.

 

[Shifty Geezer]

The people who viewed the Demo have stated that the resulting divits in the ground actually effects the steering of the vehicle. Meaning when you create a path for yourself in the Mud someone whos behind you gets effected by the trails in the Mud...

Without a force feedback wheel, won't that just be annoying? You push left and don't go left because of the road, but wouldn't feel the resistance of the road as you would with a wheel, so it would just feel like inneffectual steering IMO.

 

[London Geezer]

As a matter of fact, in Waverave GCN your wave did affect other player, and you were affected by it.

Thanks for confirming it, i wasn't 100% sure.

 

[Titanio]

As i said, of course Waverace had a primitive way to do things compared to Motorstorm, but that's due to technical differences that are obvious.
I'm quite sure that either Waverace or some other water-scooter racing game had racers that did create waves

I wasn't sure about this, but was referring just to 64. But as is, if you have an engine that can spawn waves from a given point which I think would be fairly easy to do, doing so in the wake of players isn't so difficult, you don't subsequently have to worry about what happens to them any more than any other waves, and I don't think you're actually deforming any existing waves as such (but maybe blending some new ones in with others?). And these games weren't exactly using complex fluid dynamics or the like.. ; )

One imagines dealing with 'mud' and having mud that can dry out either splattered on cars or as more persistant grooves in the track itself, with the potential for getting stuck in the mud even etc. is somewhat more complex. We haven't seen a (off-road) track racer with that level of environment interaction, as far as I'm aware?

Well, to composite the clouds, you need depth info on the scene. It makes more sense to let RSX render the frame and then read the framebuffer and Zbuffer data in small blocks into the SPUs. There's probably some more optimizing involved though, as it'd need quite a few hundred megs of data per frame, especially with 4x AA (4 times the Z samples).

Then again, the cloud data only requires a depth input, and produces a color output, so if they let RSX to do the final blending, then they won't have to read the framebuffer into the SPUs, only the Z-buffer. So the SPUs could render into a texture, that RSX would simply read in from the XDR RAM and add on top of the framebuffer. Still, GDDR bandwith is probably scarce enough already, with 4x AA, so maybe they can do some trickery to minimize the Z-reads too...

Hadn't thought about the first way before..in some instances it might be better. In the second, though, I doubt the bandwidth required by RSX for the final blend would really be all that much at all?

Now, about the raytracing part, why can't they call it raymarching, cause AFAIK that's the proper name?

I've seen ray-marching referred to in the context of cloud and atmospherics rendering before, so it's quite possible this is what they're doing. edit - though doing a google, is ray marching perhaps not ray tracing but accounting for light scattering..? Does this allow you to introduce some assumptions that simplifies things a bit?

I remember reading a small bit about Heavenly Sword in a Ninja Theory profile in Edge some time ago, and they were talking about using Cell for, amongst other things, clouds. I wonder if they're doing anything quite as ambitious as this? (Yeah yeah, I'm teasing for info..for all I know they might 'just' be generating some perlin noise on the CPU to feed to the GPU ).

edit - side note, but harping back to the argument of "what can physics do for my game?", but this is a really neat comparison using GRAW on a regular CPU, and on a PhysX-enabled PC:

http://physx.ageia.com/footage.html (<-- footage there)

That's 'just' effects physics. Though I'm sure it's not exactly pushing it to the limit, the PhysX-enabled version clearly looks an awful lot better than the regular PC version, and it's a fairly clear demonstration of how physics can help visually, also. There'll be a lot more where this came from, and lot more sophistication going forward I think.

 

[Deepak]

Where is my beloved KZ?

:smile:

 

[War]

Pics of new Sony dev kit

http://www.ps3focus.nl/sitee/nieuws/bekijk.php?id=1073&cat=catps3

 

[Pugger]

Titanio, although I understand where your coming from on those comparisons pictures, they really arn't that impressive. I have been playing Full Auto and although the game itself is as mediocre as they come the physics engine is just as impressive as those pictures. There are no pre baked physics in the game and this really does show up, blow up a petrol station in the game and watching the resulting domino effect on surrounding objects is very impressive indeed. In fact the best thing about the game is phyics engine which I'm sure could be used in other tittles.

 

[one]

new pics/vids of the demo booth
http://ps3.ign.com/articles/698/698086p1.html

 

[Titanio]

Pugger - the point was not to hold GRAW out and say "isn't this just the ultimate application of physics ever!" This is not what I mean to say at all. The point is that it's a nice comparison between what the same developer has done with less and more physics capability. If another developer can do more respectively with both levels of capability - and I'm sure that will be the case if it hasn't been already - that's great, but it's better when we can compare what the same developer has done at those levels, as the above comparison does.

 

[Platon]

edit - side note, but harping back to the argument of "what can physics do for my game?", but this is a really neat comparison using GRAW on a regular CPU, and on a PhysX-enabled PC:

http://physx.ageia.com/footage.html (<-- footage there)

That's 'just' effects physics. Though I'm sure it's not exactly pushing it to the limit, the PhysX-enabled version clearly looks an awful lot better than the regular PC version, and it's a fairly clear demonstration of how physics can help visually, also. There'll be a lot more where this came from, and lot more sophistication going forward I think.

No doubt physics can enhance games, both from a gameplay point of view, as well as visually. However, I really don't get the above example. What is done with the PhysX could be doen even without it through scripting, sure, it would not be realistic physics but the visual impact could be more or less that same...

 

[London Geezer]

Yes, that example was obviously a vastly inflated PR kind of comparison. Any high-end computer today can get explosions that are much much better than that first screen, even without the PPU.
Obviously AGEIA want people to believe that the PPU can do miraculous things, by showing examples to suit their marketing.
That example for instance used the "medium" settings on the PC version without the PPU, and the "hardware" settings for the PPU version, which is just a totally different path that makes use of the PPU.
We've been getting explosions that are much better than that PPU-less video for years even on old consoles. Black on PS2 and Xbox comes to mind.
I mean look at the first screen, there's nothing there!! The explosion is a couple of low-opacity alpha polygons and that's it. Really not indicative of real capabilities of PCs today.
Sure the PPU will makes things explode better, but that example is very misleading.

All these companies are in it for the money, AGEIA is no different, they will show us videos to "prove" that you can get prettier explosions in games. I think they need to do much better than that or they won't sell much.

Their dynamin fluid demos were more impressive, but still i'm not sure they will convince a lot of people that the PPU is actually needed. At least not any more than NVIDIA and ATI can convince gamers that they "need" their ridiculously priced highest-end GPUs.

 

[mckmas8808]

Pugger - the point was not to hold GRAW out and say "isn't this just the ultimate application of physics ever!" This is not what I mean to say at all. The point is that it's a nice comparison between what the same developer has done with less and more physics capability. If another developer can do more respectively with both levels of capability - and I'm sure that will be the case if it hasn't been already - that's great, but it's better when we can compare what the same developer has done at those levels, as the above comparison does.

Exactly. Imagine when devs start to use gameplay physics. I can't wait for the game that allow me to blow up a truck like that and the tires, grill, and other pieces of the truck flying off can actually kill or hurt the enemy instead of my having to shoot him.

Effects physics are nice and I love them, but gameplay physics is where it's at.

 

[Phil]

No doubt physics can enhance games, both from a gameplay point of view, as well as visually. However, I really don't get the above example. What is done with the PhysX could be doen even without it through scripting, sure, it would not be realistic physics but the visual impact could be more or less that same...

Scripting sequences takes time and may or may not look fake depending on where the sequence is triggered. It's not that much different than comparing this generation where most 1st generation racers had fake reflections running in the car's windows vs. what could be done with either ray-tracing or other methods. It's not hard to say that the latter would be much more realistic, but also the more sophisticated and technically demanding approach.

The same applies to dynamic physics or prebaked sequences. The former certainly will be much more impressive and there probably aren't many artists that are willing to make the latter make as good looking as the former (by making hundreds of sequences for every single action the user may or may not perform).

 

[Phil]

All these companies are in it for the money, AGEIA is no different, they will show us videos to "prove" that you can get prettier explosions in games. I think they need to do much better than that or they won't sell much.

Sure, that goes for any company that wants to sell their product. Yet, I must say I find the physics-processing-unit a great idea to have developers improve graphics and gameplay through physics on a whole different level. Sure, you can use *any* processor outthere to perform complex physic calculations, yet depending on how well suited that processor is, it may or may not be feasable. By having a processor that can be used for dedicated physics number-crunching, you can sort of have the best of everything without some trade-off.

This is where I find the concept of CELL and the direction Sony is taking with it to be a brilliant one: I want more interactions in my games [through physics] - and given CELLs strengths in performing such calculations, I can see many developers actually taking advantages of those strengths. At the end of the day, game development and the games that result through them are a direct result of the hardware. Just as Nintendo's Revolution will attract innovation by having developers use that new and (r)evolutionary controller, I see many developers on PS3 that are going to innovate by using CELLs strengths. I see added physics and their interaction in games to be a direct result of that.

Hopefully developers will use that and the above described info on what Motorstorm is supposedly pushing, I am excited by what else can and will be done by very talented developers. Sure, no one is directly saying that nothing of that won't be possible on Xbox360 - that isn't really the point at all. I think it is fair to say though that developers that will use different hardware to make different games by using its different strengths, we'll see games of each kind that will only be possible on the console they were targeted. I see one of these differences to be in CELLs floating point performance.

 

[London Geezer]

Sure, that goes for any company that wants to sell their product. Yet, I must say I find the physics-processing-unit a great idea to have developers improve graphics and gameplay through physics on a whole different level. Sure, you can use *any* processor outthere to perform complex physic calculations, yet depending on how well suited that processor is, it may or may not be feasable. By having a processor that can be used for dedicated physics number-crunching, you can sort of have the best of everything without some trade-off.

This is where I find the concept of CELL and the direction Sony is taking with it to be a brilliant one: I want more interactions in my games [through physics] - and given CELLs strengths in performing such calculations, I can see many developers actually taking advantages of those strengths. At the end of the day, game development and the games that result through them are a direct result of the hardware. Just as Nintendo's Revolution will attract innovation by having developers use that new and (r)evolutionary controller, I see many developers on PS3 that are going to innovate by using CELLs strengths. I see added physics and their interaction in games to be a direct result of that.

Hopefully developers will use that and the above described info on what Motorstorm is supposedly pushing, I am excited by what else can and will be done by very talented developers. Sure, no one is directly saying that nothing of that won't be possible on Xbox360 - that isn't really the point at all. I think it is fair to say though that developers that will use different hardware to make different games by using its different strengths, we'll see games of each kind that will only be possible on the console they were targeted. I see one of these differences to be in CELLs floating point performance.

I agree 100%, we need more interaction and physics is the way to go, and the PPU will help in the PC world, i was only commenting on the example they provided, which is just very misleading.
What sets me back is that with new Quad-core AMD and Intel CPUs, these machines will have some serious power in them, and i'm not sure what a PPU can do that you couldn't do with these monster CPUs with lots of cores to spare. I guess cost comes into the equation now...

 

[expletive]

edit - side note, but harping back to the argument of "what can physics do for my game?", but this is a really neat comparison using GRAW on a regular CPU, and on a PhysX-enabled PC:

http://physx.ageia.com/footage.html (<-- footage there)

That's 'just' effects physics. Though I'm sure it's not exactly pushing it to the limit, the PhysX-enabled version clearly looks an awful lot better than the regular PC version, and it's a fairly clear demonstration of how physics can help visually, also. There'll be a lot more where this came from, and lot more sophistication going forward I think.

I actually saw this and was wholly unimpressed. Probably becuase ive seen physics equal to this 'accelerated' video in 'non-accelerated' games. I find it a little odd that this is the BEST Ageia could come up with to prove a point. Do they really think that footage is going to sell $300 physics cards to gamers?

Take full-auto as an example, the physics in that game is, imo, superior to what we're seeing in this ageia video and it couldnt save that game...

I'm all for the nirvana of fully realized photorealistic worlds and physics to match but what theyve been showing isnt that, its more like physics 'effects', like the HDR version of physics but not close to 'global illumination'. The stuff in motorstorm with deformable road and ruts in the mud seems fairly useless to me. Unless youre in a jeep and can feel the car tilt or the steering wheel jerk becuase of the ditch, is it really going to matter?

I understand where theyre going, where we would like it to be, , and this generation may be a necessary step towards that, but i think in terms of gameplay i have yet to feel like i NEED an ageia card in my PC. (FWIW i bought the first voodoo 3d card, the orchid righteous 3d, so i probably fall into the category of 'early adopter')

 

[Npl]

Im asking myself if the PPU will result in the "General Path" beeing a aftertought. Not so much when it comes to Gameplay, but Im sure AGEIA wants a distinct difference to sell the Cards.
Kinda like 3DMark dropped optimized T&L Paths (for 3DNow, SSE) and suddenly it was HW T&L vs 'runs-on-all' T&L. Needless to say H&L immediatly looked alot better compared to the single alternative

 

[Titanio]

I actually saw this and was wholly unimpressed. Probably becuase ive seen physics equal to this 'accelerated' video in 'non-accelerated' games. I find it a little odd that this is the BEST Ageia could come up with to prove a point.

It's a real world example of the difference their technology brings in that game. The best demonstrators, they might say, from a tech pov might be their tech demos.

I'll again say to you what I said to Pugger, though - put it another way, if you think Full Auto does better than this without "acceleration" (though, in fairness, there's more to work with in 360 than on a PC CPU anyway, which is what they're comparing against here), think what more they could do with it

I'm all for the nirvana of fully realized photorealistic worlds and physics to match but what theyve been showing isnt that, its more like physics 'effects', like the HDR version of physics but not close to 'global illumination'.

But you agree that the physics here makes the game look better than it otherwise would, no? And what's wrong with that? Most of the advances made to date in technology for game's sake has been "just" about making things look better.

The stuff in motorstorm with deformable road and ruts in the mud seems fairly useless to me. Unless youre in a jeep and can feel the car tilt or the steering wheel jerk becuase of the ditch, is it really going to matter?

I think the point is, with Motorstorm, that it will. The terrain as deformed by you or others will affect your handling etc. at least as far as I understand.

e.g. from gamespot:

it appears that this deformation actually does have an impact on gameplay, since the suspension and wheels of the car bounced over the ruts that were being created. It'll be interesting to see how this effect impacts gameplay. In a multilap race, you could conceivably wind up having to stick to the ruts worn into the turns if you want maximum speed, whereas pristine mud would slow you down.

http://www.gamespot.com/ps3/driving/motorstorm/news.html?sid=6146416&mode=previews

 

[Mmmkay]

Wow, reading through this thread, it looks like you guys have been having fun!

Erhum, yeah so Gamefront has a link to a video of Phil Harrison's Q&A session here. At 19:41 and Phil says:

"Software will be region free. Erm and it is possible for developers at their choice to put onto the software all of the TV formats that they want to choose; PAL, NTSC, and all the HD modes that they want to choose. But if they want to divide it by language, divide it by region that's their choice but the concept is that software is region free."

So it's not that developers have the option to 'restrict' to a region, but that they have the freedom to develop their software as they currently do and still market different regional versions. Opening and closing with 'software will be region free' suggests that it is a platform policy rather than at publisher discretion.

Also, with regards to MotorStorm not being representative of a 'game' per se, Phil explained that they stripped out most of the game for the sole purpose of demonstrating the technology involved rather than 'showcasing' MotorStorm. This is GDC folks, remember that.