Sony @ GDC: Phil Harrison's Keynote
- Thread starter Titanio
- Start date
_phil_ said:
Cheers, I did see this paper before. There's a range of approaches of varying sophistication presented even there, and I guess I'm just wondering what's most possible/likely. Are we talking about using slices, or 'full' ray marching with a high average number of steps with procedurally generated volumes, and full lighting/shadowing?
It'd be really nice if those cloud volumes were procedurally modified over multiple frames to mimic billowing and deformation etc, for example, but I'm thinking that might be less likely! But would that initial generation of the volume be less expensive than the actual rendering, anyway?
I really should just try this stuff out myself, if I had more time
A couple of more bits..
A guy from GAF says of the Motorstorm demo that the last one wasn't as barren as some people are being led to believe, that toward the end the camera panned out to a more distant vantage point to reveal upwards of 20 racers kicking around on the terrain etc.
Also, IGN's weekly show has coverage of GDC, with a brief chat with Phil Harrison and an interview with Peter Moore:
http://media.games.ign.com/articles/693/693580/vids_1.html
Boulding asks him how much of the stuff shown at GDC will be playable at E3, and Harrison replied that he doesn't see much reason why "everything" can't be playable at E3. Though Boulding follows up, asking if the devs know that, with Harrison replying that if they didn't, they will when they see the show, so it doesn't sound like it's all sorted out quite yet
Also, on Warhawk clouds, the same guy who informed about Motorstorm above, mentioned that the clouds seemed to be "static" in that they didn't change shape. Which at least answers one of my questions about procedural modification of the clouds over time (Though tbh, I'm not sure how easily you might spot deformation in the clouds in a demo, unless the speed was exagerrated..it seems like something that could be done, though, if the dev wanted, as it strikes me that the harder part would be the rendering..?)
A guy from GAF says of the Motorstorm demo that the last one wasn't as barren as some people are being led to believe, that toward the end the camera panned out to a more distant vantage point to reveal upwards of 20 racers kicking around on the terrain etc.
Also, IGN's weekly show has coverage of GDC, with a brief chat with Phil Harrison and an interview with Peter Moore:
http://media.games.ign.com/articles/693/693580/vids_1.html
Boulding asks him how much of the stuff shown at GDC will be playable at E3, and Harrison replied that he doesn't see much reason why "everything" can't be playable at E3. Though Boulding follows up, asking if the devs know that, with Harrison replying that if they didn't, they will when they see the show, so it doesn't sound like it's all sorted out quite yet
Also, on Warhawk clouds, the same guy who informed about Motorstorm above, mentioned that the clouds seemed to be "static" in that they didn't change shape. Which at least answers one of my questions about procedural modification of the clouds over time
(Though tbh, I'm not sure how easily you might spot deformation in the clouds in a demo, unless the speed was exagerrated..it seems like something that could be done, though, if the dev wanted, as it strikes me that the harder part would be the rendering..?)Megadrive1988
Veteran
Ratchet & Clank makes me want to see how well PS3 could reproduce Final Fantasy 12 CG, even if it was just a realtime non-interactive demo like the FF7 technical demo.
croc hunter2
Newcomer
!eVo!-X Ant UK
Banned
[Brick_top]
Newcomer
I know it's something that is asked a lot around here but are these confirmed to be realtime?
!eVo!-X Ant UK
Banned
[Brick_top] said:
Yes, there from a fly by cut-scene ( like the one's actually present in the current gen R&C games )
Titanio said:
I don't have a clue
However, the clouds are quite simple in strucure, like they have very low iterations on their fractal components - so it might be an advanced method used with low detail settings. They might be undersampling too (1 sample for each n pixel instead of 1-n sample per pixel). As I've said before, we need direct feed screenshots to judge details...
I think that's the problem. You seem to think of physics as just objects colliding and getting knocked about? It can also be animation, lighting, weather, and open-ended interactivity. Physics can be more materials based, allowing for softer surfaces like sand and snow to deform underfoot, adding to realism, and take energy from systems, making motion across different surfaces more tiring. For Football, physics will eventually produce true-to-life animations with more realism. As I gave in an earlier example, physics produces more open ended gameplay, making world's more interactive and open ended, which is what you want. For racing, Forza or GT can't improve much in normal racing gameplay, but can add 40 cars simulataneoulsy, and more importantly can produce off-road and non-conventional racers. Mercury shows the application of physics in a new puzzler as fluid dynamics. No-one knows what advanced fluid dynamics can add to a game because we've never had the processing power to do that. AndI think you missed the main point I was making that in having an advanced physics engine, you decrease the amount of work needed to create better games. In the football example, for 10 types of tackle you need 10 motion-captured animations. With skeleton synthesis you need one physics engine, which will produce animation for all and every tackle. With Elder Scrolls there's maybe a few mocap animations for a few different weapons. In the next Elder Scrolls there'll be perfect animations for all weapons, and the chance to use anything as a weapon, and a chance to fabricate your own weapon, all from the one engine without anyone have to decide what weapons to add, what animations to use, etc.scooby_dooby said:
The most advanced engines would allow the greatest freedom with the least creative effort. As a simple example, consider scripting 1000 dominoes in an animation, versus 1000 rigid bodies that interact with a physics engine. The latter is more versatile, adaptable and realistic. Physics can contribute to many of the things you want in better games, and other things like story don't need any advances in processing power so physics needn't be sacrificed to support them.
overclocked
Veteran
I certainly agree with that physics is the most important thing besides the base-quality overall.
Its all about better graphics and physics has everything to do with that also.
Creating waves in the ocean or hair blowing in the wind etz. Physics isnt just about an "explosion" or "correct" animation, it affects everything so i think its one of the most important things in this gen of consoles.
Just want to add that there is a confusion or rather different views what physics in games are, i dont think anyone really is right/wrong cause its a very complicated subject IMHO.
Its all about better graphics and physics has everything to do with that also.
Creating waves in the ocean or hair blowing in the wind etz. Physics isnt just about an "explosion" or "correct" animation, it affects everything so i think its one of the most important things in this gen of consoles.
Just want to add that there is a confusion or rather different views what physics in games are, i dont think anyone really is right/wrong cause its a very complicated subject IMHO.
After initial difficulties I bet better physics handling can contribute to the reduction of production cost as except for some glitches caused by unrealistic game design conditions such as too fast response to user control, physics system can construct a consistent world with reality with little designer input.
I think the problem is we are comparing lots of precanned physics to the new realtime stuff.
To give a analogy it's like comparing a model with high rez textures vrs one with complex shaders. Textures alone are not interactive or dynamic. Precaned physics are the same way they may look good but they only work under a specific situation. Real time dynamic physics should allow complex physics to apply to a virtually unlimeted number of situations. From a development perspective it's preferable to construct models that can react realistically to physics than having to script physics for 100s of different possible events.
End users and even casual gamers will notice the difference. Screenshots may not capture it but it can have a dramatic effect on gameplay. Things will look and behave more realistically in the game world.
This could also save lots of dev time, imagine if we can simulate a complex skeletal, muscular, and fat body system. It would completely change the way we animate characters, the current time consuming process could be revolutionized. One of the biggest things that breaks immersion in games is the way animation is handled.
In the long term I don't think the current way of doing things will scale well. We can't just keep adding thousands of motion capped precaned animations or even more tedious animations done by hand and expect this to simulate a dynamic interactive environment. The current method lacks consistency and realism, game players will notice an improvement. You can look at different Cg movies with different animation quality as a testament to how dramatic the effect is.
From software rendering we have a good idea on the amount of processing required to do these things. Cell from a theoretical technical point of view should be able to handle some software rendering techniques that where previously not achievable in real time. Being able to composite these effects with hardware accelerated graphics creates lots of possibilities.
Many people downplay cell and like to think it's all about the GPU. The reality is CELL does make ps3 very different from xbox360. Where as the differences in gpu architectures will not allow one platform to create fundamentally diff. content, because the GPUs are still very similiar in function(performance may vary). The only question is how much CELL will allow for unique content on ps3.
These demos show some of the possabilities. They are targeted at developers, and in that light they are impressive. In the future ps3 games will be able to have lots of software rendering as well as hardware rendering. This does not implicate a deficiency in RSX compared to other GPUs. Their are still many techniques used in the "software world" that GPUs cannot handle. We already have a good idea what types of effects RSX can produce by looking at PC GPUs running at similar rez, bandwidth, etc. These demos are meant to show us techniques that aren't currently well known and/or used. So you should not be upset or disappointed if a demo does not contain all the features you expect in a final game visual or otherwise.
To give a analogy it's like comparing a model with high rez textures vrs one with complex shaders. Textures alone are not interactive or dynamic. Precaned physics are the same way they may look good but they only work under a specific situation. Real time dynamic physics should allow complex physics to apply to a virtually unlimeted number of situations. From a development perspective it's preferable to construct models that can react realistically to physics than having to script physics for 100s of different possible events.
End users and even casual gamers will notice the difference. Screenshots may not capture it but it can have a dramatic effect on gameplay. Things will look and behave more realistically in the game world.
This could also save lots of dev time, imagine if we can simulate a complex skeletal, muscular, and fat body system. It would completely change the way we animate characters, the current time consuming process could be revolutionized. One of the biggest things that breaks immersion in games is the way animation is handled.
In the long term I don't think the current way of doing things will scale well. We can't just keep adding thousands of motion capped precaned animations or even more tedious animations done by hand and expect this to simulate a dynamic interactive environment. The current method lacks consistency and realism, game players will notice an improvement. You can look at different Cg movies with different animation quality as a testament to how dramatic the effect is.
From software rendering we have a good idea on the amount of processing required to do these things. Cell from a theoretical technical point of view should be able to handle some software rendering techniques that where previously not achievable in real time. Being able to composite these effects with hardware accelerated graphics creates lots of possibilities.
Many people downplay cell and like to think it's all about the GPU. The reality is CELL does make ps3 very different from xbox360. Where as the differences in gpu architectures will not allow one platform to create fundamentally diff. content, because the GPUs are still very similiar in function(performance may vary). The only question is how much CELL will allow for unique content on ps3.
These demos show some of the possabilities. They are targeted at developers, and in that light they are impressive. In the future ps3 games will be able to have lots of software rendering as well as hardware rendering. This does not implicate a deficiency in RSX compared to other GPUs. Their are still many techniques used in the "software world" that GPUs cannot handle. We already have a good idea what types of effects RSX can produce by looking at PC GPUs running at similar rez, bandwidth, etc. These demos are meant to show us techniques that aren't currently well known and/or used. So you should not be upset or disappointed if a demo does not contain all the features you expect in a final game visual or otherwise.
Last edited by a moderator:
As it turns out, despite the scheduling mess-up, Lair was shown at GDC. From Matthias Worch's blog (Factor5 artist):
http://www.langsuyar.com/news/blog.html
According to second-hand reports on GAF, one of the demos showed some dragons flying in a moutainous environment. Mentions also HDR and that the dragon's wings have air resistance (!)
People have been asking Worch to put the slides up on his blog since so many missed out, but he said that most of the presentation was focussed around the demonstrations, and he won't be able to put them on the web for now
http://www.langsuyar.com/news/blog.html
According to second-hand reports on GAF, one of the demos showed some dragons flying in a moutainous environment. Mentions also HDR and that the dragon's wings have air resistance (!)
People have been asking Worch to put the slides up on his blog since so many missed out, but he said that most of the presentation was focussed around the demonstrations, and he won't be able to put them on the web for now
mckmas8808
Legend
Titanio said:As it turns out, despite the scheduling mess-up, Lair was shown at GDC. From Matthias Worch's blog (Factor5 artist):
http://www.langsuyar.com/news/blog.html
According to second-hand reports on GAF, one of the demos showed some dragons flying in a moutainous environment. Mentions also HDR and that the dragon's wings have air resistance (!)
People have been asking Worch to put the slides up on his blog since so many missed out, but he said that most of the presentation was focussed around the demonstrations, and he won't be able to put them on the web for now
So sad. I really wanted to see that demo too. At least a screenshot.
Laa-Yosh said:
Can't you? I can see them. Most prominent if at about the centre of the picture, on the monster thing, on what i think is its shoulder or its upper chest...
And if i see them, you should see 10 times more, knowing you!
It is a really bad shot though, fast movement and all...Similar threads
