Jump to content

Discuss the future of game graphics


Moz
 Share

Recommended Posts

I said something similar in the Alice thread when dood uploaded a video of the psysx on display int hat game, but i think the next gen really will be all about better animation and physx style physics. Id be more than happy to take todays graphics in the next gen if they added in proper physx stuff. Seeing that shit in motion is unreal, but it needs a fair bit of grunt to power it. I makes game worlds feel much more alive and interactive. And games really will need better animation too. Not that theyrte all bad, but look at the witcher 2. Gorgeous, gorgeous game, but as soon as Geralt starts moving it breaks the immersion totally. You get used to it after a while but its still terrible animation.

Link to comment
Share on other sites

That was a great talk and it's nice to see that we have improved on Moore's initial calculations. But a lot of science fiction in there. In 8 years we will have full immersion VR and imperceptible AI personality is fantasy! I love singularity conversations and the new era of human evolution it will usher in, but it won't happen that soon. Imo.

Link to comment
Share on other sites

I said something similar in the Alice thread when dood uploaded a video of the psysx on display int hat game, but i think the next gen really will be all about better animation and physx style physics. Id be more than happy to take todays graphics in the next gen if they added in proper physx stuff. Seeing that shit in motion is unreal, but it needs a fair bit of grunt to power it.

That sort of stuff benefits from the massively parallel nature of the GFX card, which unfortunately will still be tied up with doing menial things like making the pretty pictures, the choices of CPU out there aren't exactly going to be performance competitive at doing that.

Theoretically, the CELL/Xenos combo could have done something similar this gen, but the SPUs are stuck assisting the geriatric RSX instead of more interesting things.

Link to comment
Share on other sites

That was a great talk and it's nice to see that we have improved on Moore's initial calculations. But a lot of science fiction in there. In 8 years we will have full immersion VR and imperceptible AI personality is fantasy! I love singularity conversations and the new era of human evolution it will usher in, but it won't happen that soon. Imo.

We can have a big discussion about the likelihoods of a technological singularity in a different thread. Regardless of people's opinions of the wider ramifications of his theories, Kurzweil's research into current computer speeds and their near-term future is absolutely rock-solid and proves my point that computer speed is now doubling every year.

Link to comment
Share on other sites

Bollocks, Intel disagree with your/Kurzweil assertion that we are moving ahead of Moore's Law for computing. The whole point of the original observation was to show how much power you could get into a reasonable sized/priced package over time.

The complexity for minimum component costs has increased at a rate of roughly a factor of two per year... Certainly over the short term this rate can be expected to continue, if not to increase. Over the longer term, the rate of increase is a bit more uncertain, although there is no reason to believe it will not remain nearly constant for at least 10 years. That means by 1975, the number of components per integrated circuit for minimum cost will be 65,000. I believe that such a large circuit can be built on a single wafer

NwBOw.jpg

http://www.intel.com/technology/mooreslaw/

You certainly can't buy a PC with double the computing power every 12 months for a similar price.

Link to comment
Share on other sites

  • 2 weeks later...

Seems there has been some progress on something first revealed back in 2008!

An Australian bloke by the name of Bruce Robert Dell has been plugging away at a form of Voxel-based rendering for a while, the engine currently does what it shows in the video at 20fps in SOFTWARE only, if the mooted Voxel-hardware renderer that Carmack discussed back in 2008 ever got made (or next gen GPU hardware is good enough to accelerate other forms of rendering, apart from a polygon-based one), the performance would take right off. He seems to have solved some of the problems associated with the dev pipeline and the fact the industry revolves around the polygon.

Link to comment
Share on other sites

Seems there has been some progress on something first revealed back in 2008!

An Australian bloke by the name of Bruce Robert Dell has been plugging away at a form of Voxel-based rendering for a while, the engine currently does what it shows in the video at 20fps in SOFTWARE only, if the mooted Voxel-hardware renderer that Carmack discussed back in 2008 ever got made (or next gen GPU hardware is good enough to accelerate other forms of rendering, apart from a polygon-based one), the performance would take right off. He seems to have solved some of the problems associated with the dev pipeline and the fact the industry revolves around the polygon.

When you watch the video games industry for as long as many of us have, you see many "infinite" detail engines come and go. I really would like it to be genuine of course, but I just think this would be making bigger waves if there weren't loads of caveats.

Link to comment
Share on other sites

When you watch the video games industry for as long as many of us have, you see many "infinite" detail engines come and go. I really would like it to be genuine of course, but I just think this would be making bigger waves if there weren't loads of caveats.

Well the main problem is how graphics architecture has evolved, and is all based around polys, and poly routine efficiency. That's basically the only thing holding it back.

Link to comment
Share on other sites

I think voxel-sculpting for asset creation then conversion into polygons, like the ProcWorld guy is doing, has a better chance of taking off, purely because it works with all existing tech, and because it offers productivity benefits over manual sculpting.

Voxels just seem like they're never going to take off, regardless of the benefits, since polygons are already established. Just like how we're not having this conversation on ternary computers, despite them being more powerful and power-efficient than binary computers, and that the industrial revolution never happened with the Stirling engine, because they were pretty good at building Steam Engines that didn't explode by the time it came around, despite it being the theoretically most efficient heat-conversion engine.

The best technology is not always the one that prospers.

Link to comment
Share on other sites

Whilst I'm certainly open to the idea of a revolution in graphics rendering, pretty much everything about that video smacks of snake oil. It says a lot but tells me nothing.

I'd love to be proven wrong, but to me the presentation comes across as someone sniffing for grant money, not a programmer extolling the virtues of their unique modelling solution. Perhaps those more versed in computer graphics technology can offer a different perspective, though.

Link to comment
Share on other sites

In this video (from a while ago) he states that using this technology on the Wii would enable it to surpass PS3 levels of visual fidelity. :omg:

http://www.gametrailers.com/video/technology-comparison-unlimited-detail/718415

I believe.

In all honesty though, I hope this becomes a viable alternative and/or solution to our current graphics capabilities but as the creators of the tech themselves state: If put to use by the big graphics manufactures, it would result in massive losses on their end. As there is no room for improvement it would be up to the artists to fully make use of Unlimited Detail.

Link to comment
Share on other sites

as the creators of the tech themselves state: If put to use by the big graphics manufactures, it would result in massive losses on their end. As there is no room for improvement it would be up to the artists to fully make use of Unlimited Detail.

Yeah, and fully electric cars have been around for years - but the masons keep them secret to protect the interests of oil companies. shhhhh

EDIT: Seriously this smacks of that.

Link to comment
Share on other sites

Thinking on it, though making things out of atoms (voxels) is indeed a sound idea for having Unlimited Detail, things get a bit (infinitely?) more complex in terms of animation/physics etc. For a cube, instead using 1-6 points of data for transformations/rotations/physics calculations, you're basically going to have to use however many million atoms your box is made of.

Link to comment
Share on other sites

Thinking on it, though making things out of atoms (voxels) is indeed a sound idea for having Unlimited Detail, things get a bit (infinitely?) more complex in terms of animation/physics etc. For a cube, instead using 1-6 points of data for transformations/rotations/physics calculations, you're basically going to have to use however many million atoms your box is made of.

That's all being handled by the engine though (which for the record is still running on a software based acceleration). As seen in the videos, you basically scan in an item of your choosing and the engine will take care of the rest. That rock you see in one of the videos is basically a real rock. It looks exactly the same as it does in real life. Imagine what artists could do with that when the engine takes full use of the hardware side of things.

That is, if they can deliver upon their promises and this isn't all a big hoax.

Link to comment
Share on other sites

Thinking on it, though making things out of atoms (voxels) is indeed a sound idea for having Unlimited Detail, things get a bit (infinitely?) more complex in terms of animation/physics etc. For a cube, instead using 1-6 points of data for transformations/rotations/physics calculations, you're basically going to have to use however many million atoms your box is made of.

You should read the John Carmack interview from 2008 discussing id Tech 6, which at the time was planned to be a hybrid Voxel/Poly engine, he explains why Voxels can become competitive when your triangle size becomes sub-pixel.

There are different approaches that you could wind up and try to get that done that would involve tessellation and different levels of triangle meshes and you can could conceivably make something like that work but rasterization architecture does really start falling apart when your typical triangle size is less than one pixel. At that point you really have lost much of the benefits of rasterization. Not necessarily all of them, because linearly walking through a list of primitives can still be much faster than randomly accessing them for tracing, but the wins are diminishing there.

This demo from a commercial company whose product is actually in real world use by major companies is a slightly more impressive demo for what point clouds can do:

Link to comment
Share on other sites

That's all being handled by the engine though (which for the record is still running on a software based acceleration). As seen in the videos, you basically scan in an item of your choosing and the engine will take care of the rest. That rock you see in one of the videos is basically a real rock. It looks exactly the same as it does in real life. Imagine what artists could do with that when the engine takes full use of the hardware side of things.

That is, if they can deliver upon their promises and this isn't all a big hoax.

It's a real rock, but rocks are very easy to scan. The tree was made by an artist because it would be an utter bitch to scan.

Animating with voxels is also incredibly hard to achieve and have it look good.

To boot there are big issues with how to approach dynamic lighting in games. Sure you can quite easily bake the lighting information into the voxels but i imagine the more detailed your scene gets the larger memory load it's going to throw on your machine.

The main overarching points are:

Are people really going to appreciate the molecules of dirt being individually modelled?

How many people are you now going to need to manufacture a game of even 10 times detail?

EDIT:

Just noticed:

RPS - Unlimited detail VS Notch

Pretty much confirmed my reservations.

Link to comment
Share on other sites

EDIT:

Just noticed:

RPS - Unlimited detail VS Notch

Pretty much confirmed my reservations.

I think some people have more reservations about the way Dell has gone about promoting his search algorithm, than what it actually is underneath, Carmack was also asked about it:

Re Euclideon, no chance of a game on current gen systems, but maybe several years from now. Production issues will be challenging.

The lighting demo I posted a while ago was done with SVO (Sparse Voxel Octree), which is something along similar lines. Even Dell himself admitted that his tech would probably only be useful for background stuff at first. This portion of an article which includes a new interview with Dell mentions a reason why the games industry is reticent about adopting such technologies and mirrors what Carmack said in 2008:

We spoke to an Australian physics engine developer with experience of Bruce Dell and Euclideon. His company dealt with Bruce Dell years ago, when Euclideon was seeking funding for the Infinite Detail project. Said company declined to fund the project, citing issues with memory management, particularly when it came to animations.

According to him any live demonstrations given by Euclideon featured poor art and assets, so it was difficult to gauge precisely how hardware intensive Infinite Detail actually was.

The developer in question asked not to be named, but his primary concern wasn’t with the ‘Infinite Detail’ tech itself, which he claimed could work with adjustments – the issue was the toolset and the investments required to move an entire industry across to a new standard. Currently every game developer in the world is using tools dedicated to polygons – convincing an entire industry to toss years of investment and research would be a difficult task indeed, especially with an unproven technology.

Link to comment
Share on other sites

Currently every game developer in the world is using tools dedicated to polygons – convincing an entire industry to toss years of investment and research would be a difficult task indeed, especially with an unproven technology.

I disagree entirely. If this technology was workable, someone like Bethesda would be chewing their own arm off to be the first developer using it. Imagine being able to advertise a game as "100,000 times more detailed than Oblivion" and genuinely meaning it. You don't need to move every game developer or an entire industry - you just need to perfect the technology and initially sell it to one, competent devco. The industry will take care of the rest.

Link to comment
Share on other sites

Yeah, you can't "perfect the technology" first, then watch people adopt it. People need to adopt it before it can be perfected.

I'm sure polygons had a similarly tough time getting started, but back then there was an obvious advantage to using them over sprites, so they took off. Would voxels be the same, or would a prototype voxel-based game just look (to the untrained eye) worse than what we currently have with polygons and shaders? That's the problem.

Link to comment
Share on other sites

The problem is in the production side of things, all the tools and hardware exist to work with polygons and are constantly being worked on, while few are designed to work with Voxel-based approaches to things, the Carmack tweet basically says as much. Why would people throw away the current system, unless other competing ways of doing things have a similar support network of tools and expertise in place to replace it.

Assuming Carmack hasn't changed his mind on his pet project, we'll probably have to wait for id to pave the way, rather than expect somebody from outside the industry to be able to convince people it is worth the effort and investment.

Link to comment
Share on other sites

But all current graphics hardware is designed to be optimal for drawing polys, I think that's what that bit actually meant.

It's a shame that Intel canned Larrabee. What they were proposing was a completely programmable GPU that could turn its hand to Ray-Tracing, Polygons, Voxels, whatever.

Link to comment
Share on other sites

  • 2 weeks later...

So I was thinking about this when watching the BF3 co-op footage just now, seeing as that's running on PS3 hardware (I believe) it's quite an accomplishment that they managed to make it look half decent at all really.

Would it be a long shot to say that if you only draw the visual effects and all the polygons that are currently present on screen it could result in beefed up graphics? I'm sure that Crysis (2) does this to some extent by not rendering all of the detail behind a building or other objects?

If utilized to its fullest extent, wouldn't that have a much bigger impact on the way we render our visuals for games than the use of voxels (or 'atoms') as seen in Unlimited Detail?

Link to comment
Share on other sites

  • 2 weeks later...

Square Enix announced some details and a small demo of their inhouse Next Generation engine, the demo is getting close to photorealism (but it is only having to render a relatively easy environment and nothing else)

Luminous Studio is similar to a full engine like Unreal Engine or Unity. In other words, it has all the tools for the various game development processes, from asset editing on. The developers are working to make the engine high quality, easy to use, flexible, high speed, compact, and supporting both manual and automatic methods.

The engine is a Square Enix Japan project. However, Iwasaki, Iwata and Hashimoto did travel to IO Interactive in Denmark and Eidos Montreal in Canada to see those studios' technologies.

The engine is being developed separately from IO's Glacier 2 Engine and Crystal Dynamics' CDC Engine. However, the Japanese tech staff is is sharing source code with their overseas counterparts, and will be making use of the overseas technology. An example is a procedural approach animation demo that was shown at last year's CEDEC developers conference). This was mostly technology adopted from the overseas studios. You can see a video of the procedural approach animation demo in the Impress Watch article.

Luminous Studio will be compatible with any hardware that can handle programmable shaders. It will work with PS3, Xbox 360 and Windows PC, for instance. When asked about working with hardware like Wii and 3DS that don't have programmable shaders, Hashimoto said they're actively looking into making it compatible.

Regarding next generation hardware, Luminous Studio is being made Direct X 11 generation ready to a high degree. The graphics engine component has a Direct X 11 native mode.

The Luminous Studio team is placing a lot of emphasis on animation. They believe that things like high quality lighting, shading and modeling will be standard stuff in the coming generations, and having unnatural animation will stand out. The studio will include systems for procedural animation, full body IK, and so forth.

Another major focus is artificial intelligence.

http://game.watch.impress.co.jp/video/gmw/docs/473/138/html/render.wmv.html

ix9gft.jpg

xmvr6f.jpg

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. Use of this website is subject to our Privacy Policy, Terms of Use, and Guidelines.