Realtime Raytracing

[HackerX]

To view links or images in this forum your post count must be 2 or greater. You currently have 0 posts.

"Real-time raytracing has often been called the pinnacle of computer rendering for games but only recently has it been getting traction in the field. A German student, and now Intel employee, has been working on raytraced versions of the Quake 3 and Quake 4 game engines for years and is now using the power of Intel's development teams to push the technology further. With antialiasing implemented and anisotropic filtering close behind, they speculate that within two years the hardware will exist on the desktop to make 'game quality' raytracing graphics a reality."

As somebody who's played with a fair bit of graphics programming over the years I think it's about time somebody brought this to the forefront.

Assuming they managed to pull this off, it means far more complex (far more perfect) 3d scenes, and about 10 years or so of graphics hacks down the tube (not such a bad thing). It also probably means the end of the game graphics engine competition, since raytracing basically levels the field (since currently, engines are defined by how well they can fake "realness") and hopefully means games/graphics will be defined by the content chucked at the raytracers.

I see the future where graphics cards are hardware raytracers (as opposed to hardware rasturisers) where performance (and hence price) is measured in the number of parellel cores that the card is able to utilise (since raytracing the perfect candidate for parellel processing)

Negative of all this? The same one the exists currently, speed is directly proportional to the processing power you can throw at it. So naturally, the time spent now hacking around the graphics engines today to add in features, will be spent trying to create tricks to speed up the processing. Perhaps that's where the engine competition will head?

Thoughts?

[qwerty]

As much as I'd like to see games graphics continue to get better and better, I'd much prefer to see games engines work more towards natural motion (To view links or images in this forum your post count must be 2 or greater. You currently have 0 posts. warning seriously cool).

Back on the graphics side of things: An insight to the direction of 3d programming would be to compare the current demo scene to the current games industry. Every year I am impressed more and more at demo competitions and I still am in awe of Debris. Yet every year the games industry throws out the same old crap packaged with a different name.
Ray tracing should be so much more than what it will come out as because the games developers will not harness the technology completely (setting John Carmack aside on this - the dude would do it in a heart beat).

[HackerX]

As much as I'd like to see games graphics continue to get better and better, I'd much prefer to see games engines work more towards natural motion (To view links or images in this forum your post count must be 2 or greater. You currently have 0 posts. warning seriously cool).


One of the advantages of using raytracing is that it allows you to define your scenes fair more easily as mathematical equations. This definately leads an advantage to raytracing. You'll find one of the biggest issues currently with the above is the problems that occur when trying to define how polygon distortion (and in turn, fluid motion) should work "realisticly". If you can define how the shape exists, and how it moves mathematically, then you can probably raytrace it easily


Back on the graphics side of things: An insight to the direction of 3d programming would be to compare the current demo scene to the current games industry. Every year I am impressed more and more at demo competitions and I still am in awe of Debris. Yet every year the games industry throws out the same old crap packaged with a different name.
Ray tracing should be so much more than what it will come out as because the games developers will not harness the technology completely (setting John Carmack aside on this - the dude would do it in a heart beat).

I don't keep up with the demo scene, but I can probably explain this. The problem with doing fancy stuff is that, as soon as you try to do more and more of it in a single environment (i.e. a game level) things start falling apart quickly. Especially when the effect needs to be generic enough for multiple scenerios (different lighting, different backing textures etc) around the level. It adds a huge complexity to the work involved. It ends up being maximum effort for minimum results, hence why a lot of it doesn't make it into the games. The difference is that with raytracing, all those effects are possible no matter what the suitation. Need lighting and shadows, do a pass with the lighting rays. Need reflections, do a pass with reflective rays. No matter what the ray hits, it will always respond in the "correct" way (assuming it's programmed to do so). Currently it involves programming for many complex scenarios and trying to come up with something that is best fit.

[qwerty]

As much as I'd like to see games graphics continue to get better and better, I'd much prefer to see games engines work more towards natural motion (To view links or images in this forum your post count must be 2 or greater. You currently have 0 posts. warning seriously cool).


One of the advantages of using raytracing is that it allows you to define your scenes fair more easily as mathematical equations. This definately leads an advantage to raytracing. You'll find one of the biggest issues currently with the above is the problems that occur when trying to define how polygon distortion (and in turn, fluid motion) should work "realisticly". If you can define how the shape exists, and how it moves mathematically, then you can probably raytrace it easily

Yeah well it comes down to how you program your scenes. Myself I've always worked in the vector world for objects (not that far accomplished in 3D work) and then apply raster algorithms to plot the scene for speed on display. Digitial motion shouldn't be a by-product of Ray tracing, whereas they will go hand in hand they are not dependent on each other.

Back again to the Demo Scene whereby everything is procedural from the objects to the textures, they focus on much the same problem as far as I know of converting the procedural vectors into raster outputs.

[Ijz]

Realtime raytracing is definitely the future. Not just because it looks better but in my opinion (as a developer) because of the huge paralel processing benefits. Intel is already experimenting with 80 core processors, I wouldn't mind having one of those in my PC :)

And what about AI development, this is still in its infancy.I'm really curious when this will have a real breakthrough. Will it be some new kind of neural network, some ultimate pattern matching algorithm, raw processing power or the sum of the parts? What do you guys think?