Thread: Unlimited Detail [GFX Engine]

Heya :)

Just found this and though it's all slightly unbelievable, if true this is absolutely flabbergasting.

See for yourself (article and site is in german, but the videos are in english):

- Summary of this technology (video)
- Detailed Explanation Part 1 (video)
- Detailed Explanation Part 2 (video)

- Article (german)

- Company homepage (english)

Now while the examples of classic polygon-based gfx are usually very choppy and with low resolution/anti-aliasing, the comparison of working with a point-cloud as opose to an object described by a finite number of polygons remains.

DX11 introduces tesselation, a way of automaticaly increasing the complexity of an object by adding more polygons while paying attention to the power of the host-system. They don't seem to mention this, however the inherint limitations of this tesselation technology remain. It's just another way of arbitrarily boosting the polygon count.

I can also see how nVidia, ATi or Intel would opose this technology and do their best to supress it, seeing as it could cost them billions of dollars of profit, if it made it to the consumer market.

Obviously, one must question in how far the Unlimited-Detail clips are pre-rendered, however I find it unlikely that they'd take as long as say ray-tracing (playing a pre-rendered movie in fast-forward) when promoting a product meant to be used for real-time rending.

Another issue they don't go into far enough in my opinion is the actual algorithm. They mention using search engines and compare it to google/text-based internet search, however these search engines merely precompile a long list of keywords and references (indexing) in order to improve the search-times. I'm not sure how this relates to unlimited point-clouds, I find it difficult to imagine them indexing these in any cost-efficient (space-saving) manner, however who knows...



Interesting huh? This is certainly something to keep watch of :) Let me know what you think!
~ Exa

PS. This does NOT belong in gaming ;) This article describes graphical technology and is not limited to the gaming spectrum.

yo mate... WTB translation to english :PPPPPP

nah just kidding

just watched the summary video, all i have to say is ... OMFG games are gonna be fucking awesome in the years to come (and computer build ups fucking expensive i guess)

Originally Posted by Ryuzaki

yo mate... WTB translation to english :PPPPPP

nah just kidding

just watched the summary video, all i have to say is ... OMFG games are gonna be fucking awesome in the years to come (and computer build ups fucking expensive i guess)

Less so than they are now apparently.

What they're saying is that the polygon/shader-units arms-race is over. They're claiming to have discovered an alternative that limits itself to the amount of pixels on the screen while mainting a much higher degree of detail at a lower performance cost.

I don't really understand it myself, but if it's true then not only are games gonna look absolutely awesome and way more realistic (polygons are inheritly limited in their capability to model reality), but they're also gonna require alot less processing power to achieve the same results when compared with today's technologies...

Think using a brute force algorithm to crack a password. Intel/AMD start producing processors with more and more Mhz (that cost more and more ;) and then somebody comes along with a dictionary attack that can achieve the same result with a regular old 386 :P And yes, I realize that this is over-simplifying to the extreme xD

Originally Posted by Exa

Less so than they are now apparently.

What they're saying is that the polygon/shader-units arms-race is over. They're claiming to have discovered an alternative that limits itself to the amount of pixels on the screen while mainting a much higher degree of detail at a lower performance cost.

I don't really understand it myself, but if it's true then not only are games gonna look absolutely awesome and way more realistic (polygons are inheritly limited in their capability to model reality), but they're also gonna require alot less processing power to achieve the same results when compared with today's technologies...

Think using a brute force algorithm to crack a password. Intel/AMD start producing processors with more and more Mhz (that cost more and more ;) and then somebody comes along with a dictionary attack that can achieve the same result with a regular old 386 :P And yes, I realize that this is over-simplifying to the extreme xD

ye but towards the ending of the summary video there was a graph showing that the new technology's disadvantages would be 1) the time needed to design something and 2)that the processing of it would be slower or something. So we might actually need some uber hardcore top notch hi-tech system if we want to be the FIRST to experience the new technology :p

Originally Posted by Ryuzaki

ye but towards the ending of the summary video there was a graph showing that the new technology's disadvantages would be 1) the time needed to design something and 2)that the processing of it would be slower or something. So we might actually need some uber hardcore top notch hi-tech system if we want to be the FIRST to experience the new technology :p

1) In the detailed vids they explain that you can either design directly with the voxel editor, or you can import high-def models, such as are used in 3D rendered movies (Shrek etc.).

One big advantage they mention, is that you don't need to design three-eight different models of an object, depending on the distance/detail you have to that model. Also you don't get model-swapping. Again this would save time in the development process. Current games all emply model-swapping, this technology wouldnt.

2) I think what they're saying is more along the lines of it's slower than polygons (due to the algorithm) yet alot faster than ray-tracing. If you were to produce the objects as they look with polygons, it probably wouldn't be much slower...

Either way, I don't mind the polygon-arms-race. Sure it's annoying that gfx-cards are "outdated" within a quarter of a year, however I find it even more annoying that there hasn't been a really major breakthrough in the past 15 years since DX9 started. Shader 2.X, 3.0 and maybe tesselation are all improvements on a relatively primitive technology (polygons) when compared with these voxel cloud things ^^ Besides, I recon these could scale incredibly well...

ah i see, ye you got a point there. This thing seems exciting, but as everything new, until it gets tested out we'll face some problems i guess. I really believe it will be expensive though, even if it doesn't need to be, just for the sake of "hey you consumers, we're giving you a whole new technology... put your house in a mortgage to pay for a new pc, MWAHAHAH" :P

i think you get my point