This looks interesting: Unlimited Detail Technology

Though it sounds a lot like raytracing, as that also works by computing just the amount of pixels in your screenresolution, rather than computing each forward-facing polygon's image. What's missing in the video, though, is collisions between objects - no word on how much extra computations those would throw up if you have 2 point-cloud objects running into eachother

i saw the demo on this, it seems great for static visuals, 1000s of models that are not moving, but none of their demos showed actual movement in the scenes, only your POV moving around within the world

They did mention Ray Tracing once in the video, kind of glossed over it. Not sure what exactly the video is supposed to be showing. I did not get the impression that this was not a professional endeavor. Nothing wrong with a non professional project, I was just not clear on his goals or any real use application. The future will have different graphics then we have currently. There I just summed it up

The concept sounds the same as Voxels, except more of a 3D search routine. That is just guessing as it really did not elaborate. This guys wants graphics to be rendered on a pixel by pixel basis I think.

Perhaps it's as simple as someone using CUDA or DirectCompute to build a '3D accelerated MySQL' type of system to sort a database of points really fast?

^^ LMAO,
They lost me after 'this looks interesting'.

In games, each object is built out of little triangles - the smaller the triangle, the more detail the object has. But as triangles get smaller, more of them are needed and each new triangle increases the computing power required to run the game smoothly.

The dude in the link claims he can replace the triangles with points, and no extra computing power needed, so that should give unlimited detail as you can't go smaller than a point.

right, but he's saying that only the pixels required for the current point of view are rendered, unlike with common engines where all the polygons aka objects are rendered whether you are looking at them or not

I would love to use a 3dmax clone with this technology. Now that would be really cool!.

Ah yeah, interesting thought If it's real, then the video likely shows only a fraction of what's possible with it -- realtime rendered 3D modelling on desktop computers sounds very nice indeed And it would also enable the virtual reality side of things, though the high resolutions needed, will limit that initially.

Ignoring the issues of object collision etc I think, assuming it's real, that the obstacle they will come up against is cost/benifit ratios by the leading companies. Even if you use unlimited detail the human eye/brain can only process so much as a time so will never take it all in especially when driving fast through a level or while running around trying to attack someone. Whereas the cost of the new development to make it a workable solution and the new training for graphics developer I think will outweight the benifits.

Also is it just me or does it seem strange that he says (in the second part I think) that it looks a bit dodogy at times as it was developed by programmers not graphics people. If it's a small personal project fine, but if you planning on making it a big business launch to major companies don't you think you'd hire at least one graphics expert to aid with the model development.

Hmm,i was thinking about that. But some bugs were programming. I guess the graphician will be called in at the last stage. as far as i understood they also ment that right now the graphics were done only with coding not by 3d modelling through a program like max.Even then they are very good. I guess it could be a milestone!

I think that might be what they meant to. My point was more if your going to do a press release sorta thing like that surely you want to impress from the get go, not have a so-so representation that can be improved on buy a graphics expert. If your going to do a press release to get people intrested normally you would hire a expert to design and do the graphics for the model. You would also usually fix things like the shading in the second video, even though they do say that is done by different software and dosn't have anything to do with the main product, you would think they would either not include shading if they can't get it to work properly or wait till they fix it. All just seems a bit ameaturish to me.

Sorry for resurrecting an old thread but I don't visit the games forum very often and as I read through it looks like a hard floppy topic.

It sounds great on paper and from a British guy, however when you consider the complete process there is still one HUGE factor in eliminating jagged edges and straight lines on round objects; the screen which it is displayed on. No matter what kind of computing process or power, if you're trying to "draw" it using straight edges (essentially how pixels are arranged on a screen) you will never get rid of the lines completely. It is true that screen resolutions have been slowly increasing for computers at least but this compounds the problem of requiring more computing power to render all those points/pixels.

Some companies only have so much for R&D and cannot spend what could be billions on a new [better on paper] technology which might require a huge push onto developers to learn something completely new. If every computer gamer had 4 5970's in crossfire and a 6-core i7 overclocked to 4.5ghz then the polygonal dilemma wouldn't exist as programmers could utilize 100% of the hardware's potential. I personally do not see any scalability in point cloud graphics, and looks like it would require special hardware to run well enough to be on par with current polygonal games.

I would also like to note that some games are only eye candy (cough crysis) and not enough fun value; some companies get so tied up in making things look pretty that the game itself isn't any fun. Yes eye candy is good but it only goes so far. I propose the analogy of Megan Fox: sure she's pretty but has the acting talent of an ordinary house cat.

They're back with an update: Euclideon Island Demo 2011 -- doesn't look much different to me, though, just a few new objects

Warning: that dude's voice is a mix between a commercial radio DJ and a rookie tv news reporter, so you might get physically ill from hearing it

I'll be damned if that isn't impressive to anyone that watches it. It probably means my graphics card will be rubbish in a few months though...

sounds good but the claim that my games will get better by "about one hundred thousand times" did make me chuckle, the graphics will increase but IMO a badly thought out game is bad regardless of the graphics

Notch, the developer behind Minecraft, is calling them out on their claims: It's a scam! : The Word of Notch.

Though I have to say that procedural generation may still cut down on the actual need for huge amounts of data -- I recently saw a nice YouTube video about the Fuel game for xbox360, which has the biggest game environment ever and also seems to have more detail than could actually fit on a DVD thanks to the procedural engine behind the game.

you my ape friend are what is known as a tease

Aha, I managed to lure you away from the pack

Dear god! There's a flock of evil sheep now?

Euclideon address the haters :P

Cheerz, aZen! It's nice to see them trying to counter the criticism, though they're obviously trying to shield their work as well.

A few months (years?) ago, I came across an imaging technique where you can take just a small bunch of random pixels from a picture and if you then fill in the blank space with gradients between the pixels, the end result is remarkably accurate compared to the original picture. I wonder if that could be one of the techniques that they're using (in 3D) to cut down on the pointcloud that they need to search through