Voxel based rendering has been around for years, it just isn't as efficient as rasterization, or as accurate as ray tracing. You can call it the future of computing, but its still a red headed stepchild.
what a fascinating idea. obviously it's not very sophisticated now, but when you apply physics to voxels, it completely changes the whole concept. moving objects by affecting the force of each particle that makes the object up and then calculating their affect on other particles as well as each other is the true power here
[citation][nom]Morgan3rd[/nom]Voxel based rendering has been around for years, it just isn't as efficient as rasterization, or as accurate as ray tracing. You can call it the future of computing, but its still a red headed stepchild.[/citation]
Pretty much. Wasn't there just an article a few months ago about Epic rendering an character through voxels, but admitting they couldn't get shading and animation properly done?
As said, Voxels aren't new, just doesn't afford the customisation and abilities that current rendering tech does. The future will mostly likely be a hybrid bastard child of Rasterization, Voxels, and Raytracing. Or something.
So this is not a new rendering concept.. As for the voxel calculations, i think they can become more sophisticated then the pixel or even polygon calculations.. However, the level of detail that they might generate is surely something worth looking for.. The current claims of hyper realistic detail levels and the ability to render unlimited graphics ( mot frames ) is not properly clear.. I mean a render engine draws the scene with details as specified.. Is this software meant to create detail levels on its own..!!
Call me crazy but didnt toms do a review on voxels or something similar??
I remember the conclusion more or less being that there is just no good and easy way to dynamically modify all the render tree thingys yet so its really only useful for static ground stuff.
Also, the unlimited thing is sort of misleading, I think. The video says "you obviously cant have an unlimited number of polygons" or something similar. True, thats why you only render the important triangles.
They only render 1 voxel per pixel and the rest are ignored. It's kind of like the hardware tessellation feature which makes more triangles when you get closer. I have a feeling this tessellation could eliminate the model swap problem the video makes a big deal out of on those old games.
I will be really impressed when some of the obvious shortcommings are fixed, like a video showing off crazy dynamic lighting, and dynamic destructible stuff.
After typing this, I see Kelavarus already mentioned what I was talking about
still, good on him. always good to see people trying new things. pushing tech that has just been brushed over and disregarded, who knows, maybe they missed something. sounds far-fetched atm, but so did many other breakthroughs.
also, its not that big a deal switching from polygons to voxels, as you say, voxels are just 3D pixels, all you'd have to do is "render" the poly mesh to voxels...
another thing. ok, so yeah, it wouldn't be unlimited, you could only have as many points as you could fit into memory, and you would need alot of memory, but keep in mind, its not the AMOUNT of memory at the moment that's the bottle neck. its memory BANDWIDTH, so even if you have to have 32GB of memory, if his search algorithm is really good, he would only really need enough memory bandwidth to pull one screen worth of pixels out every frame! and say you were doing it as cloud based gaming (like that thing where it renders on a server somewhere and then streams the result to any really underpowered device) if you were doing that, you could just have one level loaded into ram that could serve EVERY PLAYER IN THAT LEVEL. they could all search the same level in ram! limited use of memory bandwidth is more important that use of memory capacity at the moment.
It looks like DX11's tessellation, with 16-bit colours though. It looks more like an upgraded Nintedo rather than a high end screen, lol.
The ideal solution would be what Kevalarus suggested. A mix of all best things technology has to offer for an optimum result of image quality and speed.
Fascinating! Wouldn't this sort of turn the whole gaming industry upside down? Would this be all CPU coded or would it require a dedicated type of GPU (similar to say, physix, or EAX for audio)? More details would be awesome.
[citation][nom]omnimodis78[/nom]Fascinating! Wouldn't this sort of turn the whole gaming industry upside down? Would this be all CPU coded or would it require a dedicated type of GPU (similar to say, physix, or EAX for audio)? More details would be awesome.[/citation]
Yes, indeed it would. But this story has the typical signs of a pile of BS. The companies who get exposed as frauds or disappear after making big statements are also the same ones who are "tight lipped" about providing even the most basic information on how their stuff works.
essentially streaming the data needed for each frame from a database, sort of like video data only much more detailed. I wonder if it could be compressed using algorithms like H264. At least this is what I got out of the video. (Its not Voxels or polygons or point data.) I have thought about such a system before but I figured it would be some ways off before something like this would be feasible, well I look forward to seeing further developments in this and maybe some new uses for this tech too. Medical imaging would be nice if this could be connected to real time MRI data.
Interesting though first I want to see more animation than a camera moving through a static landscape, show me a moving actor like a person doing jumping jacks or even more impressive an explosion. All I see in this video is static objects. Easy enough to deal with, but I want to know how you coordinate unlimited points moving all at once in addition to the camera.
If I was the person who founded this breakthrough I would be tight lipped as well. There could be many possibilities why others who have "breakthroughs" disclose all information but basing this on how intense this breakthrough is, I would be weary on how much information I disclosed, especially if I didn't understand laws and politics. Maybe he wants to be sure it isn't stolen and he understands the ruthlessness of other companies. But regardless, I would have provided much better imagery. There should be full, hi-res screenshots. There should be full HD vidoes. This should be EXPANSIVELY showcased as if this is real, this IS going to flip the 3D world upside down. Does anyone truly understand the intensity of what is being claimed here???