• Unlimited Detail Technology (Ditching Polygons)
    74 replies, posted
I thought this was pretty neat [media]http://www.youtube.com/watch?v=Q-ATtrImCx4[/media] Watch till the end to see how it works
Well this totally took off didn't it.
Lettuce :D
I hope they keep building on this and make it more efficient. It looks promising.
[QUOTE=Dirf;25407102]I hope they keep building on this and make it more efficient. It looks promising.[/QUOTE] Yeah. Imagine fully destructible environments like the first Red Faction, only infinitely better
They should just hire some good artists to make some decent art to go with their tech demo, than they might actually wow people.
The problem with this is that it's just static, almost impossible to animate and even more impossible to get any physics with. And simple things as collision detection is very hard.
[QUOTE=MadCatMkII;25408922]Yeah. Imagine fully destructible environments like the first Red Faction, only infinitely better[/QUOTE] That's already been made. It's called Red Faction: Guerilla.
[QUOTE=Hesychasmos;25409213]That's already been made. It's called Red Faction: Guerilla.[/QUOTE] buildings aren't the only things in an environment
[QUOTE=Juniez;25409262]buildings aren't the only things in an environment[/QUOTE] This is true. If they made a game using this technology that was all about destroying a mountain, I would probably play it nonstop for weeks.
Worms 3D had that, but it was pretty lame.
[QUOTE=Muscar v2;25409185]The problem with this is that it's just static, almost impossible to animate and even more impossible to get any physics with. And simple things as collision detection is very hard.[/QUOTE] I suppose they could still use polygon based meshes for collisions, and only render the visual end of things with their system. I'm sure they could figure out some way to attach point cloud data to a physics enabled mesh for moving objects.
So basically this video is about Voxels?
[QUOTE=Edthefirst;25412908]So basically this video is about Voxels?[/QUOTE] He says in the video how it's different from Voxels.
[QUOTE=deathrat;25412955]He says in the video how it's different from Voxels.[/QUOTE] Yeah I posted that while I was still watching. It's still basically Voxels. They just have more efficient ways of using them.
This plus procedural generation? Actually, I really don't see how this would work together,but someone might find a way. Hopefully.
Seems like something at least worth looking into. Collision detection and physics might just counteract the huge boost in renderable points though.
The narrator in that video seemed so....... unexplainable. He seemed like he was about to go in depth at moments, but cut right out and explained something else or the same thing. The demo, interesting, but hire a better narrator next time.
I'm pretty sure the last time this video came up everyone figured out very quickly that it's essentially bullshit. Sure, raycasting to millions of small voxels in real time would be impressive if the voxels weren't all arranged in repeatable patterns... You can get a shit load more performance out of those huge scenes by cutting corners with the knowledge that all of the "geometry" is repeating. And then there's the fact that everything has to be static. Oh and also they're just barely getting workable framerates without any kind of effects or game logic... I don't think this tech is feasible, as much as I'd like it to be. At least not for a few years.
ah right i remember this. the guy's too much of a snot to realize this is way too far away and by the time we even reach it, something better will be out instead
i like how he's bagging out the polygon system and using crysis as an example but crysis still looks better than the thing he made obviously his has more detail, but crysis still looks more realistic
Rusty, he never tried to make it look realistic he just made it to show how much detail can be put into it. Also guys I have an idea, you listening? How about you don't ditch the polygons but make all the models invisible, so they would only be there for collision detection etc. and the "dots" would follow the invisible models?
Sounds like it'd be extremely tedious and annoying as hell to program. We don't only use polygons in games because they're cheaper to run than point-clouds. They're also way less annoying to work with programming wise. [editline]15th October 2010[/editline] Not to mention filesizes would be ungodly large. Also translating specific sections of polygons + calculating any sort of physics would entirely destroy that simulation. You can also see different points disappearing in and out during that video, meaning their visibility algorithm isn't that great. And say bye-bye to smooth dynamic shadows.
In this thread we pretend whe actually know stuff about real-time computer graphics while judge the entire potential of a new technlology by comparing it to one that has had 40 years to develop. [editline]15th October 2010[/editline] His voice is kinda annoying, though.
He uses strange colours for his demo's, but it looks very promising.
gimmick
[QUOTE=Muscar v2;25409185]The problem with this is that it's just static, almost impossible to animate and even more impossible to get any physics with. And simple things as collision detection is very hard.[/QUOTE] [media]http://www.youtube.com/watch?v=cF8A4bsfKH8[/media] Here's an example of a flying bird. I don't know how hard they tried to get this done. Media tags don't leave me alone.
[QUOTE=Rusty100;25414528]i like how he's bagging out the polygon system and using crysis as an example but crysis still looks better than the thing he made obviously his has more detail, but crysis still looks more realistic[/QUOTE] his demo is without any fancy shaders and his art is [I]very[/I] ugly if they hired some good artist to get a better presentation it would be much better just listen to that Beethoven shit in the background that sounds like ringtone
I would actually like to see what a professional artist can come up with using the point cloud system, if they create something that makes me go "HOLY SHIT AWESOME!" then consider me on board for the changeover. ofcourse the big factor when coming to games is you can make a game look as good as you want, but your still working to a time limit so this system is no good if it takes twice as long to create something that would take 10 minutes to whip up in polygons.
[QUOTE=Un.Hxx.Aé;25415710]In this thread we pretend whe actually know stuff about real-time computer graphics while judge the entire potential of a new technlology by comparing it to one that has had 40 years to develop. [editline]15th October 2010[/editline] His voice is kinda annoying, though.[/QUOTE] Except this isn't an entirely new revolutionary technology, and has been around for at least 12 years. Not to mention the memory foot print of one billion points ( as said in the video ). Assuming in the LOWEST possible form ( a struct with 3 bytes ( x, y, z ) ( which wouldn't even work well in itself, assuming they are unsigned they'd only have a range of -128, 127 ) ) So, for each defined point in 3D space there is AT LEAST 3 bytes, 3 * 1 billion = 3000000000 bytes = 3000 megabytes = 3 gigabytes. Now lets load all the textures and other relevant game data, and you can say goodbye to your RAM. Where with polygons you have 3 points for a triangle, no matter the size. One triangle ( of any size ) only comes out to 9 bytes in the simplest of terms. A triangle strip with 1 billion triangles ( of any size or rotation ) comes out to approx. 1.5 gigabytes because of shared vertices ( which a point cloud kind of lacks ). You know, we've had several decades to work on polygons and how our hardware works with them, damn straight they're going to better. ( point clouds can suck my dick )
Sorry, you need to Log In to post a reply to this thread.