pouët.net

Unlimited Detail Technology

category: offtopic [glöplog]

"Unlimited Detail files all its data in a strange format" - Bruce Dell, YOU are STRANGE !!

added on the 2010-04-12 18:37:17 by lsl lsl
cheers zj, that was interesting. Poor guy. He has a good resume if he's looking for a job, at least. Probably can't float a company on this portfolio though.
added on the 2010-04-12 18:50:54 by Claw Claw
yeah it's all very interesting - I think he's got funding now - so we should see something very soon (hence the New Scientist article and a few others there was also mention of a .exe soon). I'm sure something will come out of this might, be a while tho - the last vid said 16months (which I guess is them working out how to move stuff). The whole industry is seems to be heading in this direction - larrabee, the renewed interest in voxel's and SVO's e.t.c..

added on the 2010-04-12 18:55:46 by ZJ ZJ
I thought about "Molecular Animation" Quite a while ago. Basically it's the same kinda thing and idea. An object could be animated using real world data, say a rubber ball that has molecular data the same as a real world rubber ball. So instead of a physics engine, it just acts as it does in the real world. Considering processing power can be crammed into smaller spaces, it would be easy to animate, just using raw CPU power. Intel are pushing 48 cores to special universities and they will soon have over 100 cores in a single die, so we're not far off from the power needed. And as that guy was saying, it would remove polygons altogether.

8 bit - it's interesting that u mention this - because one of the limitations of the tree data structure is it acts like the real world - i.e. u can't have 2 objects in the same place at the same time. If u could create the tree in realtime - u essentially would have a very simple physics engine also - which in fact mirrors the real world - if u think about a object stored in a sphere tree - the nodes are almost a real version of a interia tensor - and so physics coding would be a joy!
added on the 2010-04-12 19:04:52 by ZJ ZJ
That's what I was thinking when I initially came up with the idea.

Just for arguments sake, a first person shooter. Shoot someone in the arm and depending how they take the shot depends on the severity of the injury and how it affects them physically. But it would be as simple as real world objects and physics. If someone as a test could animate a rubber ball using only it's physical properties and not a physics engine, would be a genius! I'd imagine it would cut down on a LOT of physics code.
sweet! it all boils down to the dynamic tree creation :) I'll investigate it after I've got my rendering working (currently got a mess of triangles on the screen :P )
added on the 2010-04-12 19:17:26 by ZJ ZJ
Quote:
But it would be as simple as real world objects and physics

?
added on the 2010-04-12 19:19:02 by stijn stijn
guys, you should found Pouët Inc.!
added on the 2010-04-12 19:20:52 by aftu aftu
Quote:
Quote:
But it would be as simple as real world objects and physics

?

...but he's right! Just make a GOD-engine and you can create everything in it. And it'll be really easy, too. No need for some complicated programming with logical constructs or whatnot. Just a tiny bit of faith.

Man, I'm _so_ patenting that stuff right now!
added on the 2010-04-12 19:31:58 by ted ted
hehe - all jesting aside one of the interesting aspects of the tree structure is it can indeed be infinite (it's how I stumbled into this mess in the first place - trying to render serpinski's/menger cube). Because the nodes reference themselves and because of using the midpoint algo (which is a self normalizing error term) it has infinite precision! (actually it's limited by the first ray cast - but meh to details :P ). I'm very interested where all this will lead (the madhouse?)
added on the 2010-04-12 19:39:51 by ZJ ZJ
QUANTUM FOR TEH WIN!!!
added on the 2010-04-12 19:41:10 by Optimus Optimus
http://www.guru3d.com/news/pointillist-style-rendering/

Someone should tell them that demos have had voxels for some time already. I'd also like to know if Agenda Circling Forth actually used any voxelly stuff.
added on the 2010-04-12 20:51:47 by msqrt msqrt
great demo! here's the tech on it:

http://directtovideo.wordpress.com/2009/10/06/a-thoroughly-modern-particle-system/
added on the 2010-04-12 21:10:49 by ZJ ZJ
whoops wrong link :)

http://www.geeks3d.com/20091028/demoscene-blunderbuss-direct3d-demo-with-one-million-particles/
added on the 2010-04-12 21:17:29 by ZJ ZJ
Thanks, seems like an excellent read.
added on the 2010-04-12 21:33:56 by msqrt msqrt
it is a great read - the light propagation system is awesome - would work in the tree system too :) think smash posted on this thread on a previous page
added on the 2010-04-12 21:39:04 by ZJ ZJ
Soon we'll have more processing power than we could possibly need. Vector rendering won't be able to keep up with processing power, so I think it's high time a new rendering technique is created to make use of this abundance of power!
I reckon the rendering will be ok, it's the artists that won't be able to keep up. We need a new method of making artists!
added on the 2010-04-13 00:01:06 by psonice psonice
And off-course, the most important thing, will this run on a browser?
added on the 2010-04-13 00:07:11 by JaK JaK
Quote:
Could some one please explain to me what memory cache is and why I have never encountered it in C programming.

[ Bruce Dell ]
[ 05-Apr-2008, 12:53 ]
[ http://forum.beyond3d.com/showthread.php?t=47405 ]


:)
however, that does not mean he hasnt developped interesting things on his own.
pwned :)
added on the 2010-04-13 06:33:59 by Optimus Optimus
Are 'cache' and 'memory cache' the same thing? I've had the intention of learning more details about the processor for quite a while but never actually got to it.
added on the 2010-04-13 09:24:29 by msqrt msqrt

login