EPTIn their example they stated their level consists of several billion of those voxels, they need to be saved and rendered somewhere. Obviously oclussion culling will make sure you only render what you see, but your oclussion culling process would take far longer with that many points.
The future of graphics in games - Page 3
| Forum Index > General Forum |
|
Slaytilost
Netherlands968 Posts
In their example they stated their level consists of several billion of those voxels, they need to be saved and rendered somewhere. Obviously oclussion culling will make sure you only render what you see, but your oclussion culling process would take far longer with that many points. | ||
|
Simberto
Germany11452 Posts
On August 02 2011 01:42 Maxd11 wrote: This video is obviously made for people that don't understand as much about computers as the average TL poster seems to. That said it would still be awesome if this can actually work the way the say it can. It makes you wonder though, how far could the "real" graphics go? Could it get to the point where they are shooting at unsuspecting actors inside of scanners to get a realistic death scene? Doesn't really seem probable. They have avoided doing that in movies for quite some time. | ||
|
LuciferSC
Canada535 Posts
Taking it straight from the video, they are claiming that u don't need a supercomputer to run that level of graphic. Seems like a different engine, or even a different algorithm of doing graphics. Can't wait to see what this really can do. | ||
|
LegendaryZ
United States1583 Posts
I could be wrong, but the entire direction of this seems contrary to where the gaming industry is looking to go in terms of graphics which seems to be finding ways to procedurally generate content rather than mapping out entire worlds in this kind of detail. | ||
|
fezvez
France3021 Posts
| ||
|
GizmoPT
Portugal3040 Posts
http://www.youtube.com/user/AtomontageEngine :O im getting really confused ;< | ||
|
Mykill
Canada3402 Posts
| ||
|
Ryhn
United States509 Posts
If they've found a way to reduce their problem to a form that search algorithms do the heavy lifting. . . that's huge. Put in laymens terms, some of the most efficient search algorithms have properties along the lines of, "Oh you need me to look through a billion possibilities? Let me get you that answer you wanted in about 10 steps" | ||
|
DeLoAdEr
Japan527 Posts
On August 02 2011 01:47 Slaytilost wrote: Its kinda silly stating something is unlimited in the IT industry. As stated above, its fake, it just uses voxels. Its never unlimited. In their example they stated their level consists of several billion of those voxels, they need to be saved and rendered somewhere. Obviously oclussion culling will make sure you only render what you see, but your oclussion culling process would take far longer with that many points. from what i've understood this is not voxel rendering, but point-cloud rendering. I agree though that this whole unlimited thing is very silly. I suppose (assuming it's not fake) they've come up with good hierarchization techniques to make the culling and rendering process fast as well as some good compression and streaming methods to handle this insane amount of point-data. | ||
|
Blyadischa
419 Posts
Does anyone think that their "atom"-based graphics look like graphics from games from like 7 years ago? | ||
|
Laurence
Ireland119 Posts
| ||
|
arbitrageur
Australia1202 Posts
"The word you're looking for is Voxels, and its a technology that's as old as polygons. You've got nothing new here. You're passing off basic fundamental well-known 3d rendering techniques as something new and fascinating by making up your own words for decades old technology and fluffing it up with... fluffy meaningless phrases like "UNLIMITED POWAR!!!". Oh well, have fun fleecing your investors then disappearing with the money as all modern snake oil salesman do. NinjaSeg 2 months ago 20 " | ||
|
dudeman001
United States2412 Posts
| ||
|
aka_star
United Kingdom1546 Posts
| ||
|
kirdie
Germany221 Posts
| ||
|
wollhandkrabbe
Germany97 Posts
On August 02 2011 01:19 DeLoAdEr wrote: interesting results for a software renderer, but the video only shows static geometry. I wonder if they're even able to handle animated objects.. and they mention the scene consists of more than 21 trillion points, lets say they store 4 byte for each point (color information, position, etc.) this would be ~100 TB of data right? so there must be some good compression going on there. ;p The objects are all referenced. All the trees are the same for example, so you need the tree geometry once and then just one ancor point for every other tree. That limits the amount of information. But I dont quite see how they can render the scene though, as that actually would require all geometry, shader etc. information to be taken into account. That amount of information just can't be handled by current current computers. Not to mention collision detection, physics, adaptive lighting and so on. | ||
|
exog
Norway279 Posts
Will be interesting to see how it goes. | ||
|
Ziktomini
United Kingdom377 Posts
| ||
|
EternalSC
Sweden313 Posts
| ||
|
Keitzer
United States2509 Posts
On August 02 2011 00:54 WniO wrote: the problem with this is they cant render it in real time, or animate for that matter. wasn't that the whole point of the video? was saying that they CAN render it real time? also... HOLY SHIT THIS IS AWESOME!!!!!!!! >mfw + Show Spoiler + ![[image loading]](http://i.imgur.com/fJavy.jpg) | ||
| ||
