Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon. Entire thread

Voxels?

Name: Name 2010-11-21 8:21

I was browsing the superinformation highway, and found something about voxels, (specifically, the voxlap library, written in C). I've learned that voxels lost the war against polygons, and now we all are stuck to them, and to silly clipping problems

But voxels are a pretty cool guy, they can represent real objects and even mechanisms with various resolutions using voxels as 'atoms', and they are not afraid of systems without 3d acceleration.

So, /prog/, tell me, in your opinion, why are voxels so ignored, and what could be the practical application that will make them a 'mainstream' technology?

In case you dont know shit about what i'm talking about, watch these

http://www.youtube.com/watch?v=TjmRPjnWJ5g
http://www.youtube.com/watch?v=Zl9CiGJiZuc

Name: Anonymous 2010-11-21 8:29

It's the same deal as with vector and bitmap graphics. Voxels take up a lot more space than polygons.
3D acceleration is useful in a complex scene (i.e. any scene) as 3D->2D projection is still required to draw anything, but seeing as most graphics cards work with polygons, each voxel is actually a cube and you may as well use polygons then anyway.

Name: Anonymous 2010-11-21 8:54

They're not really 'ignored'.  Look at idTech 6, which will probably use SVOs.  Everyone in the field are aware of the technology.
  At the level of detail we use today, polygons are simply faster and less memory intensive with better visual appearance, which is why they dominate the field.  This will become less true as more detail is added and polygons start approaching the size of a pixel, but current commodity hardware doesn't scale to that point anyway.

Name: Anonymous 2010-11-21 9:02

I'M JUST GOING TO SAY WORMS 4: MAYHEM LANDSCAPE DEFORMATION WITH CAPS LOCK.

Name: Anonymous 2010-11-21 11:30

>>3
less true

Name: Anonymous 2010-11-21 11:34

>>4
Even that kind of thing doesn't require voxels, it's just easier to implement that way.

Name: Anonymous 2010-11-21 17:56

>>5
You don't understand fuzzy logic.

Name: Anonymous 2010-11-21 22:46

>>7
Neither do you, maybe.

Name: Anonymous 2010-11-22 1:07

>>8
Note that fuzzy logic and probabilistic logic are orthogonal concepts – you can have systems with fuzzy logic but certain values, crisp logic with uncertain values, or both fuzziness and uncertainty.

Name: Anonymous 2010-11-22 1:48

>why are voxels so ignored
storing high resolution voxels requires too much space

>what could be the practical application
algorithmic generation of volumetric data (quaternion functions)

>make them a 'mainstream' technology
once the number of pixels <= number of ALUs we are talking

Name: Anonymous 2010-11-22 6:05

>why are voxels so ignored
>what could be the practical application
>make them a 'mainstream' technology
gb2imageboards, >>10

Name: Anonymous 2010-11-22 12:21




Name: Anonymous 2010-11-22 12:59

>>9
You may be fuzzy and uncertain, probably.

Name: Anonymous 2010-11-22 14:42

>>11
[quote]=====D bbcode epenis[/quote]

here have fun <°>>>-<<

Name: Anonymous 2010-11-22 15:28

>>13
Sometimes the boundary between fuzziness and uncertainty may be fuzzy and uncertain.

Name: Anonymous 2010-11-23 3:54

I forgot wallet at home and now my stomach is growling while my fed coworkers sit around /prog/ing happily. AMA

Name: Anonymous 2010-11-23 5:17

Next id software will feature polygon-based progressive detalization that will be able to display as much detail as voxels, but without the horrible computation expenses immanent to voxels.  Have you tried playing Voxenstein 3D?

Name: Anonymous 2010-11-23 18:51

>>17
The problem is that every polygon engine has to be converted triangle based models which doubles the computation effort.
In that fashion voxels are still superior.

And until now progressive detalization always required the storage of additional polygon models.

Name: Anonymous 2011-02-04 16:50

Name: Anonymous 2011-02-04 18:45

Name: Anonymous 2011-02-18 13:34

that's cool and all, but check my doubles over there

Name: Anonymous 2011-07-01 12:45

>>22
nice dubz, fucking shitfucker

Name: Anonymous 2011-07-01 12:57

voxels don't scale when you zoom.

raster in general is inferior to vector. Polygons aren't perfect, though. What we need is hardware accelerated spline drawing.

Name: Anonymous 2011-07-01 12:59

UNLIMITED POINT CLOUD DATA

Name: Anonymous 2011-07-01 13:06


15 March 2011

Dear Supporters,

We’ve been more secretive then a bunch of ninjas lately and subsequently we have received many emails asking for updates and in some cases confirmation of life.  Please know that all is well and work is progressing very nicely.  We’ve been hiring a few new people to work on tools and polygon conversion and work here progresses smoothly and according to schedule.  Polygon to voxel conversion is still young but is working as expected.

(A simple polygon object, converted to Unlimited Detail)

At completion we will be able to import objects straight from CGI movies directly in to games, which will be quite an improvement over the current system of having to rebuild them all by hand.

Polygon conversion is important to our customers as most artists have been trained in such programs as 3DS Max and Maya and we feel it will be best for everyone if we can make our technology fully compatible with what exists so that no retraining is required.  No retraining does not mean that there won’t be big changes. Artists of 3D graphics will no longer have a “Polygon Budget” they are completely free to make whatever they want with as much detail as possible. 

 We are often asked if the work is encountering any sort of problems. Whilst the engine is working fine and tools are progressing very well, we do have a small dilemma in that we wonder where we can find any content that is worthy of the technologies abilities. What I mean is : where do we get polygon trees that have as much detail as the real world ?, were do we acquire realistic polygon objects ? People have not been this free before so there was not any reason to create such objects with massive levels of detail. If anyone can assist with directing us to some immensely high detailed 3d models we would be most grateful.

As for demos we still can’t give any release date but we are working as hard as we can and we hope we will soon have something that we can humbly submit to the approval of the public and hope that our efforts have been well pleasing to you all.

Once again thank you all for your support.

      Kindest Regards

               Bruce Robert Dell

Name: Anonymous 2011-07-01 13:14

>>26

Did he just say he can't find models with enough detail? There are thousands of free CAD-quality models on the internet, and you can always have a CAD engineer draw some for you (they're working on a commercial project after all). And did he sign some forum post with his second name like he is some sort of nobility?

From the website:

Polygon to voxel conversion is still young but is working as expected.

What the shit? I know undergrads who could write a polygon-to-sparse-voxel-octree converter in a weekend.

Name: Anonymous 2011-07-01 13:30

>he sign some forum post
Thats from his new web sight. http://www.euclideon.com/

Name: Anonymous 2011-07-01 13:31

Unlimited bullshit

"Not detailed enough models" just means this retard finally discovered that voxels are not about computing time but memory space.

Name: Anonymous 2011-07-01 13:31

>>27
What the shit? I know undergrads who could write a polygon-to-sparse-voxel-octree converter in a weekend.
Bullshit. CS graduates around here can barely write guess-the-number in Java.

Name: Anonymous 2011-07-01 13:43

>>30

Enjoy your community college. At my university, freshmen write a raytracer and several glsl shaders in the graphics course.

Name: Anonymous 2011-07-01 13:57

>>31
BZZT. Wrong: I'm in one of the leading universities of my country.

Name: dubzbot-ng 2011-07-01 13:57

:GJS1M 67dcbdbce4a0b67c4b48e86a6ae29205a95e4b83024a9d947213d1231800e8d9
:42 fa4d746e69046501b5a35b1809a0b9db
:1290345671 1309543009

>>27
<-- check 'em

Name: Anonymous 2011-07-01 13:59

http://atomontage.com/?id=dev_blog#jan03_2011 This is real, voxel-based engine which isn't vaporware

Name: Anonymous 2011-07-01 14:01

>>34
I'll believe it when i get hands on a demo or game using anything "unlimited".

Name: Anonymous 2011-07-01 14:43

>>34
isn't vaporware
no demos

probably is http://thermite3d.org with fancier models

Name: Anonymous 2011-07-01 15:09

>>34
The first time I saw that "atom engine," maybe a couple of years ago, I had a boner for maybe five minutes, then just emptiness and despair because it makes the failure of voxels obvious.

They show you these amazing visuals and get you thinking, "hey, maybe voxels really are the next big thing..."  But then you see them doing some simple scene interaction and it all falls apart.

He has some command that "removes" a big sphere of voxels.  Looks cool, at first, and then you see that fragments of the world are just left floating in mid air as he removes big chunks.  Hey, why aren't those little pieces falling to the ground and obeying physics?  There's the catch.  If you're going to simulate a scene in voxels, and you want them to actually behave realistically, then you're going to need to do N2 physics calculations in every frame, where N is the number of voxels in the scene.  You can surely optimize that down, but you're not even going to come close to N.  And even doing N physics calculations is not feasible in the near future.

With polygons, you don't need to do physics on every triangle.  With voxels, you do, unless you invent something that's the equivalent of the bones/skinning system we use with polygons, and just do physics on that...  but then you've thrown out most or all of what made voxels interesting in the first place.

Name: Anonymous 2011-07-01 15:27

voxxy, the queen of /v/

Name: Anonymous 2011-07-01 18:41

I'm currently investigating voxel oct-tree raycasting and deferred shading. Don't have anything to show yet, but perhaps in a couple of weeks. I'm using OpenCL for GPGPU acceleration.

Name: Anonymous 2011-07-01 22:48

bezier surfaces anyone?

Name: Anonymous 2011-07-02 11:25

voxelstein!!!
Newer Posts
Don't change these.
Name: Email:
Entire Thread Thread List
All trademarks and copyrights on this site are owned by their respective parties. All comments and posts are the responsibility of their own posters.
- Tinychan + 2ch Mode -