Atomontage Engine - The Future Is Volumetric, Atom-Based - Unofficial vblog #1 (part 3/3;1st half)

Part 3 demonstrates simple destruction of the volumetric content.

The engine was performing mostly in real-time on 1280x720; the notebook (with Intel C2D P7550, nVidia GF GT 130M) failed completely to capture the video from screen at interactive framerates. Therefore the video was shot using a video camera (704x576i deinterlaced and resized to 1024x576p (16:9), so the apparent low framerate does not always match the framerate on the screen).

Read more on: http://www.atomontage.com
Hi-Res screenshots: http://www.atomontage.com/?id=gallery
Dev-blog page: http://www.atomontage.com/?id=dev_blog#jan03_2011
Donations: http://www.atomontage.com/?id=donate

Please, refer to this particular video if you want to refer to this multi-part video dev-blog.

Thanks!

Brano

Category:

Gaming

Tags:

• atomontage
• engine
• atom
• voxel
• based
• volumetric
• 3D
• real-time
• rendering
• physics
• destruction
• erosion
• thermodynamics
• chemistry
• dof
• bloom
• per-pixel
• geometry
• future
• game
• building
• rock
• stone
• sand
• profile
• revolution
• yt:stretch=16:9

[SLPCaires]

I do high end CGI for game trailers.

I love this stuff, its great that you aren't just repeating a small area of dense point cloud over and over again... like that irritating euclidean bs.

I believe voxels of some sort is where we eventually will arrive at.

Re the blocky look, I was wondering if this could be dramatically improved by removing vertices which are not shared by neighboring voxels? Basically turning these silluette cubes into much more forgiving wedges and pyramids and such?

[AtomontageEngine]

@SLPCaires I prefer per-pixel geometry - that means that plotting a particular atom, or something like that, remains a trivial operation. So for example a higher-LOD version of the geoemtry will be created on the fly and possibly augmented by some noise to improve visual fidelity. It is a slower method but it has greater potential to satisfy players. Could be that future implementations of similar engines will feature both methods. There are some good reasons for that.

[TheDg00]

I didnt really understand why it could be better that this engine is used more for 3rd person games?

I think that this engine could look very inpressive in fps games and i would dream about a game where you have to battle in these kind of buildings and where you could build them too!

Anyways its an incredible work!

Good luck for the future ;)

[AtomontageEngine]

@TheDg00 Thanks. The engine is becoming really potent in 1st person view. Still 1st person view is quite more resource-demanding and therefore I think that most of the first games powered by AE and similar engines will be 3rd person ones.

[lonestaremex361]

Euclidean engine looks way better than this.

[AtomontageEngine]

@lonestaremex361 In AE I could show similar scenes to theirs. In their engine they would most likely NOT be able to show what I did in AE videos. Tons of unique dynamic volumetric content will beat surface point-clouds on every field.

[AtomontageEngine]

@FastFlyerJr what exactly is fake and why do you think it is?

[lesleyhenriquez]

I'm dead broke but I'm almost tempted to offer sex instead.

You like tall men?

[cubrman]

@AtomontageEngine Thank you. It was a very informative answer.

[AtomontageEngine]

@SLPCaires Thanks. Some generators suffer from the 'floating silands' a lot but the generation pipeline fixes most of the artifacts by filtering them away. Also I prefer formulas that don't tend to generate many floating islands and other artifacts.

[AtomontageEngine]

@cubrman (cont2. >) ... without GPUs atom-based engines would have become dominant sometime between 2002-2005. With GPUs it is still not the case. If I could go back in time and change it I would have preferred a multi-core multi-CPU architecture. Having a little lower number (compared with the num of GPU cores) of fast general-purpose independent processors would probably result in a greater variability of products. Also the development of atom-based engines wouldn't be halted at all.

[AtomontageEngine]

@cubrman (cont. >) ... the capabilities of the game engines (not only the way the geometry was rendered) were pretty much determined by the architecture of the dominant GPUs. It is not a big problem now, but for about a decade the problem was pretty much devastating. I think there is a huge difference between progress made by thousands (or tens of thousands?) of independent programmers and progress predetermined by few hardware vendors. I believe that without GPUs ... (cont2. >)

[AtomontageEngine]

@cubrman Modern GPUs are okay for rendering non-poly content, complex lighting etc. But before the GPUs were programmable, developers were very limited in what they could do. The fact that texturing was smoother, FPS higher, .. was not so great in exchange with no possibility to apply self-developed methods. Commercial pressure killed long-term developments of SW based engines and that was the end of almost all progress in this field. Since then the capabilities of the game (cont. >)

[cubrman]

@AtomontageEngine I read on your page, that you believe that hardware acceleration was the wrong choice in video-games technological progress. What exactly did you mean by this? What sort of hardware or technology do you believe is needed to support highly detailed voxel games and how would it be different from existing shader-based videocards?

Thank you.

[drak8888]

Shut up and take my money!