We are all using 40-year-old technology

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

THREAD CLOSED
 
Thread Tools Search this Thread Display Modes
Old 11 November 2012   #61
My main issue, again, is with their deceitful presentation so far. They are widely considered a joke and for a good reason. It looks and smells like a scam, so it probably is. What they claim is impossible, it is as easy as that! It is no different from all the perpetual machine scams.

Sure, they might have some search algorithm to show but it sure as hell won't be what they claim. How is this hard to understand?

If NVidia or Intel claimed this it'd be a corporate suicide, their investors would get really worried. I don't see your parallel here at all. Did Nvidia or ATI claim something like this?

And no, I won't give these scammers a break, because dishonest people make baby jesus cry
 
Old 11 November 2012   #62
Against my better judgement, I will jump in here. First of all, I agree that their demos were done with a bit too much attitude and some wild claims. The guy comes across as arrogant, talking about a polygon war and all that nonsense. Itís too bad they did this, since I think it put off a lot of people to their technology.

That being said, there is a HUGE market for visualization engines in industrial and engineering industries. This engine is far more impressive visually already than some of the stuff thatís being used today. These folks have a very good chance to break into that market and make great money doing it.

I think itís unlikely you will see this used for games, especially AAA games, since the well-established paradigms and workflows are already well developed. If we do, it will be 5-10 years and wonít be used exclusively, but rather alongside existing tech. But people on this forum sometimes forget about the VAST markets for this technology outside games. Ignore all the polygon war crap, and realize that there are people clamoring for this exact solution outside of the game industry, if it can be brought to market.

If you think the game industry is the only market interested in realtime 3D technology, whatever form it takes, you should take time to expand your view of whatís really out there.
__________________
www.psvisuals.com - 3D Visualization and Content Creation
 
Old 11 November 2012   #63
Originally Posted by BrainFreeze: And no, I won't give these scammers a break, because dishonest people make baby jesus cry


I still fail to see precisely what makes you think Euclideon are "scammers".

They've shown a work-in-progress version of their engine. The demo is on Youtube. Millions of people have seen it work.

Was the detail it could handle impressive? Yes. Was the demo artistically great? No.

In any case, everything I've seen thus far from Euclideon suggests that their tech is real, and also that it is potentially pretty powerful.

Besides, what's the alternative to it on the market? Another 20 years of "shaded polygon" games and constantly buying "15% faster" GPUs built around polygon graphics?

Maybe there is a better way, and just maybe these guys have figured out a way to realize it.

Lets wait until their software "Geoverse" comes out. At that point, it should be clearer what their tech can or cannot do.

Until they fail to show their software and "disappear", as scammers eventually do, I entertain the possibility that their tech is real, and potentially good for gaming too.
 
Old 11 November 2012   #64
Ugh, ok, sure. I mentioned about 5 times why I think they're scammers. Maybe you lack the ability to read other people's posts and/or extract information from them.

Pointless to continue this conversation any further.
 
Old 11 November 2012   #65
Polygons are well known and well studied. A lot of algorithms work on polygons. You do not need millions of voxels or points to represent a surface, in fact three suffice. You can stretch and deform them, render them, displace, search and subdivide them etc.

Voxels also have their uses. A powerful mathematical machinery for handling fields exists and is often directly applicable to voxels. If an efficient data structure to load and interpolate gazillions of voxels/points has been developed by those guys: great, but its only part of the solution. If you want to replace everything with voxels you'll have to reinvent a great deal of techniques. Have you ever tried to move a hundred million point/frame data files through a pipeline? It's anything but painless. Would this be proprietary tech, protected by patents? If so, forget it, it's not gonna go very far then. Nobody wants to work with a closed, protected, underdeveloped, all-replacing technology, controlled by a three guy company in hinterbauernhof, even if it has a few great -but nebulous- features.
__________________
Homo Effectus

Last edited by Mic_Ma : 11 November 2012 at 04:41 AM.
 
Old 11 November 2012   #66
Originally Posted by DePaint: If established companies like Nvidia, ATI or Intel made the same "Unlimited Detail" claim, and showed off a similar demo, would you be doubting them, too?


I'm pretty sure we'd be seeing the same skepticism if any other developer or researcher pulled out claims like this.

It may well be the "future of gaming", but they've got a long, long, long way to go till they supplant the current state-of-the-art and whatever the future state-of-the-art is in gaming - because technology based on polygons isn't going to stand still and wait for them to catch up.

It is also going to be difficult for them to compete against larger companies like Nvidia or Intel in this field - especially if either of those two companies decide to develop technologies using voxels.

In order to supplant the current state-of-the-art, they've still got to implement things like shadowing, deformation/animation, physically-based shading. They've got to implement content creation tools or formats which can be used to create these "unlimited detail" worlds.
 
Old 11 November 2012   #67
I personally don't mind their "scamming snakeoil arrogant" style in their presentations. If their tech is legit, IMO they can do whatever they want.

They've said they don't need investors or anyone right now so why do they need to be so serious and impressive? What's wrong with their sense of humor if they're not hurting anyone?

We should encourage any small garage startup company that tries to innovate.

Granted, Apple wouldn't have been taken as serious if they marketed themselves this way back in their early days, but then again maybe these guys are more interested in being geeks doing interesting things than trying to make tons of cash
 
Old 11 November 2012   #68
Originally Posted by earlyworm: It may well be the "future of gaming", but they've got a long, long, long way to go till they supplant the current state-of-the-art and whatever the future state-of-the-art is in gaming - because technology based on polygons isn't going to stand still and wait for them to catch up.


From the looks of their demo, Euclideon are actually "far ahead" of the polygon crowd in terms of the sheer amount of visual detail they can put on a computer screen at interactive framerates. And they haven't even tapped into GPU power yet. They get 20 - 25 FPS on the CPU alone.

Originally Posted by earlyworm: It is also going to be difficult for them to compete against larger companies like Nvidia or Intel in this field - especially if either of those two companies decide to develop technologies using voxels.


Nvida, ATI and Intel are very much locked into hardware accelerating "shaded polygons" content as best as possible, business-model wise.

By the time they look seriously into using hardware-accelerated "voxels", Euclideon will likely be far ahead of them in terms of architecture, capability, speed/efficiency, and putting an actual working implementation on the market that you can use/buy.

The CEO of Euclideon says in a Youtube interview that they are talking to Nvidia/ATI/Intel:

https://www.youtube.com/watch?v=hxtuZE5pOGA

He also says that if no deal can be struck with them, Euclideon will simply push ahead with marketing its own software solution to the problem.

Originally Posted by earlyworm: In order to supplant the current state-of-the-art, they've still got to implement things like shadowing, deformation/animation, physically-based shading. They've got to implement content creation tools or formats which can be used to create these "unlimited detail" worlds.


From surveying the Youtube materials that are public, it seems to me that they are addressing all of the issues you mention. They have shadowing working, achieved skeletal animation almost 7 years ago, and can also deform and animate things.

I don't know about "shading" of the voxels, but it seems that they are building the tools and formats for getting 3D pointcloud stuff into their engine.

All public materials considered, things look pretty good for Euclideon.

Everything now depends on whether the tech will deliver the "detail breakthrough" the promise, or not...

Its going to interesting to see how all this unfolds in the realworld!
 
Old 11 November 2012   #69
Originally Posted by DePaint: From the looks of their demo, Euclideon are actually "far ahead" of the polygon crowd in terms of the sheer amount of visual detail they can put on a computer screen at interactive framerates. And they haven't even tapped into GPU power yet. They get 20 - 25 FPS on the CPU alone.


"Far ahead" is a real stretch. Lot's of points on screen doesn't make things better. So far all their demos have looked like crap - showing static and heavily instanced environments which are constantly shaded. Modern games are visually appealing, dynamic and already rather epic and detailed. I expect the next round of consoles we'll see much greater use of displacement and physically-based shading using polygons.

I did a quick browse of youtube and couldn't see any deformation examples from the Unlimited Detailers (I've seen others do it), the animation examples from 7 years ago did illicit a laugh through.

I think your under-estimating how mundane some of this all is. I can also load up reasonably heavy point clouds in ptviewer and zoom around at interactive rates - ptviewer uses (AFAIK) the CPU and doesn't do any kind of instancing (ptviewer is a utility program that comes with Pixar's Renderman - it's nothing special - I don't think Pixar spent much time developing it). I can think of a few other programs (headus's range comes to mind) which only use the CPU and can deal with very high resolution scan meshes with ease.

Originally Posted by sentry66: We should encourage any small garage startup company that tries to innovate.


Sure. Personally I think if they toned down their pitch a bit they're get a more favorable response.
 
Old 11 November 2012   #70
Originally Posted by DePaint: From the looks of their demo, Euclideon are actually "far ahead" of the polygon crowd in terms of the sheer amount of visual detail they can put on a computer screen at interactive framerates. And they haven't even tapped into GPU power yet. They get 20 - 25 FPS on the CPU alone.

I'm not convinced. There are plenty of ways to go beyond what most current games are capable of displaying, while still using polygons. For example, render-time tessellation seems to work well.

The main problem isn't really *dealing* with complexity. The main problem is *creating* it. What good is "unlimited detail" if someone has to sit there and sculpt it?
 
Old 11 November 2012   #71
Originally Posted by Mic_Ma: Have you ever tried to move a hundred million point/frame data files through a pipeline? It's anything but painless.

Actually, a single 24k poly mesh with one set of UVs, 2 1k textures and 2 512 textures is already outweighting in sheer data several dozens of millions of points with colour information, not to mention the latter can be very efficiently compressed and decompressed through a partitioning that will also be used for sorting later on.

Not that I think the infinite detail guys have much of a leg to stand on, but the presumption that point clouds will always be data heavy is a flawed premise.

The average game level chunk you keep loaded in a sandbox game would have the footprint and sorting challenges of a much, much larger set of voxels ready for shading and a faster sorting.

There are other problems with it, which, much as you say, will require original research to catch up to some things that are givens and well known with polies, but there would also be inherent advantages and solutions to voxels for polygons unsolved or too expensive when dealing with polygons.

There are also interesting possibilities with real time meshing of dense clouds and somewhat hybrid solutions. Pros and cons, pros and cons
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
Old 11 November 2012   #72
Aren't GPUs designed to process OpenGL and DirectX commands with insane efficiency? Like they can chew through the vector calculations and have much larger FP processing capabilities than a CPU.

Would this voxel technology even work very well on a modern GPU? Are they using OpenGl or DirectX to render the voxels?

I'm only a first year CS noob, so please let me know if I'm way off in my thinking...

-AJ
__________________
 
Old 11 November 2012   #73
Modern GPUs aren't strictly tailored to OGL or DX, it's the other way around

Yes, modern GPUs would be perfectly suited to deal with large sets of voxels and related interpolations. In fact, in the medical field there is plenty use of such things already on nVIDIA cards and Fermi units dealing with all kind of scan results, and CUDA is well used already for that, and I'm sure OCL might be soon, if it isn't already.

It's also worth noting that the bottleneck is often the amount, speed and movement to ram, so something that's more shading unit heavy, highly parallelised, and relies on less data can be made pretty damn efficient, probably more so than passing around a lot of raster data.
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
Old 11 November 2012   #74
Very cool, thanks for helping to dispel my ignorance Raffaele. Crazy mind expansion happening right now.

This is totally the kind of stuff I want to get into after I graduate in a few years. I've been so tied up with the maths for the past few months, I haven't had much time for graphics programming .

I've seen some other demo videos of this kind of tech, and they look pretty impressive. Euclideon inst the only company working on this. They just seem to be the best at getting press. I suppose the fact that they are using a public grant draws more scrutiny as well.

I'm sure more and more doctors will start to feel comfortable with making a diagnosis off an LCD screen, and the field will only grow.

-AJ
__________________

Last edited by AJ1 : 11 November 2012 at 11:06 AM.
 
Old 11 November 2012   #75
Originally Posted by AJ1: Would this voxel technology even work very well on a modern GPU? Are they using OpenGl or DirectX to render the voxels?


Watch this interview:

https://www.youtube.com/watch?v=hxtuZE5pOGA

In it, the Euclideon CEO says:

1) We are using only the CPU to render the 3D scene, getting 20 - 25 FPS on a Core i7 laptop.

2) The graphics card is only used to draw the CPU-calculated 3D scene image to the computer screen. So right now, they are not using the GPU/OpenGL/DirectX for any 3D calculations.

3) Their engine isn't speed optimized yet. Bruce Dell reckons that once it is speed optimized, the Euclideon engine will do 50 - 60 FPS on the CPU alone, at around 1024 x 768 or 1280 x 1024 resolutions.

4) He says at one point that they are thinking of bringing their engine to underpowered hardware like the Nintendo Wii or DS. Bruce Dell claims that the 3D graphics power of these underpowered consoles will increase significantly one "Unlimited Detail" runs on them.


I also don't think that Bruce Dell comes off as particularly "arrogant", or that he is a "scammer".

He's promoting his "Unlimited Detail" tech in a slightly clumsy way. That I can agree on.

But if the engine can really draw a game level, pushing 10s of millions of Voxels onto the screen in realtime in the process, then his claims aren't far off.

The only thing we can really do right now, is wait for them to release their upcoming "Geoverse" Geospatial Imaging software, and test that software to see how correct Euclideon's "Unlimited Detail" claim is.

I personally think that Euclideon is potentially on to something "big".

If they get it right, "UD" could potentially change the way 3D games are made and played very significantly.

Last edited by DePaint : 11 November 2012 at 12:19 PM.
 
Thread Closed share thread



Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright ©2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 01:13 PM.


Powered by vBulletin
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.