Euclideon & Unlimited Detail - Bruce Dell Interview @ [H]

[FrgMstr]

Euclideon & Unlimited Detail - Bruce Dell Interview - Euclideon has come under fire for its Unlimited Detail technology claims once again. Instead of sitting around discussing it among ourselves, we sent John Gatt to Brisbane, Australia to talk to the man himself with Euclideon, Bruce Dell. We show you the demo running in real time with hardware specs and answer a lot of questions, all in video.

 

[butterfliesrpretty]

This is a good idea and I applaud you for not calling premature bs on this like some people. The minecraft guy has a lot of balls coming out on this one. He did make a great game, but he's clearly not an expert on computer graphics.

 

[LeninGHOLA]

Thanks for the video.

 

[Youn]

This was a great response to the previous video and resulting comments. The fact they aren't looking for money hopefully calm down some of the negativity surrounding it. Why anyone would be upset over some smart dudes trying to do something new graphics-wise is beyond me...

I'm glad Carmack wasn't as mean about it as that Minecraft guy. I think his comment about "not this generation" has more to do with development issues then technical ones. I hope he and other polite techies make some comments on the validity of this, maybe even motivate them a bit to push forward and try new things?

I'm still not convinvced of the "unlimited detail" claim, but it's much more clear from this video it's more of a marketing slogan then anything else. I liked it when he was all like "nooo, don't say that, it wouldn't be good for marketing"... yeah, not surprised they are tight lipped when it comes to those things.

Thanks HardOCP

 

[kinjo]

to all the people that said it's a scam here on the forums....I...Told...you....so!!!

also awesome interview and ty to the folks at hardocp for working hard to get all the facts.

 

[Digital Viper-X-]

very nice

 

[ianken]

to all the people that said it's a scam here on the forums....I...Told...you....so!!!

also awesome interview and ty to the folks at hardocp for working hard to get all the facts.

Show it to me when all the trees don't look the same. Or the blocks. Or everything else. Point clouds. I'm sorry "atoms" suck memory like mad, and the fact that their demo is copypasta in the extreme shows that.

Later in the video they show a 2.5 year old demo and he states "there are lots and lots of objects" No. There are not. There are at most a dozen being replicated over and over again. It's a misleading claim.

They have a search that pulls an "atom" from the scene and assigns it to a pixel on the display. OK. Where are the atoms stored? in magic "unlimited" memory?

TANSTAAFL.

I'll belive their claims when I can download and run the demo myself.

 

[Raap]

Show it to me when all the trees don't look the same. Or the blocks. Or everything else. Point clouds. I'm sorry "atoms" suck memory like mad, and the fact that their demo is copypasta in the extreme shows that.

They have a search that pulls an "atom" from the scene and assigns it to a pixel on the display. OK. Where are the atoms stored? in magic "unlimited" memory?

TANSTAAFL.

I'll belive their claims when I can download and run the demo myself. Until then they sound like a division to Steorn.

I'd have to agree. I like the concept and I certainly *hope* it's as great as they make it out to be, but there are just way too many unanswered questions. So until they show that it can actually render complex animations *while still having decent fps*, that it doesn't need all the objects facing the same direction, that they can use many different kinds of objects and that they can do demos where the terrain is actually sloped and not just flat, I'll remain highly skeptical.

 

[ianken]

I'd have to agree. I like the concept and I certainly *hope* it's as great as they make it out to be, but there are just way too many unanswered questions. So until they show that it can actually render complex animations *while still having decent fps*, that it doesn't need all the objects facing the same direction, that they can use many different kinds of objects and that they can do demos where the terrain is actually sloped and not just flat, I'll remain highly skeptical.

Indeed. I think it's impressive they can replicated data on such a scale and render it all on the CPU. But an immersive world needs variety and I don't see that here.

I think this would be ideal for procedural objects.

And I hope they are just unappreciated geniuses. It'd be nice for them to delvier on their claims.

 

[butterfliesrpretty]

Show it to me when all the trees don't look the same. Or the blocks. Or everything else. Point clouds. I'm sorry "atoms" suck memory like mad, and the fact that their demo is copypasta in the extreme shows that.

Laster in the video they show a 2.5 year old demo and he states "there are lots and lots of objects" No. There are not. There are at most a dozen being replicated over and over again. It's a misleading claim.

They have a search that pulls an "atom" from the scene and assigns it to a pixel on the display. OK. Where are the atoms stored? in magic "unlimited" memory?

TANSTAAFL.

I'll belive their claims when I can download and run the demo myself.

I think the point of them doing that is saying; Hey this is this many polygons and look how many of them we put into the scene. This could all be bs but I don't think that's a fair critique, and more drawing at straws to point out a flaw.

 

[visionviper]

GPU GFLOPS are going to become a huge deal when stuff like this starts making it's way into games.

 

[Lord_Exodia]

Holy crap it's real!!!

I can't wait until the gpu comes into the picture. With further optimizations and them incorporating all other shader technology into the engine I can't wait to see this stuff at 3240x1920

 

[fattypants]

GPU GFLOPS are going to become a huge deal when stuff like this starts making it's way into games.

As will hard drive space!

 

[FrgMstr]

As will hard drive space!

Yeah, HD and memory space is so limited on the desktop now days.....oh wait.

 

[thyshallsmite]

This guy seems to have come up with a whole new way of doing things, so when you try to apply the industry standards to a whole new technique its like trying to fit a square shape in a circle hole. Just because it is impossible with the current way of doing things, doesn't mean it is actually impossible. This will shake the industry at its very core i can understand why people are screaming witch.

 

[nakedhand]

Animation seems to be the core challenge that needs addressing before any of this will fly.

Creating code for translating "atoms" and more importantly meaningful arrays of "atoms", and with deformers and perhaps soft body physics, is going to be quite the challenge. But these are smart guys. I hope they quickly starting working on how to integrate poly rigs for character animation, or come up with brand new ways of creating rigs. I wonder what is going to drive those atoms, certainly they should be able to exist with polys in their engine (I hope) but how do they interface?

Their static environments look awesome. Maybe they need to focus on providing static content, I have a feeling they wont.

If atoms can be controlled by fields and forces like particles can, this technology could make for some very original game play.

 

[theirlaw]

I don't understand how so many of you are buying into this whole thing...

If the environments they were demoing were something more than just the same objects repeated over and over again (odd that every single iteration of their technology ALWAYS repeats objects instead of having something more dynamic) then I would say that there was something interesting here... but there simply isn't.

The bottom line is this: when John Carmack rolls up and tells you what's what, you simply have to listen. And as far as I'm concerned, his opinion and what he has created is canon.

 

[sirmonkey1985]

Indeed. I think it's impressive they can replicated data on such a scale and render it all on the CPU. But an immersive world needs variety and I don't see that here.

I think this would be ideal for procedural objects.

And I hope they are just unappreciated geniuses. It'd be nice for them to delvier on their claims.

/me scratches head.. did you actually watch the whole video or just skim through it? i mean shit he spent over a half hour explaining all the technology thats used by the software, along with the laser tracing used for every objective which gives the ability to scan and add things very easily(which is true, as the people that made iracing) and the fact that the realtime demo which was put together over a 2-3 week period right before a show they were going to in germany. so of course there won't be much variety.. and i guess you missed the basic character rendering in the background during the interview giving a glimpse of whats to come.

i just don't get it how people can kiss the asses of cookie cutter gaming companies like activision(not saying you are) yet when some one tries to change the entire gaming industry to better it for the industry AND the gaming community they try to find every friggin flaw possible. FFS why not just give the guy the benefit of the doubt and see what they end up coming out with..

Holy crap it's real!!!

I can't wait until the gpu comes into the picture. With further optimizations and them incorporating all other shader technology into the engine I can't wait to see this stuff at 3240x1920

nah i want to see it in 4k resolutions

 

[NKDietrich]

Interesting video but the shortcomings of the technology should not be underestimated. There are all kinds of neat concepts out there with great fundamental ideas but some stumbling blocks that prove to be mountains.

Kudos to these guys for trying to think outside of the box, but this approach to rendering will probably come from somewhere else, and likely hybridized with a more traditional approach.

I think they'd catch less hate if they stopped overplaying it like it was a brand new idea that no one ever thought of. The marketing speak is poisonous to the merits of the tech.

 

[Derangel]

Interesting video but the shortcomings of the technology should not be underestimated. There are all kinds of neat concepts out there with great fundamental ideas but some stumbling blocks that prove to be mountains.

Kudos to these guys for trying to think outside of the box, but this approach to rendering will probably come from somewhere else, and likely hybridized with a more traditional approach.

I think they'd catch less hate if they stopped overplaying it like it was a brand new idea that no one ever thought of. The marketing speak is poisonous to the merits of the tech.

Agreed. The guy is coming off as way way too arrogant and he is kind of off putting. His attitude is hurting a lot more than helping. Calling it Unlimited Detail is just screaming for it to be picked apart.

 

[FrgMstr]

...and likely hybridized with a more traditional approach

I think this is exactly what we will see if this comes to fruition.

 

[TheWeazmeister]

/me scratches head.. did you actually watch the whole video or just skim through it? i mean shit he spent over a half hour explaining all the technology thats used by the software, along with the laser tracing used for every objective which gives the ability to scan and add things very easily(which is true, as the people that made iracing) and the fact that the realtime demo which was put together over a 2-3 week period right before a show they were going to in germany. so of course there won't be much variety.. and i guess you missed the basic character rendering in the background during the interview giving a glimpse of whats to come.

^^^ this, seriously did some of the naysayers actually watch the video from start to finish? There are some phenomenal things that are in it, that address actually exactly what you touched on in your post. Anyhow, kudos for actually watching it and drawing your own conclusions. Carmack's quote was very impressive as well.

 

[FrgMstr]

Agreed. The guy is coming off as way way too arrogant and he is kind of off putting. His attitude is hurting a lot more than helping. Calling it Unlimited Detail is just screaming for it to be picked apart.

Timmid CEOs that project only a pittance of confidence in their products are not what makes the business world go around. His attitude and claims has gotten his company all the press it is getting, don't be ignorant of the obvious.

 

[JohnnyGatt]

Interesting video but the shortcomings of the technology should not be underestimated. There are all kinds of neat concepts out there with great fundamental ideas but some stumbling blocks that prove to be mountains.

Kudos to these guys for trying to think outside of the box, but this approach to rendering will probably come from somewhere else, and likely hybridized with a more traditional approach.

I think they'd catch less hate if they stopped overplaying it like it was a brand new idea that no one ever thought of. The marketing speak is poisonous to the merits of the tech.

Your right it’s far from new the animation stuff in the video is 7 years old. What others have not got yet is the search algorithm that let them run that demo is software and not have that laptop even blink.

 

[wasteomind]

So I'm about 25 minutes into this video, and so far I'm not overly impressed. I feel like a lot is being said at the same time nothing is being said. I don't feel like I have a better understanding of his claims. I was surprised he decided to show a "real time demo" though it immediately became suspect when A he didn't know that certain keys had certain effects, and B he insisted on getting the controller back from John almost instantly after he started moving with it. I'll keep watching but these are my thoughts so far.

 

[ir0nw0lf]

There goes that damn word "unlimited" again, screwing up everything it is mentioned in yet again. It really was a bad choice of words. "Exponentially increased detail" might have been somewhat better.

 

[Willsonman]

hrm... famous last words? "When its ready" *cough* DNF *cough*

Interesting to be sure but I think timing will be everything on this project. So much has been invested in current technologies... He's gotta play his cards right.

 

[Raap]

Yeah, he talks a lot, but he's not actually saying all that much. He also appeared to show some ignorance when it comes to tessellation; it's neither a new technology as he claimed, nor is it limited to 'flat rocks'. It would work perfectly fine with the 'grass' or flower he showed from the Unlimited Detail engine, though obviously the performance hit would be noticeable.

And obviously, making a claim that something is 'unlimited' is just a silly idea, since there's no such thing.

 

[matrix563]

you guys really need a handheld stabilizer . of course you did mention you only had limited equipment so

cool interview

 

[Cerulean]

This guy seems to have come up with a whole new way of doing things, so when you try to apply the industry standards to a whole new technique its like trying to fit a square shape in a circle hole. Just because it is impossible with the current way of doing things, doesn't mean it is actually impossible. This will shake the industry at its very core i can understand why people are screaming witch.

+1

Makes me chuckle at the sheer pride, arrogance, and ignorance of all the haters out there.

 

[sabregen]

I'll be watching the interview tonight, just watched the 7 minute demo/explaination. I'd never heard of this before. Definitely interesting.

 

[magoo]

It's cool if it works with animated objects, NPCs,missiles, and physics.
Not just still objects.

Gotta love innovation though.

 

[scojer]

Haven't heard of this before either, but it seems great. It seems like the objects are solid, so to make terrain and stuff they could make hills and bam, you got rolling hills. It'll be interesting to see how this all turns out and if we'll see it applied or not.

 

[JohnnyGatt]

you guys really need a handheld stabilizer . of course you did mention you only had limited equipment so

cool interview

Yeah camera is 4 years old but it was on a tripod for most of it. 3 hours to render it too... might need to upgrade my PC...

 

[Cromulus]

For a more credible and nuanced view of voxel rendering technology see this.

Also, I don't think a lot of people here understand exactly why the repeated detail is so problematic for the credibility of Euclideon. People like Notch and Carmack are suggesting that this demo is using a sparce voxel octree (or something similar) to store voxel information. Everything Euclideon has shown is consistent with this assumption and they've never said otherwise (they always refuse to give details).

For an SVO renderer, geometric detail has minimal impact on rendering performance; hence the claim of “unlimited detail” isn't completely false. SVO data structures do, however, take up a ton of memory for this level of detail (to do a square kilometer with this much unique detail would probably require thousands of terabytes of data as Notch noted). However, if you “cheat” and construct the octree to reference the same data repeatedly you might very well be able to create something like the Euclideon demo.

 

[FrgMstr]

I'd think the impressiveness of the technology itself could gain some attention, especially if shown off at GDC in front of a crowd of developers.

<snip>

That is the problem with PR, there is a thin line that needs to be walked when it comes to public perception.

I think it is drawing attention right now. I can assure you that AMD, Intel, and NVIDIA are all now very aware of this technology.

There is no PR here. Just a CEO and a video cam. These guys are not spending resources on PR hacks....and being a middleware company, which is what it will be until it gets purchased, it truly does not need PR. Public perception has little to do with the process currently from Euclideon's point of view, they could truly care less what the public thinks. The public will not be purchasing its software. But surely the public put it in the spotlight.

 

[butterfliesrpretty]

I don't understand how so many of you are buying into this whole thing...

If the environments they were demoing were something more than just the same objects repeated over and over again (odd that every single iteration of their technology ALWAYS repeats objects instead of having something more dynamic) then I would say that there was something interesting here... but there simply isn't.

The bottom line is this: when John Carmack rolls up and tells you what's what, you simply have to listen. And as far as I'm concerned, his opinion and what he has created is canon.

you didn't watch this video clearly, or read this thread. Also John Carmack said that this won't be possible this generation but possibly the next, that's not saying this technology is bullshit in fact quite the opposite.

 

[Youn]

Everything Euclideon has shown is consistent with this assumption and they've never said otherwise (they always refuse to give details).

Well he does say it's "completely different"...

 

[L0s7 4 Lyf3]

Amazing how people here don't recognize the potential of this technology just because the same objects were repeated. Polygons represent an inherent problem with adding geometric detail. A modeler either has to add more polygons to give the object more facts, or try and make fancy textures and effects to make it look like it has more detail.

This is a fundamental change from that style. This is more akin to how things are built IRL (you remember that, don't you?). The potential for detail is truly unlimited.

Although it will require a boat load of processing power to run something as complicated as the latest games, holy moly is this an impressive step forward. This has the potential to be far more realistic than anything we've got so far. This is just a showcase, it's no where near ready for real environment building. If you're not smart enough to recognize what can be done that's your problem.

 

[TheOneKnownAsMe]

For a more credible and nuanced view of voxel rendering technology see this.

Also, I don't think a lot of people here understand exactly why the repeated detail is so problematic for the credibility of Euclideon. People like Notch and Carmack are suggesting that this demo is using a sparce voxel octree (or something similar) to store voxel information. Everything Euclideon has shown is consistent with this assumption and they've never said otherwise (they always refuse to give details).

For an SVO renderer, geometric detail has minimal impact on rendering performance; hence the claim of &#8220;unlimited detail&#8221; isn't completely false. SVO data structures do, however, take up a ton of memory for this level of detail (to do a square kilometer with this much unique detail would probably require thousands of terabytes of data as Notch noted). However, if you &#8220;cheat&#8221; and construct the octree to reference the same data repeatedly you might very well be able to create something like the Euclideon demo.

I vaguely remember him mentioning something about memory "compression" or something during the first 10-15 minutes of the video. I can't be sure, but perhaps that might explain why they're able to cram so much detail onto the screen without the need for several terabytes of memory or data storage.

And hell, even if this is nothing more than a fancy SVO renderer, the fact that effort is being made to move beyond the basic polygon is nice, and can only lead to greater things.

Moving beyond polygons will be one of the greatest things to ever happen to the industry. John Carmack noted during his keynote at Quakecon that we're fast approaching a time and age where games will take nigh on 10 years to complete. A large majority of time in modern game development is spent developing models with an extremely high polygon count, and this has the potential of removing that inhibiting factor all together.