• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Unlimited Detail Technology

Status
Not open for further replies.
I've watched the videos aswell and they look like just about any other promo vid. Trying to garner interest and excitement with distorted facts about the product.
anyway all i'm saying is that unless you have access to their product you cannot say that it is absolutly a scam. you are welcome to your opinion ofcourse but it's just that until proven other wise and no you cannot prove, without a doubt, that it is a scam unless you somehow got access to thier product.
*i'm not defending this thing just saying that it may not necessarily be a scam. my stand is cautiously optimistic. who knows, they may just be able to pull a rabit out of the hat.

Isn't the CEO the same Dell that started DELL the company?
Narrator is Australian but yeah has a weird Aussie accent.

Also the universe itself is technically infinite... but thats not part of the discussion.

No, it's not just "my opinion", it's definitely a scam. There's enough evidence of it, as has been explained over and over on here. Of course they're not gonna come out and admit it, are they? :rolleyes:

Man, his accent sounds exactly the same as the voice I use when prank calling someone. Sounds like a p*sstake voice, like Lloyd Grossman. Or like Fonejacker.

+1 He has that piss-takey sound about him alright.
 
It's one thing to render the same few objects over and over again. It's a whole other entierly to rander thousands of objects that doesn't have anything in common with each other.
 
I saw this in my computing class this monday, my friend showed it to me and i was like woah!!! far out.. i m wondering if its real or fake though?
 
No, it's not just "my opinion", it's definitely a scam. There's enough evidence of it, as has been explained over and over on here. Of course they're not gonna come out and admit it, are they?

i think it's more likely a case of pulling a Peter Molyneux, eg. promising the world and only being able to deliver the moon.
 
I don't get how people keep buying into this garbage
 
130980902319.jpg
 
Well that was fairly convincing. It will be quite amusing if they can finalize a dev kit before the next xbox and ps3 launch. When every game looks like real life on any old hardware why buy some new $500-800 console? Incentives will have to move to ai and physics processing, not to mention a lot of useless bells and whistles. I'm looking at you Wii U.
 
meh seems more like the guy the sent got the same rigamarole as everyone else and dosent understand how it all works in the first place

again others before Euclideon have gotten further but still havent managed to get anywhere its nothing but the same things instanced over and over and over and over.

and Carmack himself says its not possible right now, its not IMPOSSIBLE but its not possible right now, which is what pretty much everyone has stated. and there just reusing the same footage they've already released. no actual relevance to a game,

until they have footage of Characters running around in a level that dosent reek of reused instanced items then ill be impressed, as of right now even older games from 2000 have more variance in objects.
 
Damn, it's a video... and it's blocked at work. :mad:

I'll have to critique it later.

So, to those that can view it, does thing seem real?

Just the same thing, only you can see them playing the demo and Dell answers some stupid and irrelevant questions, as in repeating what he said on the previous demo: nothing new about the technology is shown and the only relevant thing to the video IMO is that Dell has no f...g idea of what a voxel or tesselation is, while essentially confirming that they are doing Sparse Voxel Octree, even though he has no idea of what that is and says they are not doing that. Judging by the video, he basically thinks that voxels are just small blocks and that tesselation is just a fancy type of bump mapping.

Well since they play the demo we can see it's not prerendered at least, but I don't think anyone thought that. The interview does look like it has been staged anyway.

All in all 40 mins of my life wasted. Oh but I still recommend watching it.
 
Thanks for the info Crazy, Bene.

So the HardOCP guy was bamboozled by this BS? I guess this helps explain why HardOCP benchmark graphics cards by manually playing games through every time, instead of using scripted tests that give nice consistent results. :rolleyes:
 
Phew, just got through 40 minutes of watching it and listening to that guy's odd voice. :shadedshu and I can now give the critique you've all been waiting for.

Here's some observations I made while watching it:

- Still claims "unlimited detail" without qualifying it. In fact, he said near the start that he'd explain it, but I don't remember seeing one

- One of the original claims were that all this could run on a simple mobile phone. Yet, the hardware demonstrated was a high end gaming laptop with GTX 460 graphics. No mention of having this run on low powered hardware was made. He should have been challenged on this

- Animation “7 years old” and really low res – admitted it looks awful. Heck, Half-Life is 12 years old and looks miles better! Didn’t show later version of the animation, but excused it as not being ready and won’t show something unfinished. The world looks complete so what's so hard about putting together a basic animation to show the interviewer?

- “No polygons”, but there are. Pause the HardOCP video at time index 16:36, where the big green round thing is showing. Look carefully at the edges, especially on the right hand side and you will see some polygons ie straight edges. They’re not very clearly defined, but present nonetheless. His "point cloud" shouldn't show any artefacts like that. The fact the screenshot is taken with a video camera really helps to hide them, huh? This could be damning and needs to be investigated

- Zooming in real close to a palm tree did make the textures look blurry, just like in a conventional game. “Unlimited detail” would retain it’s clarity and reveal more detail as you got closer. This detail clearly had a limit


- Dell Does finally claim that the tech is voxels, but in “unlimited quantities”. Once again, the claim is unqualified. Give me a break

- A leaf on the ground got zoomed in close and looked really detailed, but could it have just been a texture, given the way the tree looked when zoomed in close?

- I noticed how he avoided zooming in real close to anything generally and the couple of times that he did, it looked very much like it was done with polygons.

- Several times he referred to the incredible number of polygons that were on the screen. But I thought you use "atoms" not polygons...?

- Once again, the amount of repetition is phenomenal and not something you expect to see with "unlimited detail"

- Didn’t say how much memory all this takes. This is a critical parameter and not once did the interviewer ask this really obvious question. From what people on this thread have explained, the memory requirement is crippling

- Zoomed out from the elephant until it disappeared. It didn’t change, like you tend to see in modern games. It just got smaller. This is a good point in its favour, but it can be achieved by simple scaling

- Said that it's all running in software, but will use the GPU once the tech is refined. Yeah, maybe, but I wanna see proof. This laptop had a powerful graphics card. Dell claims that it would run just the same on a 1994 graphics, basically anything that can display a picture. Prove it by running it on Intel integrated graphics

- Dell's explanation of tessellation was off, like benetengia said

- The demo was only running at a lowly 1024x768. The laptop looks like it has a native res of 1366x768, so why wasn't the graphics mode changed to use the whole screen? Is it because the whole thing would have run a lot slower and less impressively, perhaps?

- Benetenegia reckons that the interview was staged and I tend to agree. There were cuts in a few places where Dell was starting to get into an explanation and the video was abruptly cut off and jumped to something else. I really wanted to see exactly what this explanation was

- The "unlimited" claim was never explained or qualified. A dead giveaway that he's hiding something. Once again, you can't use infinite anything in this universe

- The interviewer was constantly bowled over and fawning at Dell and seemed afraid to challenge Dell with his own tough questions. He only quoted other people that asked them

- Dell plays Carmack and Notch off against each other, saying that they are claiming opposites, but his argument didn't hold

- Dell's manner didn't come across as terribly sincere and it did feel like he was selling snake oil. This is an impression though, so don't take it as hard and fast

In summary, I still think this is fishy. I want to see some independent and respected third party (not HardOCP) pick apart this technology and verify the claims, if they're ever given the chance.
 
I know nothing of creating a game engine, but it would be infinitely stupid to make all these claims when you're receiving GOVERNMENT funding, and ending up with nothing to show for it.
 
Just the same thing, only you can see them playing the demo and Dell answers some stupid and irrelevant questions, as in repeating what he said on the previous demo: nothing new about the technology is shown and the only relevant thing to the video IMO is that Dell has no f...g idea of what a voxel or tesselation is, while essentially confirming that they are doing Sparse Voxel Octree, even though he has no idea of what that is and says they are not doing that. Judging by the video, he basically thinks that voxels are just small blocks and that tesselation is just a fancy type of bump mapping.

Well since they play the demo we can see it's not prerendered at least, but I don't think anyone thought that. The interview does look like it has been staged anyway.

All in all 40 mins of my life wasted. Oh but I still recommend watching it.

Yeah, the demo was definitely working, but he seemed remarkably oblivious to... well, just about everything. He made it sound like he did the programming initially, but his explanations of things were pretty horrible. he didn't sound like he knew what he was talking about at all.

ERM LEVEL OF DISTANCE (erm, detail???)
WE HAVE KEYBOARD SUPPORT???
UM YEA WE'RE DOING GOOD ON MEMORY...
CAN I DRIVE CAN I DRIVE LET ME DRIVE I MADE IT MY SISTER PLAYS THE SIMS
OH GOD DON'T CRASH OH GOD... finally it happens and after making it seem like he was hiding something, they don't have collision detection. Err, ok... not sure what he was so worried about with getting right next to something.

Well ok, he sure didn't explain a whole lot. I can see some stuff making sense, like the WOO IT WILL WORK THE SAME ON ANY GPU. That's logical enough considering the GPU is probably just handling the blit / flip and nothing else. Not surprising.

The interviewer was kind of busy riding his nuts more than anything though rather than asking interesting questions, like what happens when you run out of detail? Do you get gaps, does it try to fill in the blanks, or do your atoms take on a more typical voxel look to make up the size gap?
 
Last edited:
I still believe polygons are crap and are going to be phased out but not any time soon, usually with things like ray tracing the engine needs to be powerful and have powerful hardware to run it.

Point cloud data and a search algorithm to determine only what you need sounds good in theory but tbh there isn't any real evidence like everyone else in this thread has already agreed. If it really was that great then you would have thought someone else would be looking into it.
 
I still believe polygons are crap and are going to be phased out but not any time soon, usually with things like ray tracing the engine needs to be powerful and have powerful hardware to run it.

Point cloud data and a search algorithm to determine only what you need sounds good in theory but tbh there isn't any real evidence like everyone else in this thread has already agreed. If it really was that great then you would have thought someone else would be looking into it.

Well, someone has to be first :confused:
 
Last edited by a moderator:
Well, someone has to be first :confused:

It's been 7 years+ in the making! Something good would have interest from others and you'd see parallel development, but there isn't any.
 
Well, someone has to be first :confused:

Sometimes you get people simultaneously developing the same thing. Companies are leaning towards ray tracing ect not points, there must be a reason why and it cant just be well they wouldn't be able to sell graphics cards any more.
 
In reality many people are looking into it. That's why we are skeptics. It's not that no one is doing it, plenty of people are doing it, just not at the level of performance they are claiming, and not something that coould be used in games. That's the key, dozens of proven graphics experts have been looking into this kind of rendering since a long time ago. John Carmack included, who has been looking into this kind of thing since the 90's and always returned to polygonal rendering because the technology is just not there*, it just cannot catch up polygonal rendering let alone surpass it. Every time that it may look like it could, games improve again and the posibility dissapears. UD is far from being posible on current hardware because of the many things mentioned already.

For example, they continue to show the demo at 1024x768 and get 15 frames per second. Since the tech is entirely based on pixels, at 1920x1200 they would need 3x times the power, plain and simpe. Still the demo lacks animation, proper shadows, post-process effects and hell even shaders so that wood actually looks like wood and rocks like rock. That makes up >90% of current games performance demand and they completely lack it. Sure they are still not using the GPU but there's absolutely no proof that such a thing would yield any real or significant gain in performance. GPUs are not designed for "search algorithms".

* and it never is. Just consider what I said above about resolution, just like any backwards rendering tech (i.e ray-casting or ray-tracing) it's weakness is that it scales "poorly" as resolution is increased. So just imagine that in 3 years their tech is improved, PCs are faster and they are ready for HD rendering, but oh too bad the norm now is 4000x2000 OLEDs and you are behind the curve again.
 
Last edited:
because ray tracing is achieveable within the next year or 2 with top end desktop equipment

aka GTX 680 sli 7970 xfire would be if what is expected comes true powerful enough for ray tracing etc in a current gen game, thats why were moving toward ray tracing, polygons as shown by tessellation can be pumped up rather liberally at the high end spectrum, with little consequences, add in displacment + normals then add in truly dynamic ray traced lighting and shadows you can get a much better image, Voxels can do some crazy stuff but its still 5-10 years away , ray tracing is 2 years away. Quake 3 engine and i believe Half Life 2, can be run with Ray Tracing on current hardware today, Voxels are all well and good but while things might look a bit better, without realistic lighting and shadowing its going to be a small improvement over polygons.
 
I don't think ray-tracing is any closer than voxel rendering tbh and in fact they are not mutually excludable. Actually, for pure ray-tracing voxels (SVO) are far better than polys, but ray-tracing is not the be all end all of game rendering anyway. Hell it's not even the be all end all of CG movies.

I'd like to see ray-tracing on games soon, but most people, Carmack included again, have pretty much demostrated that it's not the best option for games yet. A mixture of raster for direct light/shading and ray-tracing for indirect light and reflections is not out of the question though, but most papers coincide in that once you port some of your rendering tasks to ray-tracing, you may as well move everything so as to mantain coherence and make performance expectations more stable (imagine the 2 engines fighting for BW).
 
He's panning running around the tree and claims to be rendering "21 trillion polygons". Notice anything wrong with that?

1 A trillion anything will bog down a supercomputer, let alone a laptop. There won't be enough memory to hold so much data, either. Time index 25:33

2 Polygons? What happened to "atoms"?
 
Status
Not open for further replies.
Back
Top