This is the biggest issue. A lot of hype, but their technology isn't very useful unless they've managed to solve some major, longstanding problems. It's extremely misleading.
That was my biggest gripe with it too. They just "sell" it as being the second coming. It's great, but when you choose to ignore everything else, it comes across as being dishonest. Maybe it's my lizard brain speaking, but I automatically distrust everything else said just because of that stance.
I don't think that people have issue with your factual analysis of the Euclideon system. We more have an issue with your use of the word scam. Every time I watch american television, I am inundated with ads for various medical products that claim that they have been scientifically PROVEN to do x, y or z. I know for a fact that science does not give proof, it only provides evidence. This is why we call it the theory of evolution or the theory of gravity.
Now, does that make these medical products a scam? Not really. At worst that makes it false advertising.
I think the evidence you have gathered indicates strongly that there is a flip side to their advertisement, that there is still a lot of work they have to do and that their claims are probably only barely true, or are subject to strong interpretation.
But a scam? I think not. What they are doing is still pretty cool, and I am looking forward to whatever product they release.
I have the highest respect for you Notch and personally know people on your team (Carl), but you're out on a limb here.
Sparse Voxel Octrees are frickin awesome and Euclideon aren't wrong for saying so. There are drawbacks with them, but they are one by one being picked off by research. Animation is one example, dynamic lighting and shadows can be done in screen space. There is a real chance that soon enough voxels are going to be the shit for realtime rendering.
You even agree with them that rendering octrees can be thought of as a search algo but you still hold it against them? What is up with that?
Euclideon are not saying they are more awesome than everyone, they are saying they are up against the polygon industry. And in the video they are very careful to not cross that boundary, for example when the claim they are rendering at much higher geometry resolution they say "except other examples of procedurally generated geometry".
Because that's just what this is. Of course they are repeating elements, of course they are using common knowledge in their work. So are you and so everybody else. Blizzard doesn't point that out in their marketing either.
Granted, Euclideon's marketing is using a different tone than your craftsmanlike humbleness. But that doesn't make them a scam. Soon enough their code is gonna test it's mettle ut in the wild. It will be interesting to see what happens.
You are above ridiculing people based on what can be gleaned from a marketing video. Give them a break until you can ridicule their actual code. Let them have their fight against the polygon industry.
I didn't know that there is a specific time limit to launching? I must have missed something in startup school.
Determining if something is a scam is not possible until you actually have the facts and aren't just guessing.
<oversimplification of the situation>And it's a really weird argumentation going on as well. "This is a scam - it is totally possible to do". Normally you say that something is a scam if you think it's not possible to do.</>
>I didn't know that there is a specific time limit to launching? I must have missed something in startup school.
If you're going to promise specific features(in this case, dynamic lighting, shadows, and animations), then you damn well had better deliver some part of least one of those after a year. It doesn't matter if it's handwavey or anything, you should show that the last year's efforts and money haven't been wasted.
><oversimplification of the situation>And it's a really weird argumentation going on as well. "This is a scam - it is totally possible to do". Normally you say that something is a scam if you think it's not possible to do.</>
It's also a scam if you take the results of other people's research and development(ATI has published code for a raytrace/voxel engine) and try to push it of as something brand new and innovative.
I didn't mean to imply that they are copying the open source code directly. Without a copy of their code, it's wrong to make that assumption. I'm saying that their entire rendering process, as described in the video, sounds suspiciously like what other people have released in whitepapers and source.
It probably is a completely original implementation, but I do not believe that any of the algorithms being used are original. That is where I think Notch is coming from when he calls it a "scam". It's piggybacking on other people's work, while claiming that it's their own.
To me the video is a scam. It gained massive attention attempting to compare itself to contemporary game engines without clearly representing any of it's limitations that prevent it from being used for anything. To me is screams as a way to get enough attention to continue funding of the project with no short term result (at least in the video game field that they are contrasting themselves with.)
They display two scenes side by side, but to the untrained observers eye, their solution looks fantastic, but there are important details that they never even touched upon. Dynamic lighting, keyframe/bone/ivk animations, physics/collisions, deformation. These are all extremely important elements in a modern video game to create the experience that consumers expect.
Someday this technology may replace the current approaches in video games, and it certainly has useful approaches in static 3D captures of real world scenes (crime scenes?) but the computation systems we will need for that aren't here.
Additionally, video gaming systems jump around in capabilities. This is becoming very relevant as casual gaming is making large waves in the handheld space. If the power/performance ratio of this approach can't beat more traditional polygon rendering, game developers won't work with it. Game companies are companies, have market demands to meet, not just play with the latest and greatest in technology.
I respect your opinion. If the points you make are true, I would however phrase that as a lack of product/market fit, not a scam.
Each of us put different meanings into words. I want to reserve the word scam for when there is provably no intent to create real value. To me there is no conclusive evidence of malicious intent here, only a heap of speculation.
I did the same math as Notch, independently of him, and came to the same result: This will never run on current consumer hardware because of space and computation constraints. I don't think this is a massive stretch considering the numbers involved.
The math Notch did was to prove that they are using procedural generation of geometry, which makes it feasible. In his mind it also makes it a scam. Beats me why.
Euclideon are not saying they are more awesome than everyone, they are saying they are up against the polygon industry.
The point is that they're the only ones being deliberately misleading in order to attract investment, which is pretty easy to see as unfair. There's no mention that the variety of what you're seeing is severely technically limited by space, despite specific attention being given to the fact that a "real video game would use artists and everything would look a lot better"
Calling their tech "infinite detail" instead of "voxels" feels no different than calling the iphone screen a "retina display" instead of listing its resolution details. Good marketing is not inherently evil.
Here, though, they're suffering a huge PR backlash, which I suspect is because they're out-marketing the work of passionate amateurs and academic researchers instead of e.g. big phone companies.
It admittedly does feel dirty not to give due credit in this scenario, but I can also empathise with their desire for a descriptive term which can resonate with the general public.
On an unrelated note, I'm often impressed by how large a portion of the game development community manages to stay aware of the bleeding edge in their field. I haven't seen any other arena where that's quite so much the case.
"Infinite" has a specific meaning. Had Jobs said "infinite pixels" he'd have suffered the same backlash. "Infinite detail" is deceptive because we know there's a practical limit. "Retina display" has no similar connotation. They chose that phrase so they could then give you a definition. "Infinite" already has a definition.
Wouldn't a fractal voxel octree generation algorithm technically be "infinite", though? You could zoom into objects as far as you pleased, and never run out of detail—which seems to satisfy any practical usage of the term "infinite detail" I can think of.
This gets into subjective perceptions of the detail in question. Mathematically, you are correct that fractals have infinite detail. Visually, however, it's noticeable when you've run out of interesting fractal detail and that the technique has failed to produce more detail (that the brain will acknowledge as such.)
Had they said "fractal detail" not only would they have received less grief for it, but they would have provided us with an explanation for part of their system. That last part seems to not be something they're inclined to do.
The detail needn't be fractal to be infinite though. It could be procedural e.g. a rough surface could be given infinite detail by using a pseudo random algorithm.
It's trivial to get 'infinite' detail when the data is repetitive. The way they claim it, they should be able to zoom into molecules, atoms and then quarks and so on. Anything else is not really infinite detail in the context of graphics.
They probably can zoom into atoms because there are only so many different types of atoms and there are only so many quarks. So it seems even then their claim holds. On the large (buildings) macro level that might hold as well. Many details in a building, in a city are similar. A window on the left side of a house is the same as windows in the right side of a house. Even trees are similar (hell, they are self-similar, you can use a fractal to build them). Textures are often similar and repetitive.
Yeah there is this intermediate level of detail between the molecules and whole buildings (like surface imprefections, dirt) that might not be as repetitive. On the whole however I don't think their marketing claim is that fraudulent.
>Here, though, they're suffering a huge PR backlash, which I suspect is because they're out-marketing the work of passionate amateurs and academic researchers instead of e.g. big phone companies.
From my perspective, I'm more frustrated with how they don't seem to be producing anything. They just release a PR video every year that is about the same as the PR video they released the previous year.
Until they actually produce something useful, I have no problems referring to this stuff as vaporware.
I think they aren't calling it voxels because it works differently. I have a feeling that while the data is likely stored in a voxel-like fashion, the way they are reaching in to the data and extracting it is completely different.
If you recall that tech that was shown a few years back of being able to quickly zoom around a ridiculously large image, because you only need to look up the pixels that are going to be on screen, I think they are doing the same thing with the voxel data here. Which does make it new tech, and does make it 'infinite' in that it could probably handle an infinite data set. Hardware can't, but that's a different issue.
However, the problem with their tech will be anything dynamic, as most people have already said.
I think Notch's point of the memory requirements are critical as well. For non-repetitive detail, at the levels they are talking at, it would require google size data centers to just store the data for a single 1 km x 1 km game level.
But that also makes the incorrect assumption that each of those voxels must be populated -- which isn't necessarily the case. In fact, more will be unpopulated than populated, I would expect.
> I’d just like to say that I would absolutely LOVE to be proven wrong. Being wrong is awesome, that’s how you learn.
This is the best possible attitude to have. I think calling it a "scam" is perhaps going a bit too far, I believe they can actually render what they are showing us... but I agree that this isn't really that exciting, and definitely isn't practical for the applications they talk about.
I think the biggest problem that people aren't talking about is that LOD for geometry simply isn't as important as texture quality when getting close to a surface. I'm pretty sure the reason they have so many duplicate objects in the original presentation is because they simply ran out of memory to store all the texture data for the scene, because you need really high rez textures to be able to zoom in so close to each object.
Notice that most demos which zoom in on the model don't use textures http://www.youtube.com/watch?v=VpEpAFGplnI, because they don't have textures with high enough resolution.
The implication in the video is the are no texture maps. The geometry (I agree with those who've said voxels) contain the texture information in three dimensions. That's why zooming on the palm tree actually looks good, unlike traditional polygon/texture map rendering.
That's good to know, am I still correct in assuming this still comes down to a memory tradeoff, whether you want large textures+polygons, or voxel data to represent a high detail model?
Absolutely. Information needs to be stored somewhere, whether in the voxels or the polygons/textures. I do wonder if these guys are doing something like levels of detail to allow the close-up detail without destroying performance at the macro level.
It's weird because from the tone of the presentation it sounded like a parody. Then it felt like a running gag everyone seemed to be in on. So they were actually serious?
Voxels have a bad reputation in the graphics community. Back when I was absorbed in this and writing my 3D 'engine' I got to see this debate in its infancy.
There is an intermediate form, non-uniform rational b-splines or NURBs, which the first nVidia card rendered directly. That made it possible for the NV1 graphics card to render a perfect sphere with just 6 NURBs vs triangle based engines which required dozens if not hundreds of triangles to approach the visual quality of that sphere.
Of course 3Dfx (and Matrox, and NEC with their PowerVR architecture) made triangle engines, they needed fewer gates and had far easier texturing pipelines. Once the NV1 was retired I haven't seen hardware NURB support re-emerge (although it would make for a wonderful thing given a multi-billion transistor GPU).
Voxels make the argument that once you get below the oversamping fraction of a pixel, be it 1, 4, or 16, you don't need a complex structure. And while geometry culling using a z-buffer and clipping rectangle is well trod, voxel culling has some subtleties that make it more challenging.
Voxels have, to date, over promised and under delivered.
Notch's analysis is pretty much spot on, with one minor exception, the math. Just like polygon engines don't bother creating layers of polygons to represent solids, a voxel 'rock' or other solid doesn't need to represent the 'inside' with voxels, it only needs to represent the surface of the solid to a degree where you can't see between the voxels. As long as voxel resolution is comfortably above the pixel resolution of the 2D project plane (aka the screen). So lots of data to represent surfaces, but they can be 1 'atom' thick so less than the petabytes he supposed.
Rendering then is the process of taking a rectangular solid which is 1x1 pixel at the screen surface and then expands based on the field of view out from the eye, and once you've gone deep enough into the geometry such that the cross section parallell to the screen has no gaps, you can compute and render that pixel and move on. And, as the Euclideon people point out, since you're not really doing polygons at all, changing the 'complexity' (in terms of surfaces) on the scene doesn't change either the render time or the effective voxel count.
Yes, its pretty abusive of memory bandwidth, since at range the number of voxels that have to be looked at to fill a single pixel can be large. Imagine a 747 airplane flying across the sky at a visual 35,000', if its only 2 x 3 pixels by the time it renders, its still made up of potentially billions of voxels which form the surface you can see.
> a voxel 'rock' or other solid doesn't need to represent the 'inside' with voxels
This seems to contradict Notch's (and Wikipedia's) description of a voxel. They indicate that a 3D space described by voxels has a voxel for each unit in space. The is analogous to having 1 pixel per unit of display space. If this is correct, then a rock would indeed have the inside filled with voxels.
I don't think you are misunderstanding so much as conflating two things, the co-ordinate space / construction of the world and construction of objects within that world.
It's correct to say that every 'point' in a voxel based world can be represented by a voxel in that point. It's incorrect to say that every point in a world has voxel data associated with it. Perhaps this is the place where Euclideon and Notch diverge as well.
Using a quantized world representation view (where co-ordinates in the world are quantized to voxel boundaries) then as a voxel moves through the world space it moves from point to point in that space. So a 'cloud' of voxels has an origin and an orientation in worldspace, that allows the translation of a local voxel space (the object) into world voxel space. Since you can't do sub-voxel positioning it behooves you to have an oversampling rate between voxel space and display pixels as well.
A 'feature' here is that it also allows for fine grain collision detection (harder to do in polygon space) but again its pretty expensive computationally. (boundary surface intersection of the local voxel space with the world voxel space).
What I'm not understanding is the object space. If a voxel's coordinates are implied based on its position, then there would be voxel data for every "point" in the object space. Unless we use a sparse representation of some sort, in which case it seems that we're basically using point clouds.
as I understand it, the disparity in explanations can be attributed to that we're talking about using voxels in a data structure known as a sparse octree, that would in itself be used to optimize the system by, among other things, allowing only the surface information to be stored.
Interesting that the numbers in the defense are off. Good voxel schemes store neither color nor normal per voxel. In fact, storing a normal for a voxel makes no sense. Implicit storage and volume based materials have been the standard for a long time.
For a look at a game based voxel terrain engine that's been around for a couple of years, download and try the C4 demo game and editor from terathon.com! (That engine is just one guy, too)
Given how far polygon technology has gone, when we talk about computer graphic animation nowadays in the games domain, we are talking about something where we can actually write a system that can react to things and we don't have to individually animate each from with no particular intra-frame compression game. Compare that to a Kinectimals video. Any video, I don't care which.
In the way we mean the term in video game graphics, no, that isn't animation. And they're not going to fool anybody in the industry with that claim either. I would in fact observe that they'd probably appreciate it if you and others would stop linking that video, as it's obvious from their last video they've dropped the claim to animation, or it would have been in there. It just makes it obvious how stupid their claim to "animation" is.
The big problem is that the existing polygon-based rendering system is a moving target. I don't see anything being done with voxels that's doing stuff that the existing food chain can't already do (a) better or (b) well enough.
The one seductive component of the Euclideon presentation is that their system would free up artists to not waste time building optimized models or materials, but I strongly suspect that this would not be the case, it would just change the process of optimizing content. And given that everyone will still need to target platforms which can't handle voxel rendering for some time...
I'd have to say that the Euclideon presentation would have been a lot more impressive if the landscape weren't built out of cubes. You don't need to be a great artist to, say, randomly arrange simple objects to give a more organic look to the landscape -- they're able to zero in on particles of soil but they couldn't do nice terrain contours? That definitely supports Notch's points.
If I were going to show off an island, I'd make sure the land-water interface kicked ass. Unity's terrain demo (which is polygon based and several years out of date) kicks this thing's ass.
So, now we have either big landscapes with "infinite" details or we have animation (with infinite details?) but no landscapes at all. The problem is that games need both, don't they?
Both of them take time to explain that all of those who create games and those who create video card DON'T want something revolutionary.
Are you kidding me?
The gaming market has defined itself on change. Most games companies work really hard to throw in extra eye candy to set show up their tech against other games (including their own previous versions). You aren't going to find a more pragmatic industry than gaming.
If you show up to GDC and show a better way to do something, the next year you will have 20 game companies using the same technique.
The gaming industry is not tied to polygons. Gaming was once 2d, and countless technologies have come and gone over the years. Any technology used 20 years ago has likely been replaced 4 or 5 times.
Video card makers compete with each other on performance and features. They aren't tied to polygons either, in fact both nvidia and ati have been enabling non polygon features. Can anyone say pixel shader's or GPGPU?
So back to the original Euclidean videos. Why try to make the ridiculous point that game and hardware companies are scared of change?
Euclideon makes the point in both videos (this years and last years) so this isn't just a foot note, but part of their main sales pitch.
It also seems counter productive for their own goals. I mean, if you honestly believe that game companies will be unwilling to change since they "invested billions [in polygon tech]...", it wouldn't be very smart to create a company that sells licensees to those same companies based on the tech that they are actively fighting against?
So here is theory one...
It's a preemptive FUD strategy meant to head off any reasonable questions about the technology. I mean if someone offers legitment questions about Euclideon's tech, then they must be in the polygon conspiracy camp and are just trying to sabotage Euclideon because of all the money they've invested in polygons, right?
We live in a world where its pretty easy to prove computer technology. Just release a compiled tech demo. Companies do it all the time. Source code doesn't get released as its compiled, and it would be great marketing for the company.
No, all we get over each year is a utube video with a bunch of claims (not to mention the videos lack animation, and i have yet to see any video game, not involve some sort of 2d or 3d animation.) The lack of any sort of tech demo is significant. Most tech companies (at least the real ones) release multiple tech demos.
This all seems like smoke and mirrors. Most people who are smart enough to invent something revolutionary are smart enough to know how much bullshit is out there and how important it is to prove ones claims. In fact, if you've actually done something, the best way to different yourself from the bullshits is to prove it.
I completely agree with Notch, Euclideon's video's all appears to be pandering to investors with no real plan to deliver. Investors aren't going to question the tech as thoroughly. This is where the "Scam" part comes from.
It explains the lack of a real tech demo. It explains the videos' lack animation and it explains why the videos include odd comments about the game industry not wanting to change.
It might not of been a scam to start. It's likely that Eucldeon didn't set out to steal money from investors. I imagine they hoped to find some novel solutions and take the tech further. I'm guessing they ran into the same limitations of the tech that others have found, then panicked and started soliciting money under false pretenses.
Repeating the same fraud, one year later kind of does make this a scam.
If they can't do anything special, and they are telling investors they are producing something revolutionary, that's a scam.
You can produce a search engine that uses LSA. Do a demo, and you can get impressive results (in some cases). But don't obscure the fact that you are using a relatively well-known technique, and ask for millions in funding on the basis that your not-so-secret sauce will make you the next Google.
This is the biggest issue. A lot of hype, but their technology isn't very useful unless they've managed to solve some major, longstanding problems. It's extremely misleading.