Geez took me a min to understand what he was saying, (even though my first thought was right I just wasn't sure if that's what they were saying) a search algorithm for points, that's pretty smart though cant wait for it to become big! And for all you non believers out there saying its not real it's impossible: he may have said its unlimited but obviously it's not unlimited c'mon everyone knows that but it's just like google, think outside the box. It only displays what is needed the rest (being
anything who's explanation video says "we are going to keep it simple for the newbies, so back to the basics" is to be taken as bullshit. Give me the interesting stuff.
it can kinda be seen in the same respect as flight before we did it.. i have no doubt (well, not much) that someone has found a way to play around with the numbers for the next generation of things.. ive seen the same thing on other appications and to me this is very possible, even though i cant work out how exactly..
When will there be a public Demo rather then rendered video? also how can you create unlimited. Your basicly saying you can defy physical physics on hardware and create perpetual motion for it because of software.
This is incredible. Sadly I have my doubts whether this will ever take off. ATI and Nvidia are extremely rich companies, and money speaks louder than brilliance. If this ever goes public, they are no longer needed.
@TheReggaeMortis We might be looking into increasing pixels available to us which would probably require more power, so instead of MOAR POLYGONS we'd shift to a MOAR PIXELS kind of graphic world? Can't say I know, but I draw the same conclusion you did when I thought it through the first time round. Can't say it's not a valid point. :[
The reason why this does not look great is because it's only about geometry. If you would turn off all the lighting effects, motion blur etc. in Crysis 2 you would end up with clean geometry which looks way worse than this solution presented in the video.
Sorry but am I missing something? I am not trolling, but your demo does not look as good as current 3D engines. Do you have any realtime GAME examples using your system?
This is brilliant! I want to see more of this, and what it can do for games. This is the logical step games has to take in the future, in order to make it as real as possible. Hoping we'll take this step sooner, rather than later.
@CommonCoward doesnt minecraft already run something like this? software controlled graphics. As i get the same FPS with onboard graphics than i do with my 460gtx.
I'm not believing anything until I can download a demo or see an SDK.
I have no idea how they came up with a search algorithm that can search a billion points so damn quickly, mostly because you can't save all of those points in RAM
Also with the 1 point per pixel scheme, wouldn't you see gaps in between them when getting close?
@matieman77 Well they say an sdk in "maybe in 16 months" depending on funding negotiations. Also, while they have billions of points in the models, 720P is less than a million points, and I can imagine that with some clever work you can reduce the points searched.
And if you have a point per pixel, no zooming in would not make gaps. Each frame is rendered differently, so would grab points appropriate each individual frame.
@sausagenmuff Sorry, could you elaborate? I'm not sure what you mean by renders down to it.
What I meant was, if a wall made of 100x100 points, it would look good. If it fills up less that 100x100 on the screen. If you get closer and get 100x100 points on a 1000x1000 screen area, you'd logically get gaps.
There is a lot of speak about "unlimited" but someone still has to model the world and I doubt that there's an infinite amount of points in that "point cloud".
The only really unlimited geometry would have to be parametric.
I have to say that you should go with Nvidia so they can support you and since Nvidia have alot of influence in gamemaking companies you could have this technology to work in a few games in about 3 years.
Another thing, current games look just as good or better than what you are showing here. I realize you probably have a limited art budget, but it you showed my this and then Metro 2033 or Crysis or hell, even Spore, I would say these demos don't look as good.
Metro 2033, crysis and spore all looked like garbage in their early stages of development too, but developers don't show that part of development because it's not good publicity to demonstrate an alpha version of their game which looks nothing more than a few blocks moving around a non lit, non textured, block environment.
I don't understand why you would be so opposed to the idea of this kind of graphics advancement, you certainly wouldn't oppose it in the days of pacman, so why now?
I am not opposed to the idea of graphics advancement, I am just asking questions about what is being presented.
My question to you is why you would be so scared of someone questioning this technology?
Voxel tech isn't new, it has been around for a long time now. There was even a game released that used them. There must be a reason that the tech never became popular.
Also, I find it fishy that a few of your examples seem impressive except its just the same model over and over. Could you do that with all different models?
So you are saying that you create rules for the entire 3d scene then use a search algorithm to quickly figure out which pixels to actually show? But this changes every frame possibly. How do you keep up? Also, don't you still have to calculate the rest of the world?
mygaffer, not sure if you heard him, but several times he has mentioned that the demonstration is in it's baby stages, it's early alpha.
Did you also hear him mention that all the art in this demonstration was done by programmers? None of demonstration contents were done by an actual 3d artist.
So imagine what a 3d artist could do with this power when this system is finally polished and complete ;)
Once again, it runs slow in the city example, because it's in it's baby stages.
Do you guys think that a similar detail in polygons doesn't take as much hard drive space?
All your base are belong to detail!
err...
The point is that this probably don't take that much more space than things using polygons. Polygon things take less space because they are ugly simple blocks.
If you want a small size game using unlimited detail, then just make the environment ugly blocks like polygons.
Animating is done through sort of skeletons attached to the models. So when you move the "bone" of the arm... all the polygon vertices attached to the bone move with it. This is done the same way.... it's just there are more points. (this animation takes little time, because moving the points is easy through matrix multiplication or similar math... rendering the points / vertices would be the problem)
But would the models not require a lot more space? I mean, if the models are to be as detailed as they say, there should be a lot more information involved. This is a price I'm sure most would pay in order to get this good details.
well this will eventually see the light, right now ATI and Nvidia are both GIANT graphics cards .. and we are needing a bigger PC case... but that cannot continue. Although they are reducing processing yield constantly 65nm, 40nm and soon 28nm ..they will eventually hit the bottom
If this ever comes to term, my guess is that nVidia will focus on physics chips because from a point made earlier, something moving/animating would ruin this. ATI will probably focus on a chip that has a dedicated CPU for this.
I really don't think this is going to make it. nVidia, ATI, Autodesk, Adobe, and any other CG big-wig will have to rewrite their software all at once for this to become mainstream.
The same reason X86 still remains the Undisputed KING since the 70s.
Even though better architecture came afterward people didn't like the idea of totally reeducating themselves to use another architecture for a little extra speed.
Tessellation which already shows impressive results will render this technology less attractive to migrate to.
BTW the guy who made this presentation already Shot himself in the foot.
Oil companies and car companies like Ford, Toyota, Volkswagen killed the electric car. Why?, because the electric car would bankrupt them. The day this great technology becomes household use is the day i get to drive my 100% electric car that costs 10.00 US a month to fuel made by Toyota and I fear that will never be a day in my lifetime.
you will never see any of this hit the market because if any of those models moved, their "3d search algorithm" would utterly fail.
notice the polygon game? see what the female character does? she walks around.
the only movement of the "unlimited detail" was the camera moving. recalculating a point of view isn't difficult, it's tracking all the point cloud data that's difficult.
@Malikat42 If the point search algorithm is pre-baked, they could implement movement by using one camera position per model or bone. To smooth out bones at the joints they could use X number of cameras per joint, although this could cut the model up into slices if not done properly.
@moises0369 That's just using technobabble to avoid an issue. It's like if they said "We can travel anywhere in the world, because we can read maps!" without explaining WHO is going to make millions of square miles worth of maps. Just because they can DISPLAY the positions easily, they still have to CALCULATE any changes, and that work grows exponentially with increases in detail. @Whitedragon103 this is why your reply was irrelevant, it's not the cameras that matter, it's the models themselves.
@Malikat42 - What if each non-static object was it's own pixel cloud? Maybe you could first determine what pixel to display from the world cloud, then iterate through each non-static object and determine for that pixel, what pixel is closer to the viewer and render that one. Maybe each object, including the world, could be entered into another cloud (not a pixel cloud, an object cloud) to determine what objects are closer to the viewer, so you don't have to check every object for every pixel.
@slydog349 - Well my first part sounds like it would work, but creating another type of cloud to hold every object would fail as the objects can move, which is the original problem.
@slydog349 well, that sort of iteration is called occlusion, and it's ridiculously simple. The difficulty of making unlimited detail is the difference between moving an arm bone that is a line from point a to point b - a basic polygon type of thing, and moving an arm bone that is 10,000 tiny cubes, which is how voxels work. Once you've moved the cubes, rendering them is child's play by comparison.
@Malikat42 Good points. They would have to define models with bones, as they do with polygon models. And each point would be associated with a bone and move the relative distance as the bone would move. This isn't perfect so maybe you have also define the relationship with it's neighbors? Wow, this is going to be complicated. And much slower than just transforming some polygons. The skin only has to be one pixel thick right? Could we use both methods, one for the world, one for the objects?
@slydog349 you missed my point entirely. With polygons, you can have a bone of sorts that it is associated with, and when you move that bone, you do relative calculations for the thousands of polygons associated with it. However, with voxels, you have structures to fill VOLUME, not to cover AREA, and volume grows as the cube of the resolution, whereas area is the square - in short, for the detail of thousands of polygons, you have hundreds of thousands or millions of voxels.
@Malikat42 That would be talking about two different things. The graphics are a different system than animations entirely. Heck, shadows are completely different than polygons too, and in one of their videos they talk about the trouble with their shadow-rendering that they didn't have time to fix
But either way, that's beside the point. Most polygons on a game are stationary: trees, rocks, bricks, houses, statues...
ATI and Nvidia will not allow this. If this ever is adopted into video games computer manufactures would go bankrupt because people would no longer need to buy a powerful computer or upgrade one.
You seem to think this model would take no processing power at all, but it would indeed take quite a bit. On top of that, there is way more to computer advancement than just graphics cards and display power. Saying 'This one facet with obselete the WHOLE THING!' is like saying "This program will make every computer able to calculate Pi instantly!"
It's simply not true, and shows that you don't really understand that much about computers.
The methodology they're describing is one of the ways video games used 2D sprites to fake 3D graphics back in the 80s and early 90s.
Fifth, graphics aren't the only reason why people upgrade systems, and a new graphics technology wouldn't change that one bit.
Sixth, gas has an energy content of 36.6 kWh/gal, while electricity costs about $0.075/kWh. That means that on an energy cost basis they're about the same. Electric is a lot more efficient, but nowhere close to the $10/mo range.
If this technology had a name it would be 'Shitty McCrappypants'.
BudzMcgr33n 2 months ago
But it's only one or 2 animals just copied 1000s of times?
Make a single "animal" of which every single point is unique. Do that for a billion points...
19RedLineR74 2 months ago
Interesting, but will it work on conventional hardware as well as the current graphics do? Or will the hardware have to be changed?
OldNations 2 months ago
Geez took me a min to understand what he was saying, (even though my first thought was right I just wasn't sure if that's what they were saying) a search algorithm for points, that's pretty smart though cant wait for it to become big! And for all you non believers out there saying its not real it's impossible: he may have said its unlimited but obviously it's not unlimited c'mon everyone knows that but it's just like google, think outside the box. It only displays what is needed the rest (being
TheFatCatFelix 3 months ago
LAAAAAG im going to play mario
tsdoom32x 3 months ago
Infinite geometry! U jelly ATI?
keoni29 4 months ago
DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO! DEMO!
mrb4ng92 6 months ago
Only processing points for each pixel on the screen... GENIUS!
Now how long until we see games with this?
protox4 6 months ago
anything who's explanation video says "we are going to keep it simple for the newbies, so back to the basics" is to be taken as bullshit. Give me the interesting stuff.
JagermeisterDetox 7 months ago
@JagermeisterDetox the interesting stuff is a trade secret ofc :) I think I get the idea though , shame I suck at C++ atm :(
betastate 7 months ago
You sir, are a genius
Kid0fThePast 7 months ago
it can kinda be seen in the same respect as flight before we did it.. i have no doubt (well, not much) that someone has found a way to play around with the numbers for the next generation of things.. ive seen the same thing on other appications and to me this is very possible, even though i cant work out how exactly..
maxxi11002244 7 months ago
Imagine Minecraft with Unlimited Detail Technology. Notch would be so......oh wait.
flyingemuzZZ 7 months ago
When will there be a public Demo rather then rendered video? also how can you create unlimited. Your basicly saying you can defy physical physics on hardware and create perpetual motion for it because of software.
Sypran 7 months ago
"I've been given 5 minutes, no more."
8 minute long video
tekproxy 7 months ago 8
@tekproxy obviously hes not paying attention to the details.
havabighed 2 months ago
How should we improve from out last Graphics card? MOAR POLYGONS
ThatAdelaideGuy 7 months ago
This is incredible. Sadly I have my doubts whether this will ever take off. ATI and Nvidia are extremely rich companies, and money speaks louder than brilliance. If this ever goes public, they are no longer needed.
They won't let that happen. :[
TheReggaeMortis 7 months ago
@TheReggaeMortis We might be looking into increasing pixels available to us which would probably require more power, so instead of MOAR POLYGONS we'd shift to a MOAR PIXELS kind of graphic world? Can't say I know, but I draw the same conclusion you did when I thought it through the first time round. Can't say it's not a valid point. :[
Luponious 7 months ago
Give me a trial, don't care if it only lasts 30 min, and one secene.
If this is true, I will personally be your servent for life.
Jabberwockxeno 7 months ago
The reason why this does not look great is because it's only about geometry. If you would turn off all the lighting effects, motion blur etc. in Crysis 2 you would end up with clean geometry which looks way worse than this solution presented in the video.
varfare 9 months ago
I'd buy stock if this company was publicly traded.
CyanideXCloud 10 months ago
Sorry but am I missing something? I am not trolling, but your demo does not look as good as current 3D engines. Do you have any realtime GAME examples using your system?
ky0t0kid 10 months ago
This has been flagged as spam show
@snowballeffect Random shit, you know?
BudzMcgr33n 10 months ago
@snowballeffect Random shit, you know?
BudzMcgr33n 10 months ago
@snowballeffect I was sent here from a forum on Gamespot regarding voxel graphics. Random shit, you know?
BudzMcgr33n 10 months ago
I'm surprised this hasn't gone viral yet.
PSPfanBOY22 10 months ago
Where can I buy stock?
spetsnaz5 1 year ago
Hate the way he says data.
RaggyFTWW 1 year ago
Do you want to know what i like about polygon graphics? not being restrained by the patents like the patents that cover this technology.
*zing*
ethoxyethaan 1 year ago
This is brilliant! I want to see more of this, and what it can do for games. This is the logical step games has to take in the future, in order to make it as real as possible. Hoping we'll take this step sooner, rather than later.
RoflJoRemi 1 year ago
1 word....
Minecraft...
CommonCoward 1 year ago 21
@CommonCoward doesnt minecraft already run something like this? software controlled graphics. As i get the same FPS with onboard graphics than i do with my 460gtx.
connell256 1 year ago
@CommonCoward Minecraft is more like Voxels...
Tikvanaya 9 months ago
@CommonCoward 1 acronym... GTFO!
euroreicher 7 months ago
@CommonCoward 4 words... Notch fanboys get annoying
chaquator 6 months ago
@CommonCoward Minecraft still has render limitations and runs using a Voxel renderer.
I believe that what is being shown here is possible. Furthermore, I think it's about damn time we switched to something like this, something better.
smatteringofapplause 4 months ago
How about a system that makes use of polygons, but also bezier curves.
This way you would have round geometry but still potentially have a low poly count.
P4INKiller 1 year ago
I'm not believing anything until I can download a demo or see an SDK.
I have no idea how they came up with a search algorithm that can search a billion points so damn quickly, mostly because you can't save all of those points in RAM
Also with the 1 point per pixel scheme, wouldn't you see gaps in between them when getting close?
matieman77 1 year ago
@matieman77 Well they say an sdk in "maybe in 16 months" depending on funding negotiations. Also, while they have billions of points in the models, 720P is less than a million points, and I can imagine that with some clever work you can reduce the points searched.
And if you have a point per pixel, no zooming in would not make gaps. Each frame is rendered differently, so would grab points appropriate each individual frame.
malimbar2 1 year ago
@matieman77
lol you missed the point completely,
He didn't mean they search and find one point per pixel, it more Renders down to it.
plus YOU CANT SEE GAPS BETWEEN PIXELS< BECAUSE THERE PIXELS
sausagenmuff 1 year ago
@sausagenmuff Sorry, could you elaborate? I'm not sure what you mean by renders down to it.
What I meant was, if a wall made of 100x100 points, it would look good. If it fills up less that 100x100 on the screen. If you get closer and get 100x100 points on a 1000x1000 screen area, you'd logically get gaps.
matieman77 1 year ago
@matieman77 Isn't that the whole point behind -unlimited- points?
SentientSapient 1 year ago
There is a lot of speak about "unlimited" but someone still has to model the world and I doubt that there's an infinite amount of points in that "point cloud".
The only really unlimited geometry would have to be parametric.
Pylo01 1 year ago
I have to say that you should go with Nvidia so they can support you and since Nvidia have alot of influence in gamemaking companies you could have this technology to work in a few games in about 3 years.
Svennetony 1 year ago
this seems like every personal computer will need hardware like "google" has :-PPP
povokol 1 year ago
Notice that they don't say a word about lighting.
3243F6A8885 2 years ago
@3243F6A8885
Isn't that Obvious?
With "Unlimited Detail" you'll get "Unlimited" Complex lighting and shadowing calculations.
So in fact you gain an edge on one side but you lose on the other.
Also how will this technology behave with smooth and reflective surfaces like car bodies or transparent surfaces like glass bottles.
From what I see here nearly everything looks fuzzy.
tamenga88 1 year ago
Also, why is there so much lag when you pan across the city example after you mention Rome: Total War?
mygaffer 2 years ago
Another thing, current games look just as good or better than what you are showing here. I realize you probably have a limited art budget, but it you showed my this and then Metro 2033 or Crysis or hell, even Spore, I would say these demos don't look as good.
mygaffer 2 years ago
Metro 2033, crysis and spore all looked like garbage in their early stages of development too, but developers don't show that part of development because it's not good publicity to demonstrate an alpha version of their game which looks nothing more than a few blocks moving around a non lit, non textured, block environment.
I don't understand why you would be so opposed to the idea of this kind of graphics advancement, you certainly wouldn't oppose it in the days of pacman, so why now?
otreum25 2 years ago
I am not opposed to the idea of graphics advancement, I am just asking questions about what is being presented.
My question to you is why you would be so scared of someone questioning this technology?
Voxel tech isn't new, it has been around for a long time now. There was even a game released that used them. There must be a reason that the tech never became popular.
mygaffer 1 year ago
Also, I find it fishy that a few of your examples seem impressive except its just the same model over and over. Could you do that with all different models?
mygaffer 2 years ago
So you are saying that you create rules for the entire 3d scene then use a search algorithm to quickly figure out which pixels to actually show? But this changes every frame possibly. How do you keep up? Also, don't you still have to calculate the rest of the world?
mygaffer 2 years ago
mygaffer, not sure if you heard him, but several times he has mentioned that the demonstration is in it's baby stages, it's early alpha.
Did you also hear him mention that all the art in this demonstration was done by programmers? None of demonstration contents were done by an actual 3d artist.
So imagine what a 3d artist could do with this power when this system is finally polished and complete ;)
Once again, it runs slow in the city example, because it's in it's baby stages.
otreum25 2 years ago
There are many other industries (besides gaming) that use computer graphics for real-time applications.
stdTrancR 2 years ago
i wonder about lots of things like environmental effects too
stdTrancR 2 years ago
Do you guys think that a similar detail in polygons doesn't take as much hard drive space?
All your base are belong to detail!
err...
The point is that this probably don't take that much more space than things using polygons. Polygon things take less space because they are ugly simple blocks.
If you want a small size game using unlimited detail, then just make the environment ugly blocks like polygons.
Tsunoflare 2 years ago
Animating is done through sort of skeletons attached to the models. So when you move the "bone" of the arm... all the polygon vertices attached to the bone move with it. This is done the same way.... it's just there are more points. (this animation takes little time, because moving the points is easy through matrix multiplication or similar math... rendering the points / vertices would be the problem)
3laserbeam3 2 years ago
awesome!!!! never again buying new expensive graphiccards!
feroxxx 2 years ago
I will believe it when I see it running on my own computer.
SnowWolf91 2 years ago 28
what about tesselation, that holds lots and lots of triangles, to make it look alot more smoother to the eye.
ValiumMm8 2 years ago
If you look at the website, it tells you the pyramid things made of animals was run on a REGULAR LAPTOP with ONE CORE.
Amazing!!
hoboX10 2 years ago
just give me nvidia's Nalu mermaid with unlimited detail and I'll be happy :D
ronpack 2 years ago
At 3:55 it looks as if they're using Voxels...
However, I don't know if their approach will ever make it but this demo seems not appealing to me at all :-/
gehirnmuchte 2 years ago
@gehirnmuchte
It's because these are Voxels.
Nothing revolutionary here besides the fact that the guy who presents this constantly repeats the word "Unlimited" to make it sound revolutionary.
To attract investors obviously.
He created an algorithm to boost voxel generation by culling.
Type in "Voxel Octree" on youtube and see the similarity.
Nvidia made an even more impressive scenery using CUDA on their video cards.
The problems with voxels is that it tends to make object look fuzzy.
tamenga88 1 year ago
Amazing! Can't wait for it!
MisterModder123 2 years ago
But would the models not require a lot more space? I mean, if the models are to be as detailed as they say, there should be a lot more information involved. This is a price I'm sure most would pay in order to get this good details.
BrokenBjartur 2 years ago
well this will eventually see the light, right now ATI and Nvidia are both GIANT graphics cards .. and we are needing a bigger PC case... but that cannot continue. Although they are reducing processing yield constantly 65nm, 40nm and soon 28nm ..they will eventually hit the bottom
pacinox2 2 years ago
If this ever comes to term, my guess is that nVidia will focus on physics chips because from a point made earlier, something moving/animating would ruin this. ATI will probably focus on a chip that has a dedicated CPU for this.
I really don't think this is going to make it. nVidia, ATI, Autodesk, Adobe, and any other CG big-wig will have to rewrite their software all at once for this to become mainstream.
Just my $0.02
sporkinatorus 2 years ago
@sporkinatorus
The same reason X86 still remains the Undisputed KING since the 70s.
Even though better architecture came afterward people didn't like the idea of totally reeducating themselves to use another architecture for a little extra speed.
Tessellation which already shows impressive results will render this technology less attractive to migrate to.
BTW the guy who made this presentation already Shot himself in the foot.
You cannot IMPROVE beyond unlimited.
tamenga88 1 year ago
@sportkinatorus
So in other words... What you'll see here is what you'll get forever and ever with no additional potential for improvement.
Tessellation on the other hand didn't make that dumb claim.
There is always room for improvement until you get Billions of triangles per frame so small your eyes can barely discern them.
One of the reasons why engineers of the 2 "polygon" Companies (ATI & Nvidia) rejected this technology.
Less or No work for them.
tamenga88 1 year ago
Pause at 3:55 - They could only make billboards look that good? If this demo is made of low res graphics like this, what else looks bad close up?
DoubleTRex 2 years ago
@DoubleTRex
Voxel graphics are the 3D counterpart to Bitmap images.
Both look crappy when you zoom in.
Unless he created a separate algorithm that manages to Multiply voxels the same way Tessellation Multiplies triangles when you zoom in.
But how will the CPU be hit by that?
tamenga88 1 year ago
Oil companies and car companies like Ford, Toyota, Volkswagen killed the electric car. Why?, because the electric car would bankrupt them. The day this great technology becomes household use is the day i get to drive my 100% electric car that costs 10.00 US a month to fuel made by Toyota and I fear that will never be a day in my lifetime.
fox23vang 2 years ago
you will never see any of this hit the market because if any of those models moved, their "3d search algorithm" would utterly fail.
notice the polygon game? see what the female character does? she walks around.
the only movement of the "unlimited detail" was the camera moving. recalculating a point of view isn't difficult, it's tracking all the point cloud data that's difficult.
Malikat42 2 years ago 2
@Malikat42 If the point search algorithm is pre-baked, they could implement movement by using one camera position per model or bone. To smooth out bones at the joints they could use X number of cameras per joint, although this could cut the model up into slices if not done properly.
WhiteDragon103 1 year ago
@Malikat42 didnt he say that they figured out a way to only use the point cloud data per pixel on the screen? Im guessing thats how its possible.
moises0369 1 year ago
@moises0369 That's just using technobabble to avoid an issue. It's like if they said "We can travel anywhere in the world, because we can read maps!" without explaining WHO is going to make millions of square miles worth of maps. Just because they can DISPLAY the positions easily, they still have to CALCULATE any changes, and that work grows exponentially with increases in detail. @Whitedragon103 this is why your reply was irrelevant, it's not the cameras that matter, it's the models themselves.
Malikat42 1 year ago
@Malikat42 - What if each non-static object was it's own pixel cloud? Maybe you could first determine what pixel to display from the world cloud, then iterate through each non-static object and determine for that pixel, what pixel is closer to the viewer and render that one. Maybe each object, including the world, could be entered into another cloud (not a pixel cloud, an object cloud) to determine what objects are closer to the viewer, so you don't have to check every object for every pixel.
slydog349 1 year ago
@slydog349 - Well my first part sounds like it would work, but creating another type of cloud to hold every object would fail as the objects can move, which is the original problem.
slydog349 1 year ago
@slydog349 well, that sort of iteration is called occlusion, and it's ridiculously simple. The difficulty of making unlimited detail is the difference between moving an arm bone that is a line from point a to point b - a basic polygon type of thing, and moving an arm bone that is 10,000 tiny cubes, which is how voxels work. Once you've moved the cubes, rendering them is child's play by comparison.
Malikat42 1 year ago
@Malikat42 Good points. They would have to define models with bones, as they do with polygon models. And each point would be associated with a bone and move the relative distance as the bone would move. This isn't perfect so maybe you have also define the relationship with it's neighbors? Wow, this is going to be complicated. And much slower than just transforming some polygons. The skin only has to be one pixel thick right? Could we use both methods, one for the world, one for the objects?
slydog349 1 year ago
@slydog349 you missed my point entirely. With polygons, you can have a bone of sorts that it is associated with, and when you move that bone, you do relative calculations for the thousands of polygons associated with it. However, with voxels, you have structures to fill VOLUME, not to cover AREA, and volume grows as the cube of the resolution, whereas area is the square - in short, for the detail of thousands of polygons, you have hundreds of thousands or millions of voxels.
Malikat42 1 year ago
@Malikat42 this is not voxel technology.
snowballeffect 1 year ago
@snowballeffect You are wrong and fail alot.
BudzMcgr33n 10 months ago
@Malikat42 That would be talking about two different things. The graphics are a different system than animations entirely. Heck, shadows are completely different than polygons too, and in one of their videos they talk about the trouble with their shadow-rendering that they didn't have time to fix
But either way, that's beside the point. Most polygons on a game are stationary: trees, rocks, bricks, houses, statues...
malimbar2 1 year ago
ATI and Nvidia will not allow this. If this ever is adopted into video games computer manufactures would go bankrupt because people would no longer need to buy a powerful computer or upgrade one.
fox23vang 2 years ago
@fox23vang
Hardly true.
You seem to think this model would take no processing power at all, but it would indeed take quite a bit. On top of that, there is way more to computer advancement than just graphics cards and display power. Saying 'This one facet with obselete the WHOLE THING!' is like saying "This program will make every computer able to calculate Pi instantly!"
It's simply not true, and shows that you don't really understand that much about computers.
Szriko 2 years ago 2
@fox23vang
First, take a look at the leaf obscuring the frame at 3:55. That's the closest look we get at ANYTHING in this video, and it looks pretty horrible.
Second, ATi and nVidia have started to push the GPGPU market pretty hard, and the opening volley of that tech war is physics acceleration.
Third, ATi and nVidia don't have any control over what software people create. That's why we HAVE open standards in the first place.
Fourth, this 'technology' - ALL of it - has been around for decades-
qzmoiwxn 2 years ago 2
@qzmoiwxn
3:55 thanks for pointing that out!
And yes it does look Horrible.
It reminds me of Voxelstein.
Notice how the person who presents this never mentions the word "Voxel".
tamenga88 1 year ago
I tried to notice how this person didn't mention 'voxels' but failed when he mentioned voxels.
Shyarm 1 year ago
@Shyarm
The whole unlimited Point cloud Data is a scam.
Nothing revolutionary and in fact it looks exactly like SVO (Sparse Voxel Octree)
And you can easily run this on GPU hardware as opposed to CPU software as pointed out by the narrator.
This Demo with seemingly unlimited detail ran on a Nvidia GTX 260 months before this film showed up out of nowhere.
watch?v=xUe4ofdz5oI
I'm seriously not impressed with the demo here.
The only strength voxels have over Polygons is DMM physics.
tamenga88 1 year ago
@fox23vang
The methodology they're describing is one of the ways video games used 2D sprites to fake 3D graphics back in the 80s and early 90s.
Fifth, graphics aren't the only reason why people upgrade systems, and a new graphics technology wouldn't change that one bit.
Sixth, gas has an energy content of 36.6 kWh/gal, while electricity costs about $0.075/kWh. That means that on an energy cost basis they're about the same. Electric is a lot more efficient, but nowhere close to the $10/mo range.
qzmoiwxn 2 years ago
Served.
sporkinatorus 2 years ago