Unlimited Detail
polycounter lvl 11
Some people may remember a certain video on youtube a while back. It was a presentation of new technology that would allow game developers the freedom of unlimited detail when making 3D assets for games, throwing polygons away for a point-cloud system.
There's been no updates since that first video, but very recently this popped up:
[ame]
http://www.youtube.com/watch?v=UKUuUvDSXk4[/ame]
I have very little knowledge on the subject, so I don't really have a real opinion. It certainly looks too good to be true. One question I have is how will they handle animation?
Anyways, what do you guys think?
There's been no updates since that first video, but very recently this popped up:
[ame]
http://www.youtube.com/watch?v=UKUuUvDSXk4[/ame]I have very little knowledge on the subject, so I don't really have a real opinion. It certainly looks too good to be true. One question I have is how will they handle animation?
Anyways, what do you guys think?
Replies
lower the quality, (here: level detail, not model detail)
making a fuzz, dissapear, thus generating discussions over the world
revive the fuzz to generate more discussion and attract business people
profit
???
release?
You really just want to slap him the second he starts speaking...
file>import>real world
OH YEAAH !
Personally, right now it really doesn't look like something substantial and usable.
OMG Multiple shades of shadows running at 20fps!! *fart noise
Tesselation hasnt been all that, its barely noticable.
Ray tracing is where its going to be really improved, if anyone can be bothered to get it working.
But I'll keep my eye on it. The fact that they're working on polygon conversion tools might be the selling point.
They're still one year into not having given us a proper answer on how they'll handle fully skinned meshes.
Marketing 101 :poly142:
right now their stuff looks like chrome spheres rendered over checkerboards from the year 1993, all around terrible
literally a pile of shit with unlimited level of detail on each part of it
+ lighting, animation and physics questions are still there too
But yeah, nowadays polygon count is very manageable, and alot of the processing power could be gone into better lighting and physics/interation technology IMO
always nice to see folks trying new things.
Looks exciting, no idea if it will work as an overall package, but ditching low poly and exporting your zbrush models as point cloud data sure sounds nice to me, as does scanning real world objects
Share the same reservations as you guys, especially about lighting, skinning and animating, but I am all up for people trying new things, especially as if it fails, sucks for them, not for us, and if it is the next big step, great for them, great for us too
All that said...even if it is totally static, its still potentially interesting for certain applications.
This new vid is way better than the old, and hopefully they release this SDK to the public so we can see for ourselves.
Still don't have any tools to test...
If it takes 4x as long to make a game with "unlimited point clown data" then that's a deal breaker.
Still doesn't show any physics or animation....
If you're going to give players millions of tiny atoms then they will expect them to behave that way. Why use trillions of tiny points to make something look only slightly more rounded than a polygon model? Doesn't DX11 tessellation fix this same problem?
"But if you turn on cheats, and fly around to an area normal players can't get to, behind the model, you see its been optimized...FLIMFLAM! THEY'RE LYING TO YOU! CHARLATANS!!1" Exactly ass clown, why are you making the ground out of trillions of points you'll never see? Maybe using a plane that selectively tessellates the area around you using proven lighting and bump techniques to push the effect over the top? Maybe that will look better than rendering trillions of points you'll never see. Using tools that have yet to be invented.
Maybe the solution to mip and lod jumping isn't to make things out of atoms but to create systems that transition more smoothly. Maybe create systems that automatically lod materials and meshes on the fly so instead of 6 lods full of pops we have a smooth transtion... oh wait we have that already and he's choosing to show the worst case vs his best case...
Does anyone else find it mildly humorous that they spent the last year creating a polygon to point cloud data converter? It seems like guys that trash talk polygons so much wouldn't begin their new pipeline by relying on them soo much?
-> medicine
It is because this thing will cure cancer AND aids!
If we really want to see a dramatic improvement in visuals, companies need to focus their efforts on better lighting models. A dynamic polygonal object in dynamic ray traced lighting, at par with a software renderer, is more valuable than pebbles made out of a million voxels lit by a GL point light with some ambient occlusion.
Raycast/Indirect Illumination and Shaders with Radiosity and aren't limited to 16 texture samples.
Give me those, and I can paint an entire fucking island in a day.
Speak how you want to be spoken too. If he didn't talk down to his audience, backhand the entire industry he's trying to sell his tools too and behave in an unprofessional manor he would probably not get such responses and might find a few more friends. People have alergic reactions to snake oil, not so much to the oil itself but to the people who peddle it and then peel out of town faster than you can say "hey this tastes like caster oil!?"
The delivery method aside, he's missing quite a few key elements to make a successful game. I'm glad they're working on it, but as any inventor will tell you, every invention even the successful ones are met with scorn and mockery until they prove themselves useful, especially if they aren't coming from someone that has a proven track record of producing the next big miracle invention.
... in software mode? yeah, that is not that bad.
Im actually kind of annoyed by peoples lack of vision, wtf is up with the bashing? Yes, it looks like SHIT, the guy sounds like a douche, but... it looks like shit because of the lack of lighting and shaders, not geometry. Put lowpoly meshes in unlit mode with the same textures, see how it looks... If we replaced the polygons in an UDK game with this, it would look amazing. So..the technology obviously works, what remains to see is how it works with collision/shaders/animation/physics etc.
much more impressive! I like how the truck leaves tracks behind in the sand!
And that's the big issue here, the things that make this tick also makes other things incredibly hard to implement, you can't just simply replace polygons in a game with this.
People get pissed off because he is selling the idea like a revolutionary tech without flaws, while it in fact has a lot, many of those which are too great to actually make this as flexible as todays tech is.
"here you have a typical game asset, blocky, ugly and made out of these little flat things called 'poly-gones. What I'm suggesting is to use a bunch of these little atoms to build objects instead of stupid cardboard"
"here we have a typical house hold vacuum that uses a bag and filter that gets clogged. What you need is our new revolutionary cyclone technology that gets clogged in a different way..."
Same pitch, same douchebag attitude, same negative response from me.
here is a link to carmacks proposed tech.
http://artis.imag.fr/Publications/2009/CNLSE09/GigaVoxels_Siggraph09_Slides.pdf
http://www.flashandmath.com/flashcs4/bunny/StanfordBunny.html
maybe, maybe not. Depends on the pipeline or/if it was a hybrid renderer that could handle both points and polygons, or convert point geometry to cheap physics polygonal models, etc. We will see.
here is the vid
[ame]
How does this tech work with textures? streaming? UVs?
I suppose this voxel tech would require some sort of 100% procedural texture method which doesn't really get me all that excited.
This is what I'm talking about, this tech is young and might be feasible within a decade, it's been proofed and has a more robust ground working on how it works. I can get behind this!
I don't understand why people are defending Euclideon on their tech so dreamily in anyway or form, they literally gave us no specifics on the subject matter other then taking stabs here and there on how the game industry could be better if they got behind them, running in software mode can mean a bagilion things depending on the computer they're using to run what is the working behind it, Crysis can run several millions polygons on movie quality without a hitch if you plan on making a movie, calling new form of polygon creation "atoms" is just not correct, and comparing medical field of study and the software they use for games is just incorrect as my arse to the far side of Europa.
Let me make this straight, in movies, this tech could work, even be the next big thing for movies like Avatar 2, but how the bloody hell is this going to be a viable solution to games?
How much do you bet that since games are more 'sought' after then movies in this day and age, they just decided to give games as example instead of movies?
This the same stupid argument about why games can't have PTex all over again, it's not feasible, if it was, I'm pretty sure John Carmack would have been all over it.
Yeah, atomontage is pretty nice, it uses polygons to display its voxel data though.
Unlimited detail uses a per-pixel kind of search, for each pixel on the screen to display its data. (lots of which is instanced)
And yeah, carmack for one would been all over anything that would revolutionize video game tech, and he is with his megatexture work.
Even voxel-like tech is being used in many games today, caves in cryengine comes to mind.
Obviously it's not finished yet, but I agree that it's an interesting angle to watch and see how it develops. It sounds like they want developer input when they are ready to receive it (he directly said they are excited to see what real artists can do with the technology).
There's just no sense getting upset over it, I doubt they're trying to offend anyone. All of the examples they give are legitimate downsides to polygon restrictions. It's always a bummer when you have to block off a vista or the back side of any object because you didn't have to budget to detail it all out accordingly.
Not that it's thrilling to think of how long that will make environments take to build...but it sure would look nice.
J/K
I agree with one of the comments, lets see it handle animation, physics, and collision and still be playable real-time