Unlimited Detail: Bye, bye Polygons?

Unlimited Detail is a new technology for making realtime 3D graphics. Unlimited Detail is different from existing 3D graphics systems because it can process unlimited point cloud data in real time, giving the highest level of geometry ever seen. More info, video and screenshots inside!

Read Full Story >>
The story is too old to be commented.
retrofly3119d ago

If its true, then we will finally have grass at long draw distances, so we can hide in grass, yay!

Obviously there will be other benifits :P

TOO PAWNED3119d ago

yes this could be next big jump, like from 2d to 3d

ThatArtGuy3119d ago

I'm thinking if this goes through as presented, OnLive is probably going to be shaking in their shoes.

thor3119d ago

Unfortunately you can't animate this stuff very easily... so your nice blades of grass won't look very realistic when they are completely static.

phosphor1123119d ago

Hmmm, I don't have much to say other than it seems really interesting. My question is... so does this rendering just take models made by polygons but renders them in real time as points?

maxbyte3119d ago (Edited 3119d ago )

sorry but this looks lame. Reminds me of novalogic's voxel rendering.

SOAD3119d ago

They've had this technology for 6 years. It's very crap tech. Firstly, it's impossible to render on current CPU technology (only super-computers can process billions of dot data objects), and secondly, it's not as great as it makes itself out to be. Polygons are still a safer bet because as they get smaller and more numerous, objects begin to look more and more realistic. I saw this video yesterday and was shocked by how mis-informative it is. It was on Digg. Polygons help render geometry much better than dot data. Polygons help create the 3D effect that we love seeing in new video games, and even more importantly, realistic lighting effects and volumetric techniques work better with polygons than with dot data renders. All in all, this technology is not going to trump polygons. It's a cool little idea, but here's the thing. The positive spin on it is that it will allow many more objects and longer draw distances. The fact, however, is that the visuals will look crappier and more unrealistic than ever.

catguykyou3119d ago

Hey SOAD, nice to see you can read an article. The biggest benefit of unlimited data is they created a sdk that allows point rendering without a huge cpu hit. In fact, they said they are able to do it through software with little time at all. They can get this running on the DS for goodness sake. They talk about the algorithm that lets them do it in the videos on their site. The art looks bad in the images because it is programmer art and well, it is bad.

The break through this company has had deals with how to render the dots without it being costly to cpu. In fact, they are claiming better performance than what is possible with Polygons. It even takes things out of the equation like Anti aliasing since there are no jaggies to smooth out.

SOAD3119d ago (Edited 3119d ago )

I didn't read the article because I watched the video YESTERDAY on DIGG. Guess what. Most of the articles that are approved on N4G are actually found elsewhere on the internet.

I found out about this technology a long time ago. Don't fall into the hype man. The guy in the video spends a third of the time bashing graphics card companies for shunning their dubious tech. The guy even compares the dot-data rendering algorithm to webcrawler algorithms used to power Google and Yahoo. He's ridiculous and his company is ridiculous.

And like I said before, dot-data rendering has severe drawbacks that no one has been able to work around yet. They don't react well with realistic lighting effects. Volumetric fog and lighting do not act on the dots in a realistic manner the way they do with polygons.

Finally, these guys want to completely change a reliable method of graphics rendering even though polygons have only one drawback. The only drawback that polygons have had is that the less polygons you have, the flatter your textures look. But guess what, every year, graphics cards get better and better, and we have already reached a point where this is no longer a problem. We can cram billions of polygons on screen and create perfect 3D textural geometry. DirectX 11 is already paving the way with this.

So, explain to me why this piece of crap dot-data processing method is somehow going to revolutionize an industry that has no drawbacks. Dot-data rendering is not even as promising as ray-tracing. At least ray-tracing may actually be a norm one day.

ChozenWoan3119d ago (Edited 3119d ago )

You know how pictures on newspaper tend to look... spotted, that's how this engine appears to render.

I don't know, with some more work I could see this being useful for environments. If it means that the environments are fully manipulative, otherwise....

Keep those polygon crunching vid cards coming boys.

TheIneffableBob3119d ago

Yeah, the problem with this is that there might be issues with animating (notice how everything is static in the video?).

+ Show (7) more repliesLast reply 3119d ago
D4RkNIKON3119d ago

The PS3 and Xbox 720 can both take advantage of this tech ;P

ironmonkey3119d ago

lol thats funny how you said 720, what happend to 360? LOL

catguykyou3119d ago

The Ds and cell phones can use this tech. Read the article.

Akagi3119d ago

That was really interesting, but how long till its implemented?

ThatArtGuy3119d ago

He says in the demonstration that if the SDK goes through it will be seen probably in 16 months time.

St03119d ago (Edited 3119d ago )

If it's possible with today's consoles and hardware, why hasn't anyone else like Nvidia thought of it yet?. It'd be nice, but I think there will need to be new consoles to take advantage of this tech. Hope I'm wrong though

catguykyou3119d ago

Hard ware demand for normal dot rendering is huge. More than a super computer can handle in real time. That is why ATI and Nvidia haven't done it. This company came up with a way to process the data without it costing the cpu hardly anything. They use a similar algorithm as a search engine would use to only render what you would see at that time. With these techniques they are able to run this on cell phones and even a DS. Imagine better than PS3 graphics running on your DS.

lzim3119d ago

watch the videos.

Nvidia and AMD/ATI want to keep selling more polygons per second. That's the propaganda. Thankfully their cards have more ram and math power so the processing power can be applied to other things (physics/AI/animation).

After that is that Unreal Engine is kinda good at pushing many textured polygons that are really nicely shaded. You'd have to ask Epic Games to license or clone this technique. But they'd probably not drop textures and polygons until developers asked for something like this. They'd need to be able to use it better than textures and polygons before anything changes. Epic could put support for Unlimited Detail's point-cloud rendering into UE4 for devs who wanted to license it.. but that still won't make hardware makers care. They'll just fight over how many points they can render and animate per second.

+ Show (1) more replyLast reply 3119d ago
Timesplitter143119d ago (Edited 3119d ago )

How do you model stuff with unlimited detail?

Surely this has to be limited by what modelers can accomplish? I'm pretty sure no one can really come up with a mathematical expression that automatically models what he has in mind.

But hey... if games start looking like Avatar it'll be just fine!

Ju3119d ago (Edited 3119d ago )

Probably the traditional way.

You need to define that somehow, then "bake" it, so that the "renderer" (it still renders) can process those data fast.

... well, I guess ...

If that works what they say, then how this works is, like using a "look ahead algorithm" and find the final pixel on screen before you draw them. The data have to be prepared in memory to allow the search algorithm to find that pixel fast. But, still, it will contain some 3D position data + color information (must have).

What they solved is, using unlimited geometry and render without overdrawing. Genius idea.

For the data stored, imagine the time when video streaming became a reality. Encoding took forever, but decoding can be done in realtime. Same here. Geometry might need to be "encoded" to the "render stream".

Geometry will not disappear. We will need it to compose the scenes.

psman0123119d ago

How about, like Natal, they scan objects? Obviously, this would be hard to bring a tree inside and scan it, but somehow...

lzim3119d ago

still find it foolish that Natal hasn't been presented as a wireless device.

lzim3119d ago

Most 3D models for games and movies are made VERY detailed and dumbed down like crazy before a final mesh is made for a game.

Some studios already employ model makers who sculpt models and scan them in, then dumb then down to create polygon meshes.

+ Show (1) more replyLast reply 3119d ago
Abriael3119d ago

I actually watched it a lil while ago. The demo looks very fishy, and it's implementation in real games seem to pose quite a lot of problems.

And for goodness' sake. They were so good to create such an allegedly "advanced" technology, but they can't even design an half decent website? Paint me very, very dubious.


Abriael3119d ago

Also, forgot to mention. ALL the models they're showing look very organic. There are no flat or precise surfaces. This really makes me doubt the application of this technology to real gaming and it's precision.

If it's so good, why not showing us some flat surfaces too? If the technology can draw organic (and pretty much granulous, i think they look fugly) monsters, but it can't draw a car with the necessary precision, then it's useless.

Ju3119d ago

Its good enough that someone should invest and research if that can have a future, IMO. At least a fresh approach since a long time.

Abriael3119d ago

Fresh approaches are all nice and good when they have some actual application. From the demo it's applicability is still very very much in the dark. It's all static stuff, all suspiciously organic, and the detailing looks nice at a distance, but not that much from nearby.

At the moment I'm unimpressed honestly. Maybe if they show more, with animated content, better detail and precise surface, I might be interested, by so far it seems like a gimmick.

lzim3119d ago

Turst me, they did awesome by rendering curves. You might be to young to understand that curves are some of the most impossible geometries to put into a game.

Making straight lines in their technique would be a joke. All the advanced per pixel shader stuff you find in professional games should be very easy to implement. Beyond that they can accomplish all kinds of awesome stuff like particle effects given the whole system is based on particles. Particle effects include real volumetric smoke that has proper lighting (shadows), and fire, water, blood on NPCs etc. All that stuff is hard to do with textures and polygons.

If their system animates properly in realtime for stuff like fire/water/smoke, these guys stand to trully revolutionize building and rendering.

JoelR3118d ago

well the big advantage to something like this is
hybrid engines. you could do a poly engine for close range animation and collision detection on the unlimited detail objects at range...
that means great looking closeups with impressive and immerse long range draw distances.

+ Show (2) more repliesLast reply 3118d ago
Show all comments (38)