Jump to content
Spartans Home

The Future of Gaming?


MH6~SPARTA~
 Share

Recommended Posts

Euclideon has done some impressive work in the area of voxel rendering, and has recently released a video showcasing their efforts.

 

http://www.youtube.com/watch?v=00gAbgBu8R4

 

 

While the video is impressive, I'm still a bit skeptical. It'd probably take a few years before any major game companies started using a similar rendering pipeline.

Link to comment
Share on other sites

Ohhh this is amazing!

 

This is a great innovating technology. I commend these guys to push toward this direction.

 

I hope the Big Companies will look at this way and possibly the greed for business will not just keep such innovative way under the heels.

 

Unfortunately is not what is good making money, but what makes money is good, therefore if this requires a massive transormation on how Videocards handle information (Polygons vs Atoms), it could be really tough converting a million dollar business... but well, we ll see :)

 

Good luck to this project :)

Link to comment
Share on other sites

I just wonder how much GPU horsepower that will actually take to run. They dont tell you what they are running it on.

 

It's entirely software-based, apparently. So given some hardware acceleration it'd probably be pretty damn good.

Link to comment
Share on other sites

That would be pretty amazing. The industry is going to need this kind of innovation here soon. Games are getting so advanced and large that anything that cuts down on the number of man hours needed to make the game and not make my GPU burst into flames is needed to keep the industry moving forward. I think this and even more so computer generated environments (or just the technology behind it) are going to be huge in the in the very near future.

Edited by PvtHopscotch
Link to comment
Share on other sites

Awesome,but 20fps on software.....thats fucking amazing....in 3 years you will be able to run this on your pc at 60fps and Nvidia is bankrupt.....wtf.

 

Im sure its gonna take a long time before we see this on our pc's,if its gonna happen anytime soon it'll be a small or medium sized developer\publisher....games like Cod and BF will sell millions of copies no matter what....so big company's like EA dont give a shit about new technology.

Link to comment
Share on other sites

looks like minecraft with gfx update :D

 

think the demo is a bit vague about the tech (ok it's a short vid), afaik none of this stuff can be animated either. might have a future somewhere along the line if consoles can handle it and a game dev invests $xxx millions, but who would risk it?

Link to comment
Share on other sites

looks like minecraft with gfx update :D

 

think the demo is a bit vague about the tech (ok it's a short vid), afaik none of this stuff can be animated either. might have a future somewhere along the line if consoles can handle it and a game dev invests $xxx millions, but who would risk it?

 

Yes, animation is a problem. Which is why polygonal models would most likely be used for any animated characters, and the voxel/point-cloud rendering would be solely for static props and objects in the level, including the terrain itself.

 

The great thing about having a voxel landscape is that it makes it very easy to have both destructible environments and also to have true 3D landscapes, in that you can have caverns and cliffs and all the nifty features you can find in games like Minecraft without having to create custom 3D assets for it.

Link to comment
Share on other sites

The great thing about having a voxel landscape is that it makes it very easy to have both destructible environments and also to have true 3D landscapes, in that you can have caverns and cliffs and all the nifty features you can find in games like Minecraft without having to create custom 3D assets for it.

 

 

 

I hope they do make something of this but I am pessimistic.

 

 

 

The problem with 'Voxel' technology is the same 'NOVALOGIC' found in the late 90's & early millennium, that is, it takes alot of power to process the landscape aswell as all the objects on that landscape, not to mention how AI would deal with the complexity. Today AI fail in the most part to deal with any landscape with objects, bumping into things, failing to navigate around them or recognise the use of an object.

 

For example, take a simple set of stairs, how many games have you seen where object collision is poor of fails because the physics don't know how to interact the two on several occasions. Sometimes you get it to work other times it breaks through. NOVALOGIC used a simplified arena to play on and even then they had problems.

 

http://en.wikipedia.org/wiki/Voxel

 

A graphics card main load comes from the mathematical calculations it has to do on the coordinates of 3d objects made up of pieces, commonly polygons (as they were the closest compromise to a circle without looking too boxy. If you had partials (molecules) the mathematics in calculating the points is tremendous.

 

The main advantages is the landscapes & objects can look very realistic or organic as they are based on fractals or natural algorithms. Either way, all I think will happen is the polygons will get smaller & you will see less of the lines of structure (the polygons themselves).

 

OR a hybrid technology step where voxel is the landscape & polygons is the active objects, player, vehicle, destructible elements in motion.

Lastly, don't forget memory, to store any point you need a min. amount of memory, now as you increase the number of points what do you think will happen on the memory required???

 

On the fly realtime compression? why not use that on polygons? they already do? shit.....

 

Check Nvidia CUBA.

 

Only place I know where VOXEL tech is of use is in medical scanners where high resolution molecule styled images are used. and we know how slow they are..... MRI/ CT.

 

 

 

 

 

 

ANYWAY, back to these guys 'UNLIMITED DETAIL'

 

http://unlimiteddetailtechnology.com/description.html

 

 

 

 

Atomontage Engine

 

 

 

 

NVIDIA

 

http://www.youtube.com/watch?v=lpfaFrazOn4

 

 

 

KEY WORDS:

 

VOXEL/ ATOM / UNLIMITED DETAIL / POINT CLOUD.

Link to comment
Share on other sites

The problem with 'Voxel' technology is the same 'NOVALOGIC' found in the late 90's & early millennium, that is, it takes alot of power to process the landscape aswell as all the objects on that landscape, not to mention how AI would deal with the complexity. Today AI fail in the most part to deal with any landscape with objects, bumping into things, failing to navigate around them or recognise the use of an object.

 

For example, take a simple set of stairs, how many games have you seen where object collision is poor of fails because the physics don't know how to interact the two on several occasions. Sometimes you get it to work other times it breaks through. NOVALOGIC used a simplified arena to play on and even then they had problems.

 

Tech. has come a long way since the early days of Novalogic, especially in regards to AI pathfinding. It'd be very easy to create a navigational mesh to handle a voxel landscape. Having the AI act intelligently is another matter, which programmers have been trying to solve since the dawn of computing. Getting the AI just to move around though is relatively simple--in fact, one of the best navigation libraries I've seen uses voxels to generate the path data before transforming it into a polygonal mesh. The AI then use this mesh to determine where they can walk, including up/down stairs, across bridges, under overhangs, etc., similar to the pathfinding used in Crysis. In fact, voxels are already used in many games, but the voxels are never actually seen--instead what is seen is the polygons that were created from the volumetric data.

 

A graphics card main load comes from the mathematical calculations it has to do on the coordinates of 3d objects made up of pieces, commonly polygons (as they were the closest compromise to a circle without looking too boxy. If you had partials (molecules) the mathematics in calculating the points is tremendous.

 

The main advantages is the landscapes & objects can look very realistic or organic as they are based on fractals or natural algorithms. Either way, all I think will happen is the polygons will get smaller & you will see less of the lines of structure (the polygons themselves).

 

Yes, calculating a trillion+ points like Unlimited Detail claims seems to be an impossible task, though they explain their tech a bit better in another video. We will eventually get to the point where we represent entire objects by the digital molecules they are made up of, but this will not come to pass for a long long time--and it is around this time we will begin writing a system to mimic the atoms and photons in our own, real world--and in real-time.

 

OR a hybrid technology step where voxel is the landscape & polygons is the active objects, player, vehicle, destructible elements in motion.

This is what I hope to see in the future (though the destructible elements would be better suited as voxels so long as they are static, such as buildings, trees, the environment in general).

 

Lastly, don't forget memory, to store any point you need a min. amount of memory, now as you increase the number of points what do you think will happen on the memory required???

 

Voxel data can be stored on the hard drive and loaded on-demand as needed, to prevent taking up large amounts of RAM. Most sparse voxel octree systems take this approach

Though I still think procedurally generating voxels is usually the best method, as far as landscapes go.

 

On the fly realtime compression? why not use that on polygons? they already do? shit.....

 

Check Nvidia CUBA.

 

CUDA allows you to harness your GPU's power to crunch numbers unrelated to polygonal data, and is very good tech. And yes compression is needed when working with voxels, but such is also the case with polygons so doesn't much matter.

 

Only place I know where VOXEL tech is of use is in medical scanners where high resolution molecule styled images are used. and we know how slow they are..... MRI/ CT.

 

Yes voxels are great for any kind of scanner due to the nature of how they work, but voxels are also used in pathfinding, subterranean landscapes, dynamic fluid/cloud rendering, etc. Voxels give you volume--that is where they shine, representing volume. This is why most 3D games where you can build physical underground tunnels and cliffs (and aren't just prebaked 3D assets an artist manually had to place) use them, because it gives you that freedom, that extra depth. Are voxels generally blocky? Yes. But that can be fixed with higher-resolution voxels or by using the Marching Cube or other algorithms to "wrap" a polygonal mesh around the voxel volumes (which can produce pretty damn good results, and you get the best of both worlds)

 

I just find voxels to be very interesting, and I think they can be used to make games better and better. In fact I plan on writing my own voxel-based game engine before I'm out of college, as I love the idea of being able to punch holes in the very ground itself. I'd naturally still be using polygonal characters/vehicles though.

And I remember seeing that Atomontage engine before, still just as impressive now.

 

Procedurally generated voxel terrain:

http://www.youtube.com/watch?v=tWN17_CNFPg&feature=related

Link to comment
Share on other sites

I suppose dreams can come true.... I hold no real prospect for this tech, like the USA debt crisis, the big and powerful will rein not this. All this tech has done is lay the base for some sweet landscapes, the numbers needed to be crunched are very very big.

 

The memory needed has to be fast, faster than a SSD on steroids, even video memory @ sub 4nS is pushing the possibilities. IF they load sections then they face pop'up problems, if they reduce the partial count then it gets like MINE CRAFT.

 

Nvidia GTX580 pushes a max of 200Gbits of data, but the guys talking about this partial engine have to push many times that. In real time, @ 30FPS min & with active models/ elements inc players (32 preferred).

 

This makes good reading

http://research.nvidia.com/sites/default/files/publications/laine2010tr1_paper.pdf

 

It basically says the data ratio of poly v voxel partial is 5:1 the voxel needs 3.5 to 5X the storage. Storage aside there are other factors pushing away from this type of tech. Like the electric/ hybrid car, the petrol/gas car does the job & is simply more efficient today and cheaper to make.

Link to comment
Share on other sites

 Share

×
×
  • Create New...