Basically, this video:



Replacing polygons with atom-based rendering. Unlimited detail models that use single points instead of flat planes (polygons), rendered in real-time. I'm sure there are plenty of ways to reduce the processing power needed, too. This is a huge advancement in real-time graphical rendering technology (according to them, at least).

This will be very interesting if it catches on.

not only interesting
but finally worth buying high end configurations to play games

Looks promising...

Maybe this is why they keep delaying Half-Life 3

I bet NVidia and ATI etc won't be happy. In theory if they've hit the limit of maximum detail, once we get a rig that can play games smoothly we shouldn't need to upgrade. Of course there's the possibility of advancement but yeah, just imagine less upgrades to play games

Don't you mean Half-Life 2 Episode 3? Also, I wish that was true...wait, no I don't. I don't know if my computer could HANDLE detail like that!

I believe Valve said they were done with producing episodes.

I'm highly skeptical. It sounds more like a investment pitch than a demo of proven technology. Another thing they don't explain is if any major graphics card company has shown interest or if they are planning on releasing their own silicon.



There will always be room for processor improvements. Even in an "atom" based system the complexity and detail will always be on the rise and require stronger and stronger hardware.

I can see how ordinary graphics cards could handle this though. With a system like this, polygon-based rendering is trashed. Therefore, as you get further from an object, the fewer of these 'atoms' are considered for rendering (the distance between atoms gets bigger as a result of this, and the shape will stay the same). Remember, it's probable that only the model data will be atom-based. From there, lines are drawn between the atoms for rendering. (This is how I see it, anyway.) So, things will look infinitely-detailed because more detail is produced as you get closer, and it will be smooth, constant lessening/expanding of atoms considered for rendering, instead of non-smooth transitions between models of different polygon counts.

I dunno if that made any sense or not, but it made sense in my mind. With a good engine, this could be used well.

Sure, processor improvements but when you can render the maximum limit of atoms at a fast rate in smooth gameplay will in theory mean you can't really go any further than you need to.

Processors will probably get more and more capable "per square inch" rather than generally more capable (i.e. they will become more efficient, and smaller).

LoL It's FAKE
Read this http://notch.tumblr.com/

Try animating a lot of these models efficiently at the same time as you would do in games. This kind of technique is great for static things where you instance the same object 1,000 times (such as the same tree on the zoomed out island) so you don't hog memory but add many unique objects and try to animate them and you'll have a game that takes an insane amount of disk space and RAM.

EDIT: Forgot to mention Euclidean hasn't seemed to make anything truely new in this area, this has been around for a while but no company wants to use it (after all there is a lot more to games and graphics than just how your static models look - you want them to be animated and have interactivity with realistic physics simply and on restricted hardware).

Anyone here remembers programming with voxels?

Heh, not fake, but the vid does have misinformation, it seems. They just "accidentally" "missed out" some parts, it seems.

voxels are still being used today

rarely tho, but they were great at 90-ies