Whittacker's Shiturn

oceanmaster:

nitrox72:

oceanmaster:

diarrheaheartfailure:

Um. Videogame evolution to the tune of 2D evolving into 3D. These graphics are 100,000 times more detailed than Crysis 2 or any leading game of today. How detailed is that? The individual grains of dirt on the ground have seperate geometry.

And how? By replacing polygons with atoms.

NVIDIA IS FINISHED. NO, eVERYTHING IS FINISHEd

Spongebob_opens_all_his_wallets.jpg

I don’t know how tighter we can make these graphics, man. They are off the CHAIN.

Oh man I remember seeing this tech last year, so glad they’re still going through with it.

Makes me wonder how they’d handle particle effects like fire and rain and whatever.

i have absolutely *no idea* how this is going to even begin working

how it doesn’t manage to completely fry the gpu is already baffling it in itself, let alone individual grains of dirt with unique geometry

I’m going with the approach that the Unlimited Detail engine renders what you see, rather than what’s in view. I’m basing this on what I’ve seen from older videos as well, though I don’t think this has actually been explained in depth.

Straight up: Today’s engines renders triangle and quad polygons. Unlimited Detail renders dots. It’s like going from a papercraft model (a series of folded edges to produce the shape), to a sculpted figurine (atoms).

Engines these days tend to render everything in a select area. So say, if you have an enemy in your view, it’ll render the back of the enemy as well as the front, even though you can’t necessarily see the back. If something is highly detailed, it will take the time to render every single detail, even if you’re too far away to really appreciate it (unless the developer makes the engine swap high detailed models out with low detailed models when you’re far away, ie buildings in GTA). Sometimes, even anything that’s behind you will be rendered. You can’t see that shit, it’s entirely pointless to render, but the engine does it anyway.

Instead, with Unlimited Detail (and this is speculation from observation), because you can only see what’s facing you, everything that’s facing away from your point of view isn’t rendered, ever. If it’s partially hidden behind a wall or a tree, that hidden section isn’t considered. If it’s behind you, the system doesn’t care for it.

Check out this much older and scrappier video from last year - Do your best to ignore the music. Around 0:55 onwards, look at the bottom and side edges of the screen. You can see points disappearing because they’re going out of view. If you look carefully at the features creeping past the bottom of the screen at 1:08, you can see that they’re actually made up of billions of small dots; Almost literally 3D atoms.

In any case it’s all very impressive blah blah blah DEFINITELY still requires a higher-end machine to handle it, but probably only as high-end as you need to run the latest games out on the market. You could do it on an XBox or PS3 no sweat.

edit: Al-Ek in the notes brought up a good point. Because the engine is using dots, instead of models with polygons that are connected to one-another, animation is a different story. It’d be interesting to see how this engine handles animation, since you can apparently simply model your characters as usual and convert them to a series of points. There’s the question of physics, too.

Listen to Notch:
http://notch.tumblr.com/post/8386977075/its-a-scam
Scam may be too strong a word, but he’s right. 

(Source: diarrheaworldstarhiphop)