Biggest Computer Graphics Advance Since 3D?

[[aps]gnat 28]

When you compare those to the output of this company, do you not see any difference?

LOL .... yeh, sure, in all my examples, except maybe early INTEL, I'd suggest the PROGRAMMING was 100's of times easier for their first products.

[celery 8]

Gnat;1995463']LOL .... yeh' date=' sure, in all my examples, except maybe early INTEL, I'd suggest the PROGRAMMING was 100's of times easier for their first products.[/quote']

I was thinking more along the lines of what they've actually put out. I see nothing more than a hype commercial that could be complete bullshit. And seeing how they've shown practically the same thing over a year ago with no noticeable difference, I don't see a reason to get excited until there's solid evidence to back their mighty words.

[dmarkwick 261]

and lol at "games will be 100,000 times better)

He didn't say games, he said graphics :) if people wish to scorn quotes, at least get the quote right :) there is a whole world of contextual difference there.

[sabre4809 0]

There is an old animation video btw. Came about when the first batch of videos did.

[Montanaro 0]

He didn't say games, he said graphics :) if people wish to scorn quotes, at least get the quote right :) there is a whole world of contextual difference there.

To be honest, I was only half listening (and for good reason)

Why don't they bring an "artist" in and have him create something really cool?

I really wasn't impressed by anything I saw except the ground and some rocks and this is the first I've seen or heard of this tech.

[dmarkwick 261]

To be honest, I was only half listening (and for good reason)

Why don't they bring an "artist" in and have him create something really cool?

I think that will be their next step, to make a "proper" island environment instead of that blocky tech one.

I really wasn't impressed by anything I saw except the ground and some rocks and this is the first I've seen or heard of this tech.

Well if you were focussed on the content you were not appreciating the tech... :) not surprising if you were only half listening :D

But anyway. Each cubed millimetre is divided into 4x4x4 segments, that's pretty small segmentation. Only time will tell how well this can be moved, collision-detected, animated, lit, controlled etc.

Edited August 2, 2011 by DMarkwick

[Sniperwolf572 758]

Some more food for thought from Notch and Carmack (2).

[dmarkwick 261]

Some more food for thought from Notch and Carmack (2).

Indeed :) and I guess Carmack knows more than most about game engines. But is he familiar with the inner workings of the Euclideon engine? But, I would guess that a working title for this engine is 5 years away at least, so maybe he's good at hedging his bets :D

I do think though that Notch is being rather disingenuous with his assumptions:

Let’s assume a modest island height of just eight meters, and we end up with 0.008 km^3. At 64 atoms per cubic millimeter (four per millimeter), that is a total of 512 000 000 000 000 000 atoms. If each voxel is made up of one byte of data, that is a total of 512 petabytes of information, or about 170 000 three-terrabyte harddrives full of information. In reality, you will need way more than just one byte of data per voxel to do colors and lighting, and the island is probably way taller than just eight meters, so that estimate is very optimistic.

So obviously, it’s not made up of that many unique voxels.

That sounds like an attempt at baffling us with numbers. That would be like saying the ArmA2 Chernarus map cannot possibly have 1,000,000 rocks, trees, bushes & grasses in it because there's not enough room in memory to hold that info. Quite obviously there is optimisation, instancing (or whatever the analogous procedure is called) and reduction going on, why would he even make the ridiculous assumption that a 1/64th cubic millimetre detail is being rendered at 1km?

As I said, it's disingenuous at best. At worst it's a spoiling tactic aimed at clueless people.

Edited August 2, 2011 by DMarkwick

[soul_assassin 1723]

Gshc8GMTa1Y

voxel terrain engine. Quite interestingly the car is polygonal (see other videos)

Indeed :) and I guess Carmack knows more than most about game engines. But is he familiar with the inner workings of the Euclideon engine? But, I would guess that a working title for this engine is 5 years away at least, so maybe he's good at hedging his bets :D

I do think though that Notch is being rather disingenuous with his assumptions:


That sounds like an attempt at baffling us with numbers. That would be like saying the ArmA2 Chernarus map cannot possibly have 1,000,000 rocks, trees, bushes & grasses in it because there's not enough room in memory to hold that info. Quite obviously there is optimisation, instancing (or whatever the analogous procedure is called) and reduction going on, why would he even make the ridiculous assumption that a 1/64th cubic millimetre detail is being rendered at 1km?
As I said, it's disingenuous at best. At worst it's a spoiling tactic aimed at clueless people.

Im sorry m8 but you are wrong, what Euclideon are doing is not supersecret tech, its based on well published papers. Just a few off the first google page:

http://www.tml.tkk.fi/~samuli/publications/laine2010i3d_paper.pdf

http://www.daimi.au.dk/~aquak/MasterThesisKristofRoemisch.pdf

the numbers are a bit confusing as that is not the core of the technology. Obvioulsy it cannot be unlimited unique detail. The key is in replication. Once the object is loaded in memory it can be replicated without overhead. So in the OP movie he could have a billion of these trees and still have top fps. The points are ideally streamed from harddrive (thus requiring massive storage space) and only the points in the frustrum are displayed (loaded into gpu ram) at set distances the detail degrades (moves up the octree).

Nvidia vid from a year ago:

lpfaFrazOn4

lol this unlimited detail vid has footage of arma2 :)

JWujsO2V2IA

Edited August 2, 2011 by Soul_Assassin

[celery 8]

Minecraft creator Notch expresses his opinion:

It's a scam!

Edit: 2slow :p

Edited August 2, 2011 by Celery

[demonized 20]

beginning construction of my "holo deck", and getting my symbiotic holo impressionating tool together.

sceptically optimistic, as far as i can see from these videoes, they are just 3d Max or Maya rendered videos.

the PC technology have been stagnant for a few years, i am waiting for the next leap, though this one might not be it.

[dmarkwick 261]

Im sorry m8 but you are wrong

I don't see anything in my post that is wrong??? I mention the probability of instancing, which is what you mentioned also?

[sabre4809 0]

I'm sceptical at best along with almost everyone, but good/ground breaking developments, more often than not, are picked apart by nay-sayers. Its a vital part of development in anything, because it highlights limitations or shortfalls that can then be overcome.

Thats what innovation is, taking a huge idea and making it possible, rather than taking what is already done/easy and making it better, the latter is safe but very limited. If someone isn't picking at problems in your idea then its not good enough.

Also, everyone is saying how its just voxels, but in the old video they stated that while what they are doing is a point-cloud system, it is not voxels in the old sense and pointed out that old traditional voxel systems are very bad for what is needed in realtime graphics.

They seem to say that the kicker is the 'search engine' they use or summint, can't remember but it was something along those lines.

Surely if it worked out things like grass VD in ARMA would be a none issue, LODs, etc

Ah, we'll see. The Aus Government has some faith in them.

Edited August 2, 2011 by SAbre4809

[jagheterjan 10]

By the time they release some sandbox app and a SDK to toy around with this might become interesting, until then it's barely more than static noise.

[soul_assassin 1723]

I don't see anything in my post that is wrong??? I mention the probability of instancing, which is what you mentioned also?

I highlight :)

Indeed :) and I guess Carmack knows more than most about game engines. But is he familiar with the inner workings of the Euclideon engine? (See the voxel engine developed for RAGE) But, I would guess that a working title for this engine is 5 years away at least, so maybe he's good at hedging his bets :D

I do think though that Notch is being rather disingenuous with his assumptions:

That sounds like an attempt at baffling us with numbers. That would be like saying the ArmA2 Chernarus map cannot possibly have 1,000,000 rocks, trees, bushes & grasses in it because there's not enough room in memory to hold that info. Quite obviously there is optimisation, instancing (or whatever the analogous procedure is called) and reduction going on, why would he even make the ridiculous assumption that a 1/64th cubic millimetre detail is being rendered at 1km?

As I said, it's disingenuous at best. At worst it's a spoiling tactic aimed at clueless people.

The point isnt about rendering. His point is that yes you can instance a tree and a rock but if you are going to create a unique terrain model all those unique points will have to be stored and streamed from somewhere. An area of any substantial size would need too much hard drive space. I have a pointcloud of a city section (in voxel format) of total area that would be useless for gaming, 2 or 3 streets (with incomplete scan information i.e. shadowing, angle sub-sampling) and it takes up about 80 gigs of HDD space. Also without an SSD the streaming would struggle to keep up with so much data. So you see that guy is speaking some truth. He should know :) Minecraft is probably the most famous unknown voxel engine :)

[maturin 12]

Just like early 3D, it looks like crap.

[dmarkwick 261]

I highlight :)

Ah, I see, using a competitor's findings to judge a newcomer's results :) I'm going to go out on a limb here and say that Carmack is probably going to be right, and that I'm probably going to be over-optimistic. But, I've seen many, many times where one expert says something cannot be done and then someone simply does it. One thing I do have in common with Carmack is that neither of us has seen more than that video demo as far as Euclideon tech goes..

The point isnt about rendering. His point is that yes you can instance a tree and a rock but if you are going to create a unique terrain model all those unique points will have to be stored and streamed from somewhere. An area of any substantial size would need too much hard drive space. I have a pointcloud of a city section (in voxel format) of total area that would be useless for gaming, 2 or 3 streets (with incomplete scan information i.e. shadowing, angle sub-sampling) and it takes up about 80 gigs of HDD space. Also without an SSD the streaming would struggle to keep up with so much data. So you see that guy is speaking some truth. He should know :) Minecraft is probably the most famous unknown voxel engine :)

I'm just going to wait & see what, if anything, happens. I'm one of those people who believe obstacles can be overcome by ingenuity. Most of my posts reflect this, I try not to actively nay-say. Unless a thing is clearly impossible :)

[-)rStrangelove 0]

I really dunno what to make of all this tbh. They called out about the gfx revolution in 2003 already and still the only thing they have to show is a video full of promises.

Unlimited detail in 2003:

1. 2003 - 2011 is a long time so why aren't they in a position where they can offer demos for download?

2. Why are they still such a small team if the tech is so impressive? Students & others must be knocking on their doors every day.

3. They know they're lacking art talent in their team, so why don't they work with a team of game developers? It can't be the lack of outside interest after showing these videos for years can it?

Forum post in 2010:

I humbly apologize that I am not going to discuss UD further here, though I would certainly love to answer each and every one of your questions.

Look out for future demonstrations including art from real artists, moving lights and objects, animation, lovely colors, unique geometry (not just thousands of the same rocks/trees/weird-animal-thing). Hands on demos at future trade shows, eventually an SDK to play with yourself.

This is just the beginning. Please be patient. :)

Main website atm:

This site is currently unavailable

Hmmmmm ... i guess the revolution will take more time.

Edited August 3, 2011 by ])rStrangelove

[dmarkwick 261]

Well there's plenty of about-to-hit-the-consumer tech out there, most of them stuck in development hell. Three I can think of right off the top of my head:

Infinity Engine.

Outerra.

Euclideon.

Edited August 3, 2011 by DMarkwick

[Sniperwolf572 758]

Another post from Notch about this.

[PRiME 1]

Can't stop laughing at the MANY yelling Voxel VOXEL, old shit, crap. Guys get educated, it is similar to sparse voxel but it is not a Voxel based engine and has the potential to be much better.

My main concern really is clearly Animation/deformation, but also the size of datasets, they mention allot of procedural stuff used so I guess if a game like Infinity and procedurally make the universe (no joke) then I suppose they could do it with this system as well and ad procedurally generated variations to things so its not so much like a repetition all the time.

Also why people listen to COMPETING companies negative input is beyond my understand. Every man and his dog whom stands to loose out over this new engine idea is going to try hard to deflate the bubble, if UD are FOS then time will reveal that.

For a engine tech company making something truly ground breaking 5-20years development is probably expected, look at infinity engine. Also there seems to be allot of people whom do not understand the word of procedural replication, ell well, too bad. Back to the hole in ground with those lot.

Edited August 4, 2011 by PRiME

[Cookieeater 10]

Indeed :) and I guess Carmack knows more than most about game engines. But is he familiar with the inner workings of the Euclideon engine? But, I would guess that a working title for this engine is 5 years away at least, so maybe he's good at hedging his bets :D

I do think though that Notch is being rather disingenuous with his assumptions:

That sounds like an attempt at baffling us with numbers. That would be like saying the ArmA2 Chernarus map cannot possibly have 1,000,000 rocks, trees, bushes & grasses in it because there's not enough room in memory to hold that info. Quite obviously there is optimisation, instancing (or whatever the analogous procedure is called) and reduction going on, why would he even make the ridiculous assumption that a 1/64th cubic millimetre detail is being rendered at 1km?

As I said, it's disingenuous at best. At worst it's a spoiling tactic aimed at clueless people.

There's a really large chance that Unlimited Detail is using Sparse Voxel Octrees which means that they haven't invented anything new. Sparse Voxel Octrees:

+Unlimited geometry detail

+Automatic level of detail, all models resize dynamically in detail(No more popping trees in ArmA II)

-Limited by memory space

-Animation is very inefficient, animation techniques currently break level of detail optimization

Unlimited Detail seems to have both of these positives and negatives

Here's a very good read on how Sparse Voxel Octrees work:

gNZtx3ijjpo

1sfWYUgxGBE

xUe4ofdz5oI

[Minoza 11]

You guys watched new demo/interview with some explanations?

Edited August 24, 2011 by Minoza

[max power 21]

I think a less staged video would have worked better.

[dm 9]

I cant force myself to watch it, his voice is like nails on a chalk board to me...