So, we've all probably heard about the (in)famous Unlimited Detail rendering system that debuted a few months ago, developed by a small Australian company called Euclideon. In case you haven't, http://www.youtube.com/watch?v=00gAbgBu8R4 they put out about it.

When they first appeared in the public spotlight, they were harshly criticized by many well known game developers, most notably Notch, who had http://notch.tumblr.com/post/8386977075/its-a-scam about them. Other entities like http://www.gameinformer.com/b/features/archive/2011/11/22/exploring-unlimited-detail.aspx firmly believe that this is legitimate and could vastly change the way that the system works.

Going by information listed in that Game Informer article, this engine is supposed to not only create photorealistic environments, but do so without needing a GPU whatsoever. Naturally, this would mean a huge paradigm shift in the world of gaming, if we are no longer tied to the hardware.

This is a now-dated debate, but upon reading more coverage about it, rather than just Notch's professional opinion, I have become more intrigued in the matter. It's been a little while, so I figured I'd ask again. Do you think that Euclidion's claims hold any water, or are they just full of hot air? If this is a legitimate technology, how long before it becomes feasible?

Now about Notch's claim:

Whose to say that the island wast hollow? What would be the purpose of storing the information of a thing you will never see?

Also, why would the "atoms" the engine uses need to be the size of a real atom? At the detail most games are 1 "atom" could = 2.5 mm and the game would still have amazing detAil.

I'll wait until I see moving grass on a demo.

Also don't listen to Notch.

Aren't we going to move back to software-rendering CPUs at some point anyway? I'm pretty sure that's what John Carmack has said, in observations on hardware trends and where things are going.

This obvious scam again?

Valid point imho, brings it down to a "more manageable" 15TB of data. More likely something above that.


I hate to break it to you, but there are more than 64 atoms per mm^3. Far far more.

Like Notch said (Say what you want about the man, but he does know his voxel technology) it's easy to render static meshes with this technology but there is no way to render animated objects and large scenes with many different objects

I can't really take the guy on the video seriously, he sounds ridiculous.

Also, the part where they "imported"a rock made me laugh, just seems a strange turn of phrase.

So for people who know far more about this than me, is this complete BS or feasible?

Euclideon claims they can animate. They claim a lot of things, but are keeping things secret. I wouldn't jump to any conclusions until they truly reveal what they have or admit defeat and disappear.

I'm not taking any sides. But http://www.youtube.com/watch?v=hxtuZE5pOGA is a link to a 40 min interview with the Euclideon guy and includes a live demo with the interviewer.

The way I see it, if it's real - awesome. If not - I don't lose anything.

Its a scam. They had an old vid saying studf would be ready in a few moths, now tht time was long ago, still nothing.
I also hate how some folks think its a scam just becuase notch says it is. Yes it is, but damnit notch's word isnt the word of god.

I'm sure "atom" is just a term to describe how small they are, because ya'know, atoms are pretty small.


This is how I see this whole ordeal

Careful....Notch's army of 13 year olds has eyes everywhere.....we dont want these forums coming under attack for heretical thoughts...... :unsure2:


Also, there was an interesting http://www.abc.net.au/tv/goodgame/video/default.htm?src=http://forums.bethsoft.com/tv/goodgame/video/xml/20111108_2030.xml&item=05 on this on a gaming show I watch called Good Game. It looked very pretty, but I'm unconvinced as of yet that its a viable technology or if it's truly capable of what the developers are claiming.

There's a method of data processing that's been around for a long time -used in real world technologies (including sonar, if I remember correctly)- for dealing with large quantities of stuff, which is relatively low on processing requirements. So it's not completely implausible.

Remember, they're probably not using every single vertex/possible polygon at the same time. I expect they do a search for the most appropriate ones and use them, which means they can have things get more detailed when you get closer without having to replace the entire object with something new. Still, this is a lot of information to handle... but if the above mentioned methodology is applicable, it may actually be possible.

'Course, I'll maintain reservations until I've seen it in an actual game.

So because it might be taking longer than expected, it's a hoax? That's pretty horrible logic..

Well, The Elder Scrolls IV: Oblivion was a hoax, so it's understandable that people wouldn't want to get burned again.

Oblivion was Oblivion because it was rushed. So maybe this will turn out like Skyrim instead.

The only way I could see this working is that it renders only the needed 'atoms', which is redundant because it would need to render the frame in some way to figure out which atoms are needed. Even if it's real, it'd be useless.

If this technology works the way I think it does then animations require entire prerendered frames like a .gif animation.

Here is a tool that demonstrates that. Though much less detailed...

http://xhp.xwis.net/downloads/VXL_Editor.zip

EDIT - Apparently this version only edits existing voxels. So you'll need to download a Red Alert 2 or Tiberiun Sun mod (one with vehicles) to play around with the voxel technology concept.
http://www.cncgames.com/modseditors.shtml

Indeed.

If this technology was so grand, they really should release a demo.

Also, if this technology was really so great and "revolutionary", why haven't other major studios attempted it?

Not completely implausible, no. But in the various demo videos, we see it running on a low-end looking laptop. Even if it is running only on the CPU, let's assume a very conservative 5TB of data representing the island. Compress that by 75% and we get 1.5TB worth of data. That's still too much for (most) hard-drives to store, let alone easily decompress on the fly, load into RAM, etc. And 75% compression ratio is being very, very generous as well.

Representing a static island made of of "atoms" isn't the issue - computers are designed to run millions of operations per second. It's the issue of storing all that data, and not just in the hard-drive, but in RAM as well. Animations pose yet another issue - even if you use something like keyframe animation, that's still a massive amount of data that needs to be processed. Let alone collision detection with other objects.

It would be a waste of time when they can make a profitable game the traditional way.


Compression is easy when a lot of the objects are clones. Nobody really pays attention to how the dirt and leaves are different. Just offer a little variety, and just like real life, they're not really noticeable unless you look for it.

Remember the age old proverb: "If it sounds too good to be true, it probably is"

The sheer number of clones as shown in the video would not work in an actual video game. There's -maybe- three trees. Possibly just one tree. It doesn't really matter how great one tree looks - it's still one tree.

A Demo would be the easiest way to prove to the world that this is real. I'm asking myself the same question.

As for why other studios don't use it, that's actually logical to explain. This is a breakthrough technology, if it is real, and would require a huge amount of investment to utilize. You'd need to essentially experiment forever to see what it can actually do for your games, train all of your developers and artists to adapt to this new system, get it to work on existing technology, and find a publisher who would be willing to fund it all. Above all that, it's not even a finished product, so we're still using polygonal games for now.