Unlimited Detail, octree traversals

While I have no idea about DICE it's been reported that Euclideon presented their tech to game developers in the past. That only means the companies were intrigued enough to hear their pitch
It's more than that. There was an agreement between Euclideon and DICE. DICE would use UD in their next game, should UD be ready.
Isn't Battlefield 4 "the" next DICE's game? As for UD, it's quite ready.
What, in the above, is incorrect, repi?
 
Sometimes it's pretty hard not to get a presentation on something even if you're absolutely sure you're not interested.
 
It's more than that. There was an agreement between Euclideon and DICE. DICE would use UD in their next game, should UD be ready.
Isn't Battlefield 4 "the" next DICE's game? As for UD, it's quite ready.
What, in the above, is incorrect, repi?

I think the key in that agreement, if it exists, is ready. As in, all the inherent issues with the technology resolved to the satisfaction of the developer and sufficient evidence shown to prove so.
 
repi
More constructively, are the octree traversals discussed in the other thread related to UD?
 
It's more than that. There was an agreement between Euclideon and DICE. DICE would use UD in their next game, should UD be ready.
Isn't Battlefield 4 "the" next DICE's game? As for UD, it's quite ready.
What, in the above, is incorrect, repi?

repi will of course not discuss any future plans of DICE's tech beyond what has already been made public. I'm a bit surprised you even got a "bullshit" from him.
 
This most recent line of discussion has been terminated. An unwillingness to discuss his POV, presumably because he was completely wrong and unwilling to admit it, means Dag has no place on this board.

Maybe in a couple of years when Unlimited Detail is running on POWER7 based Wii U's and the like, all these arrogant preachers who never bother to answer simple questions that challenge their viewpoint can come back and feed me crow?
 
It disappeared from the game industry. It was revealed as a real-time game tech but ended up being an offline render tech. UD has gone the same way. I kinda feel a lot of these new techs don't appreciate gaming requirements, and any revolutionary tech not born from gaming ( like megatexturing) is going to take 10+ years to have direct inclusion in games, although ideas will filter down.
 
Isn't this completely pointless now because of tessellation?

Is it?
Doesnt this allow even more geometry density than what tessellation can allow? What about memory consumption? ;)
This is supposed to work on hardware that doesnt necessarily support tessellation either. Right? So that is probably a simpler but very good solution.
 
Is it?
Doesnt this allow even more geometry density than what tessellation can allow? What about memory consumption? ;)
This is supposed to work on hardware that doesn't necessarily support tessellation either. Right? So that is probably a simpler but very good solution.

Tessellation can get silly, EVE demo said to pushing 500 million triangles being generated with tessellation.

While UD might allow tessellation like detail on machine that doesn't support it don't forget you'll need a beefy system even run the thing, tessellation units are here and even in low end cards they're still extremely powerful.
 
The rendering technique is somewhat orthogonal to the issues of data storage and retrieval. Megameshes would accomplish the same thing as UD, but the use of the word 'unlimited' in the name gives the Euclideon tech a grandiosity that ignores the constraints of acquiring and storing that much detail. If UD was offering incredible data compression then it'd have something to offer, but as it is, it's a technique to access very large datasets very quickly with certain access constraints, coupled with a rendering system that matches the data format (as I understand it). Concepts like dirt in very high fidelity need either lots of dirt models, or lots of repeated dirt granules in UD. The same visual result can be applied with displacement on the ground based on distance. There's an example comparing a UD tree to a Crysis 2 tree showcasing UD's superiority. Well, durrent displacement-map trees are every bit the equal of that UD example, and use of LOD applying successive (megatextured?) maps could add detail to whatever level UD could manage too. The limiting factor would be cost to produce the assets and speed at which data can be accessed. Without a compression advantage, UD doesn't offer and advantage in the latter. Perhaps it offers an advantage in the former if you can just scan and use, but as the renderer is so limited, that's hardly of value to a game.
 
Tessellation can get silly, EVE demo said to pushing 500 million triangles being generated with tessellation.

While UD might allow tessellation like detail on machine that doesn't support it don't forget you'll need a beefy system even run the thing, tessellation units are here and even in low end cards they're still extremely powerful.

But I am not referring to UD. I am referring to CNCAddict's link :)
Is it the same UD technology?
 
Isn't this completely pointless now because of tessellation?
Tesselation has still a lot to evolve before it can be universally used.
IE.
Dependency of base mesh tesselation, factor of 64 is not enough.
Also proper surface and view dependant tesselation is needed.
 
Not enough how? Factor 64 still gives extremely high amounts of polys unless you go back to 1990s-level base models...
 
You need tessellated assets to tessellate them...
(Obviously you'll still amplify, but you need to have a finer grained mesh to begin with.)
 
Back
Top