The "Unlimited Detail" Guys Are Back

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
Remember these guys? Well, they are back and they have a new video! Then again, that's all they had last time too. Here's to hoping they don't disappear again for another year. Thanks to Kaos_Drem for the link.
 
going to need a lot of ram to run in high detail without repetition like in the demo

very exciting though
 
note that there's no animation.

such is probably the crushing fault.
 
That looks pretty cool to me. I wonder if it can be applied to moving elements in the game? Character models, and what the physics code would look like to blow some of that stuff apart.
 
I am impressed... but I won't be excited until some artists show me what it can really do.
 
Very cool. It'll be interesting to see if this came live up to the massive hype the guy is giving it.
 
Anyone else think that the video was spoiled by the guys voice?

Yeah. Maybe I'm immature, but I f'in lol'd at the top comment, it matches the guys voice perfectly:

"WE NO LONGER MODEL DICKS BY HAND. NO MORE DICK POLYGONS.

WE SCAN THEM IN. EVERY SINGLE PENIS ATOM IS RENDERED.

TRUE DICK TECHNOLOGY."
 
note that there's no animation.

such is probably the crushing fault.

Calculating collisions is probably pretty fun too.

They sure are good at the attention whoring though, they should sell ad space in their pre-rendered fly throughs.
 
note that there's no animation.

such is probably the crushing fault.

That's the firs thing I noticed as well. It looks like an entirely pre-rendered environment. Like the kind of environments you find in point and click games like Riven. I mean really, go back and play Riven. It's over a decade old and still one of the best looking games in existence - because all of the environments are all pre-rendered. The only time things look like shit is when animations start-up.

That being said, the environment itself doesn't actually look anymore impressive than the Unigine Demo. Sadly no game in existence has implemented DX11 to the same extent that the Unigine engine does. Although there is an interesting RTS/Tower Defense hybrid game coming out in the next few months from Unigine. I'm definitely looking forward to see how that turns out.
 
How does this relate to ID's tech 5?

Doesn't their engine do something that allows unique textures with hogging resources?

Would these engines be viable in conjunction? At least the technologies?
 
shit still looks janky as hell. Not feeling the straight edges at all. Damn, round some corners out d00ds and add in some players and weapons. Anything.
 
i'm for this, everything new is strange at first, it only must be adopted, or fail.
 
The only way i think this could work is if they mix this tech and polygon tech together.
 
If its bad for animations you can still use it for things like the ground and rocks/trees and buildings that aren't used(GTA 4 the surrounding city that looks like shit comes to my mind :D)

It seems pretty cool, and I think it work too. But probably needs more optimizations. This kinda seems like Gasoline versus Solar/other methods of fuel. Its better, but would require a lot of work and change and I don't think everyone is willing to do that. All depends though.
 
note that there's no animation.

such is probably the crushing fault.

Possibly. However the technology could be used for non-destructible objects and for the landscape itself, and then be combined with old-fashioned polygon models as has already been suggested. I think that would work well for flight/helicopter sims and open-landscape games (Elder Scrolls, Stalker etc.).
 
Anyone else think that the video was spoiled by the guys voice?

Glad I wasn't the only one. :eek:

If it wasn't for the fact that I wanted to hear his content delivery I would have muted it.
 
These guys seem to be pretty confident in their tech. Sure it doesn't include such things as animation and physics in the demo, but I'm sure they wouldn't just be sitting around with their thumbs up their asses in this regard.

Be as cynical as you like, I'm lookin forward to seeing what else they can do :)
 
I'm cynical as hell about this, as his environments are all repeated terrain and repeated blocks and sectors of land, with repeated tree's and other repeated objects

If the guy could release some HARD data about it, like, that elephant, what is the memory footprint of that elephant? On a PC that elephant would have a memory footprint of around 50-60KB in DX11, 70-80KB in DX9-10, what is this unlimited details version of a elephant make up?

Its easy to repeat items on screen, at that point it just takes up more polygons, not memory space, like L4D, if you have two zombies using the same model, they both take up the same sector of memory, like 90KB -TOTAL- for all zombies that use that model, you can repeat that model on screen a hundred times, and it would only affect polygon counts, not memory used, which saves on videocard memory

My question is, how much memory is that elephant eating up given its complex model system?

Also, notice how everything is only 1 or 2 colors, that's another big indicator that this system is very memory intensive
 
What have they done in a year? I'm guessing they used up all thier investor money and need more. The details they showed off were amazing. But what about the water? I hope they have flowing water in thier next demo.
 
Anyone else think that the video was spoiled by the guys voice?

I'm glad I'm not the only one. It sounds too good to be true and his voice gives the whole thing a notch to the "full of shit" side of the scale. I know it's just a voice, but damn...
 
I'm cynical as hell about this, as his environments are all repeated terrain and repeated blocks and sectors of land, with repeated tree's and other repeated objects

If the guy could release some HARD data about it, like, that elephant, what is the memory footprint of that elephant? On a PC that elephant would have a memory footprint of around 50-60KB in DX11, 70-80KB in DX9-10, what is this unlimited details version of a elephant make up?

Its easy to repeat items on screen, at that point it just takes up more polygons, not memory space, like L4D, if you have two zombies using the same model, they both take up the same sector of memory, like 90KB -TOTAL- for all zombies that use that model, you can repeat that model on screen a hundred times, and it would only affect polygon counts, not memory used, which saves on videocard memory

My question is, how much memory is that elephant eating up given its complex model system?

Also, notice how everything is only 1 or 2 colors, that's another big indicator that this system is very memory intensive

This is the age old computer debate about space vs time complexities. Basically this tech is pre-rendered (space) because it is taking a model from a 3D imaging program and converting it directly into something that looks identical although static. This is actually a good thing as memory capacities are doubling at about the same rate as CPU processing power is advancing. Also to note, GPU memory Bandwidth is going through the roof with things like 5Ghz GDDR5, so having a large point cloud makes perfect sense for the static elements in you game as they can quickly be read and drawn with little delay even if stored in the system's memory. As for animation, that could still be left to being rendered using polygons, and for physics a quick polygon based clipping box can be put around the objects for simple collisions.

The more tools a developer has to create a photo-realistic environment the better, now there is simple another tool to do a certain part of the scene better. It won't and can't replace everything and I saw the demo as a great example of making beautiful static scenery.

Go back and play FF9 and you'll see what I'm talking about. Square pushed the PS1 to its limit by using static backgrounds with video overlays and with nicely detailed character models. I'm really hoping this stuff comes out.
 
Bullshit meter 100%

I don't care what they wanna say. It looks great on paper and still. Make it functional and I'll pay attention, otherwise thats just a lot of lipstick, and the pig ain't getting any prettier
 
Watch and listen again. They aren't here to make a game engine with animations. They are making a tool for Artists to be able to use to make better graphics. They want to sell this to a big company so that they can use it and include animation.

Again, this isn't about animations but about graphics/textures.
 
...Also to note, GPU memory Bandwidth is going through the roof with things like 5Ghz GDDR5, so having a large point cloud makes perfect sense for the static elements in you game as they can quickly be read and drawn with little delay even if stored in the system's memory...

Just to clarify. I don't know what one would consider these "atoms" (if anything at all - assuming this isn't complete bullshit) but it's definitely no point cloud. You could not fit enough rendering power into a single desktop to render an entire environment, as well as NPC models ALL with point clouds.

I've worked with Point clouds at my job before. And they were incredibly small compared to these environments - just the size of an oil compressor you find in a car, and yet our computers ground to a halt. 2GB of vRAM, over 4GB of system ram...and the frame rates were like slideshows.

I don't know what's more impossible at the moment for future gaming. Real-time Ray Tracing, or Real-time Point cloud rendering. Both are so physically taxing on the hardware, that running them on anything but an entire server is an exercise in futility.
 
i-want-to-believe.jpg
 
So no polygons just "individual atoms" I'm guessing you get to choose the number of atoms which will end up being the limiting factor as far as detail goes. Instantly my mind shoots to VOXELS! Now some of you might say "what the fuck is a voxel" well the first time I uttered that phrase was when the developers of Masters of Orion 3 were talking about how they're doing their ship models... and needless to say that turned out to be poo-poo in looks. But I guess if you up the voxel count you can get some handy detail, however if this really works without you needing 2 terabytes of VRAM on your card they found a way to compress the data, perhaps an equation or something which encompasses quite a few of these voxels aka "individual atoms"... and we're right back at what a polygon is.

And yes that voice was horrible, seriously sounds like one of those "making fun of infomercials on youtube" type voices.
 
Back
Top