L.A. Noire from Rockstar

Avatar = the new Toy Story. Has anyone managed to match Toy Story in a computer game yet? It's been ten years now since those initial claims....
 
ROTFL

Avatar's quality is not based on tech, but on the sheer number of man-months poured into the CG characters and their animation. No video game can afford that amount of resources, especially for as much animation as a game has compared to a 2-hour movie.

Have you watched this video? It seems face capture technology was quite important here, and it does combine face and body capture.

http://news.discovery.com/videos/tech-avatar-motion-capture-mirrors-emotions.html
 
ROTFL

Avatar's quality is not based on tech, but on the sheer number of man-months poured into the CG characters and their animation. No video game can afford that amount of resources, especially for as much animation as a game has compared to a 2-hour movie.

Err, why not? Off the cuff I imagine some game budgets are getting pretty big...such as LA Noire for one. Wasnt Killzone 2 rumored at 100 million or something? Sure GTA 4 was really high too. Just saying, not an order of magnitude difference there.
 
Avatar = the new Toy Story. Has anyone managed to match Toy Story in a computer game yet? It's been ten years now since those initial claims....

To be fair though, offline graphics rendering versus 'online', and motion capture (which in the end is both offline) aren't quite the same type of comparisons. We've seen plenty of examples showing that games here do not have to lag much behind movies.

However, the graphics that the animation is rigged to still being rendered in real-time is of course going to hold back the visual impact, which is where comparisons with Avatar may still be dangerous. But taking it purely from a motion capture tech standpoint, there's merit to Cage's remark I think, which is purely intended to refer to the quality of fully body and face motion capture.
 
Well, in some things we're quire close (it's time to watch TS1 again!) but there's stuff like general image and shadow quality or pure scene complexity that's still just a dream for realtime.

All in all though, with normal mapping and current rendering tech, I think that now we can get pretty close to the 1995 level of rendering shiny plastic toys :p
 
Have you watched this video? It seems face capture technology was quite important here, and it does combine face and body capture.

Didn't watch that certain video yet but I've learned as much as I could about Avatar's tech and approaches. Capture was an important part in the performances, but there was both a lot of manual keyframe animation on top of it and a huge amount of work in getting characters that looked and deformed realistically. Muscle simulations, thousands of blendshapes for the facial animation, polygon counts of insanely high numbers and so on. All that stuff was man-made and quite work intensive, and that is where a game can not compete with a movie's budget.
 
Err, why not? Off the cuff I imagine some game budgets are getting pretty big...such as LA Noire for one. Wasnt Killzone 2 rumored at 100 million or something? Sure GTA 4 was really high too. Just saying, not an order of magnitude difference there.

Avatar spent like $300 million on about 2.5 hours of final footage. KZ2 was like 50-60 million at most and only a fraction of that went into the cinematics and the assets used in them. And even that was extremely high for a game.

Avatar had dozens of people working on the facial rigs and animation, there's years of research into the technology, and no expenses were spared. They are also not limited by a realtime environment, a single head is like 25 thousand quad polygons (50k triangles) and uses thousands of blend shapes to create proper, lifelike deformations. Texture resolutions I can't even guess, probably multiple sets of 4K textures per face (several color layers just for the subsurface scattering shader, specular, roughness, reflection, displacement...)

I stand by my statement, Avatar levels are unreachable even for X3/PS4 systems, maybe even for the PS5 level. And by that time we'll have Avatar 2 or Avatar 3.
 
But taking it purely from a motion capture tech standpoint, there's merit to Cage's remark I think, which is purely intended to refer to the quality of fully body and face motion capture.

Well, for the body capture, Naughty Dog seems to have a studio that's at least comparable, somewhere above 100 cameras in the system for both Avatar's virtual studio and the stage they've used for UC3 (not sure if they own it or Sony).
For facial capture, UC3 uses none, only video reference; Avatar used a SD face camera and some pretty heavy custom image analysis software to track markers and translate them into facial expression data. And still it only got them 90% of the way and the rest took a LOT of manual animation to get right.

Avatar had maybe an hour or two of facial animation in the end, a game like Heavy Rain needs dozens of hours, especially because it has to cover multiple paths for each conversation. Weta could manually tweak everything to the slightest detail, Quantic will need a fully automatic solution and see if they can even afford to manually touch every sentence just once.
 
Ok, that's fair. It may help though that like LA Noire, Quantic likely will limit themselves to actual faces of their actors, which may help as then perhaps more can be automatically detected/mapped. Obviously, Cage also didn't say they were at Avatar quality now, but thought they could be in a few years, and I don't think that is totally unrealistic, if, as you indicate, you make a bit of a rough comparison overlooking some of the 'details'. :p

UC3 would definitely benefit from facial capture. It's holding them back a little now - when I look at UC3 footage, it is becoming its weakest point. You can also really tell that in its animation system, there is a lot of interactive logic going on, blending various animations at once and reacting to the environment (e.g. standing with one foot on a rock or an enemy) and improving response time from controller to movement. That makes the game stand-out in terms of interactivity with the environment and the player, but will always hold it back some versus motion capture, at least until the human body has been fully simulated and everything mapped to its constraints (which will require Drake to make slightly less outrageous jumps in the game here and there ... ;) )
 
Avatar = the new Toy Story. Has anyone managed to match Toy Story in a computer game yet? It's been ten years now since those initial claims....

I thought Ratchet and Clank TOD's cutscenes came pretty close imo, LBP2 is even closer. As for Avatar, that Samaritan footage from Epic did a very good job of closing the gap. Of course in terms of animation, CGI will always be better since people can manually tweak the hell out of them and no algorithms can match hand tweaking..........at least not yet.
 
I thought Ratchet and Clank TOD's cutscenes came pretty close imo, LBP2 is even closer.
How many visible polygons and pixelated shadows did you see in Toy Story? Including Slinky's spiral? As a general 'look', yes, we could say certain benchmarks have been reached, but if we are serious about making comparisons, games are still a long way from CG quality and it's silly for the games industry to keep making such impractical comparisons. Faultless objects in a complex world with varied textures still doesn't exist. The memory and processing requirements are still beyond what's capable. It's worth noting that TS was rendered over months on about 300 100MHz SPARCs, rendering one frame every 45 minutes. To get that to 30 fps for a game, you need an increase of 81,000x. That may be possible in the latest GPUs although RAM issue could be very limiting, but not in any of the current consoles. And certainly not in the PS2 or XBox or any other device that's claimed equivalency. So basically it's a silly claim! We still haven't got Toy Story quality in a game such that it looks like a CG movie. We may be using the same sorts of techniques, but scaled down in terms of complexity, image quality, and variety.
 
How many visible polygons and pixelated shadows did you see in Toy Story? Including Slinky's spiral? As a general 'look', yes, we could say certain benchmarks have been reached, but if we are serious about making comparisons, games are still a long way from CG quality and it's silly for the games industry to keep making such impractical comparisons. Faultless objects in a complex world with varied textures still doesn't exist. The memory and processing requirements are still beyond what's capable. It's worth noting that TS was rendered over months on about 300 100MHz SPARCs, rendering one frame every 45 minutes. To get that to 30 fps for a game, you need an increase of 81,000x. That may be possible in the latest GPUs although RAM issue could be very limiting, but not in any of the current consoles. And certainly not in the PS2 or XBox or any other device that's claimed equivalency. So basically it's a silly claim! We still haven't got Toy Story quality in a game such that it looks like a CG movie. We may be using the same sorts of techniques, but scaled down in terms of complexity, image quality, and variety.

I agree that in terms of absolute polycount, AA, shadow res and etc there's no comparison. Of course when I say they came in close I only meant in look and feel though I believe we should have reached TS quality sometime ago on the latest PC hardware with techniques like tessellation and dynamic radiosity. It may still not be a 1:1 replica but I believe we are somewhere there, aside from animation.
 
How many visible polygons and pixelated shadows did you see in Toy Story? Including Slinky's spiral? As a general 'look', yes, we could say certain benchmarks have been reached, but if we are serious about making comparisons, games are still a long way from CG quality and it's silly for the games industry to keep making such impractical comparisons. Faultless objects in a complex world with varied textures still doesn't exist. The memory and processing requirements are still beyond what's capable. It's worth noting that TS was rendered over months on about 300 100MHz SPARCs, rendering one frame every 45 minutes. To get that to 30 fps for a game, you need an increase of 81,000x. That may be possible in the latest GPUs although RAM issue could be very limiting, but not in any of the current consoles. And certainly not in the PS2 or XBox or any other device that's claimed equivalency. So basically it's a silly claim! We still haven't got Toy Story quality in a game such that it looks like a CG movie. We may be using the same sorts of techniques, but scaled down in terms of complexity, image quality, and variety.
Probably when the developers said that, talks only of general 'look' , the most of gamers are not interested of real numbers.... I mean, imho, toy story has passed in this generation from the first gears of war... obviously talking only of the look...
 
I'll get back to the issue of facial performance capture tomorrow; in short it's far from a simple case...
 
New is that it takes up 3 discs - so it's sort of a proof that they're really streaming all the facial performance from storage, frame by frame.
 
Money ready and waiting. Think i want to see some review scores first though.

Yeah but I love all things LA, so I'm getting it :) Dunno if this has been linked yet:

http://www.latimesinteractive.com/Standard/landing_pages/lanoire/

It shows locations and info on the crimes that inspired the game, pretty neat. Can't wait for this title.


New is that it takes up 3 discs - so it's sort of a proof that they're really streaming all the facial performance from storage, frame by frame.

Seems like it, sounds like there are two disc swaps in the entire game, so presumably each disc holds the face data and parts of the world needed for the cases on that particular part of the game.
 
OK, sorry if this has been answered - but how 'open world' is this game? I read that you can walk around but no 'real' interaction - but then I read elsewhere that you can do side-missions...does anyone know!?
 
Back
Top