Last of Us [PS4]

My take: the trailer runs in real-time, but they played it in slow motion n order to be able to average a bunch of frames and get some very high quality motion-blur out of it (aka they used an accumulation buffer as Polyphone did for this http://www.youtube.com/watch?v=_vZR7LNJqg8).

They could get even better DOF, anti-aliasing and shadows out of this technique, simple because they haven't done it it doesn't mean anything.

Saying it was rendered with on a single PS3 or 10 or more doesn't make any difference in terms of the final trailer looks like.
 
My take: the trailer runs in real-time, but they played it in slow motion n order to be able to average a bunch of frames and get some very high quality motion-blur out of it (aka they used an accumulation buffer as Polyphone did for this http://www.youtube.com/watch?v=_vZR7LNJqg8).

They could get even better DOF, anti-aliasing and shadows out of this technique, simple because they haven't done it it doesn't mean anything.

Saying it was rendered with on a single PS3 or 10 or more doesn't make any difference in terms of the final trailer looks like.

So, just to be clear for layman like me, you keep render frames, add them to the accumulation buffer and then push the image from the accumulation buffer to the screen? So you'd only display 1 accumulated frame for every x frames, meaning real-time at full speed would require 30x frames if you wanted to target 30Hz? If this is what they've done to produce the video the real-time quality should still be comparable with reasonable compromises to dof, motion blur, shadow filtering and AA?

Also, does it make sense to implement an accumulation buffer in a rendering pipeline for the purposes of creating a trailer? Is this buffer typically the same size as your forward buffer, or whatever you call the buffer that gets pushed to the screen?
 
Why do it real-time tho?
Why not just pre-render it like the Uncharted games?

Is it like unchy, no more. Don't let me wrong, it's undeniable what ND can achieve with the ps3 hardware but honestly I continue to ignore the purpose to claim something prerecorded directly in an hardware... to me tech wise it's pointless. I hope to see something more in game soon than later.
 
So, just to be clear for layman like me, you keep render frames, add them to the accumulation buffer and then push the image from the accumulation buffer to the screen? So you'd only display 1 accumulated frame for every x frames, meaning real-time at full speed would require 30x frames if you wanted to target 30Hz? If this is what they've done to produce the video the real-time quality should still be comparable with reasonable compromises to dof, motion blur, shadow filtering and AA?
Create accumlation buffer with higher numerical accuracy IE.fp32.
Render frames with like you normally do, but jitter time, camera/light/etc position.
Copy each finished frame into an accumlation buffer using additive blending.
Average the data in accumlation buffer with amount of frames you copied into it.
tonemap (if using hdr.)

So you have only one additional buffer and no need to store more frames, thus it's quite easy and memory efficent way to do AA/DoF and so on.
Shift polygons in viewspace within a pixel = AA,
Camera move within camera lense = DoF
time = motion blur.
Lightsource jittering = area lightsource & shadows..
etc.
Also, does it make sense to implement an accumulation buffer in a rendering pipeline for the purposes of creating a trailer? Is this buffer typically the same size as your forward buffer, or whatever you call the buffer that gets pushed to the screen?
Size of buffer is same except bitdepth. (ie 1280x720xfp32)
Fun fact, Accumlation buffer was called T-buffer by 3dfx.
 
The GT3 intro was rendered offline as well.

There was an obvious difference in image quality.

We dont know what they have "touched up" over the actual real time capabilities of the PS3 to render this.

So until we actually see the in game footage, my excitement has been reduced.
 
It's not that far fetched to see a fungal infection that radically alters your physical appearance.

tree_man.jpg


Though that just needed some surgery and Vitamin A to start to fix it.

This was a viral infection. Acommon strain of warts virus to be precise. The thing with that guy that made him unique was that his immune system was peculiarly susceptible to the virus and his body was unable to fight it off.
 
They're not zombies! It's not a zombie game.
True, they're people that got some virus infection and turned to monsters, but they're not zombies. Sure, some more originality in enemies would be welcome in games and movies, how long have they been enemy soldiers, alien monsters or zombies... hmmm.. I don't think I can even list any other types of enemies in action games and movies :)
Anyway, really looking forward to this, and even more so now that it's known Naughty Dog are the developers. Who needs next gen yet if the game sfor this gen can look as good as UC3 and what The Last of Us seems to be.

Fungal infection... not viral ;)

Also, re the realtime/offline/in-engine debate... in Naughty Dog's tweet they did say that it was "in-engine, like Uncharted cutscenes".

So i think that makes things much clearer ;-)
 
Create accumlation buffer with higher numerical accuracy IE.fp32.
Render frames with like you normally do, but jitter time, camera/light/etc position.
Copy each finished frame into an accumlation buffer using additive blending.
Average the data in accumlation buffer with amount of frames you copied into it.
tonemap (if using hdr.)

So you have only one additional buffer and no need to store more frames, thus it's quite easy and memory efficent way to do AA/DoF and so on.
Shift polygons in viewspace within a pixel = AA,
Camera move within camera lense = DoF
time = motion blur.
Lightsource jittering = area lightsource & shadows..
etc.

Size of buffer is same except bitdepth. (ie 1280x720xfp32)
Fun fact, Accumlation buffer was called T-buffer by 3dfx.

So the performance reasons for not using accumulation buffers is the additive blend plus the requirement to render more frames for the purposes of "averaging"?
 
The GT3 intro was rendered offline as well.

There was an obvious difference in image quality.

We dont know what they have "touched up" over the actual real time capabilities of the PS3 to render this.

So until we actually see the in game footage, my excitement has been reduced.

From what I understand, it's a cutscene rendered with the game-engine with game-assets realtime. :-/ So it's not beyond PS3's capabilities to do it.
For the most part you won't have the camera that close to the characthers all the time, so the game won't look like that most of the time, but there isn't many devs who manages camera-control as good as Naughty Dog.
When the camera zooms in on Nathan Drake, and company ingame, like when in close quarters crawling through tight spaces, it look just as good as the cutscenes, I'm pretty sure that this game will do that aswell. :-/

I don't think it's a horror-game, it's a survival action-game. :)
But it's their first M-rated game. I don't really care much for what genre it is, as long as it's a good game, and it's not too scary for me to play. :p
I've seen videos from a game like Amnesia, and it looks really atmospheric and fun, but there is no way I'm gonna play it. :-/
 
This is very, very impressive stuff here.

From modeling perspective ? I love the eye movement and the overall look.

First impression is: Gamers' focus will probably be on the teenage girl and the fungus people as a distant second ("They are just zombies. La la la"). Hope the game focuses less on gunplay but more on character development and variety of gameplay. Basically whatever works for the situation.

Also hope it shows other surviving human and animals, kinda like Resistance 3. Plus the relationship between the 2 characters, like Ico.
 
It's too early to tell how the entire game will look.

I think animation in the Last of Us trailer stands out. The eye movement, facial expression, posture and body language all look natural/life-like. The action and camera work are also more dynamic (for both subtle/slower and abrupt/fast-paced scenes).

The people in Heavy Rain looked less consistent and sometimes more rigid. I remember the most convincing character is the grocery store owner. Gameplay-wise, I sure hope Last of Us can be as immersive and engaging as Heavy Rain.

EDIT:
NaughtyDog interview:
 
Dunno if its that much better, shaders are little better and lighting is much better, but modeling dunno. Heavy Rains has really impressive faces.

I'm sorry but those images only go to support my opinion. Some of those faces are simply bad looking. Melted facial features, blurry textures, less polygons and so on...
 
I'm sorry but those images only go to support my opinion. Some of those faces are simply bad looking. Melted facial features, blurry textures, less polygons and so on...

Seriously?

Granted, they aren't CG movie quality, but surely the best quality characters seen this gen ...

Unless I missed something ... :???:
 
It says, "...Directly Captured From A Playstation 3". That's singular.

Exactly. I just don't believe anyone at ND is devious or cynical enough to include such a bold statement at the start of the trailer and have it anything other than a realtime capture from a single PS3.
 
Seriously?

Yes, that is my professional opinion. The truth is that Heavy Rain assets had a very mixed quality level and the mostly mocap facial animation was quite bad too. I never really understood the praise it received for its characters.

UC2 and 3 characters were already more detailed and better executed, and Last of Us is even better than those games were with a more realistic art style. I could list a lot of other games as well but let's not make it a vs. thread.
 
Back
Top