[PS3] Uncharted 2

I'm kind of curious to know how they do post processing on the SPU. I'm sure this is answered somewhere. Pardon my weak understanding of the graphics pipeline, but post-processing is done after rasterization, correct? Wouldn't that mean copying data back and forth from VRAM to main memory, since them segmented?
Probably no back-and-forth. The ideal way (i.e. minimal bandwidth) would be to transfer the framebuffer(s) from GDDR to SPE local store (in several chunks) and spit them back out to GDDR so it can be scanned out.

Post processing can refer to any screen-space operation that takes place after all objects have been rendered, i.e. after the frame is "complete", but before it is presentable. Tone mapping is the traditional example, but depth-of-field effects, MRT integration (deferred lighting schemes, certain types of motion blur) and aa resovle can all be shoved under that same umbrella.
 
I'm kind of curious to know how they do post processing on the SPU. I'm sure this is answered somewhere. Pardon my weak understanding of the graphics pipeline, but post-processing is done after rasterization, correct? Wouldn't that mean copying data back and forth from VRAM to main memory, since them segmented?

SPU Postprocessing was first revealed to be implemented in a version of PSSG. In case of Killzone 2, which does something similar, they keep a much lower res version of framebuffer (and possible some motion and depth data or may simply be gbuffer) on the main memory, and do SPU postprocessing there.

Since most of the post processing involves blur and blur variants (DOF, motion blur etc) lower resolution is very natural choice. When they are combining the upscaled postprocessing buffer with actual 720p framebuffer, they know which pixels are affected (based on motion vectors or depth) so for the unaffected pixels the lower resolution post processing has no effect, and for the rest the nature of blur hides the lower resolution.

edit:To clarify, upscaled postprocessing buffer and native 720p buffer are mixed via per-pixel weighted averaging, for which weights are derived from stuff like motion vectors and depths.
 
Last edited by a moderator:
SPU Postprocessing was first revealed to be implemented in a version of PSSG. In case of Killzone 2, which does something similar, they keep a much lower res version of framebuffer (and possible some motion and depth data or may simply be gbuffer) on the main memory, and do SPU postprocessing there.

Since most of the post processing involves blur and blur variants (DOF, motion blur etc) lower resolution is very natural choice. When they are combining the upscaled postprocessing buffer with actual 720p framebuffer, they know which pixels are affected (based on motion vectors or depth) so for the unaffected pixels the lower resolution post processing has no effect, and for the rest the nature of blur hides the lower resolution.

edit:To clarify, upscaled postprocessing buffer and native 720p buffer are mixed via per-pixel weighted averaging, for which weights are derived from stuff like motion vectors and depths.

The lower resolution framebuffer is only a lower resolution framebuffer because they don't have enough bandwidth for doing some of the effects though right? You can do a proper 720p framebuffer in XDR memory. I think (but I"m not sure) that this can even be a proper framebuffer that you can display from directly. But that's something I read ages ago, so someone may have to set me straight on that one.
 
The lower resolution framebuffer is only a lower resolution framebuffer because they don't have enough bandwidth for doing some of the effects though right?
Well, ideally "amount of effects" should not affect bandwidth because once you fetch those lines to localstore, you can do multiple effects in serial (assuming all of them fits in ~half localstore or you can cascade them using multiple spus without having to read from main mem again). I think it's more about SPU resources.
Of course lower resolution means lower bandwidth for both CELL and RSX independent of amount of effects, in addition to more SPU muscle per pixel which determines the amount of effects. Does that make sense?
You can do a proper 720p framebuffer in XDR memory.
You can do 720p SPU postprocessing at ridiculously high framerates, especially if you are doing simple stuff like 3x3 blur. But I guess you mean from RSX, in which case yes too. In case of Killzone 2, I seriously doubt they render at lower resolution though since they have to render higher resolution too. It's simply downscaled framebuffer.

I think (but I"m not sure) that this can even be a proper framebuffer that you can display from directly. But that's something I read ages ago, so someone may have to set me straight on that one.
I remember that too.

Also T.B. should be around but I think Sacred 2 divides the buffer into 2 tiles for SPU processing.
 
I like the part that playing SP gives u points to spend in MP too. I like that approach. Borderlands is using that approach too, but as co-op. But its good that devs are thinking this way now.
I always wait till I have completed the campaign before I even try MP. One reason being that I don't want all the levels that will be revealed in the story by looking at the maps of MP. That is a big bummer for me. So, later on what happens is that I reach MP a month or two late and everybody is all suited up while I am running around in my underpants. I think that my devotion to SP should be rewarded, and looks like it will be in U2 ! The virtual cash I earn in SP will be useable to buy upgrades in MP from get go ! :) Love you ND !

I think this is required, as no player should feel that his preferences were neglected by the game devs. ND is going that extra mile that makes a person a fan of a studio rather than just that game. Insomniac, ND and Sucker Punch, I love u this Gen :cool: ! (Don't worry Kojima, I love u too, but I want Peace walker on the PS3 !)
 
Yes, it's really a good thing to "encourage" SP play, I'm very disappointed buy the numbers of people who play only MP game and no SP.
 
Uncharted 2 got a cinema mode but with no export? I find that's strange, why not a youtube export?
I'm not speaking about the play game, that you can export, but about the editing tools…

http://arstechnica.com/gaming/news/2009/08/exploring-online-play-film-making-with-uncharted-2.ars

This is a really good idea, The only reason I'm understood the no export option is for limiting these tools to people who really want made something well made.

Yes, it's a shame. Writing a high quality video exporter is not so trivial and probably need to pay additional royalties ? They can spend the resources on game development instead. Youtube would be a good distribution mechanism though.
 
Cinema Mode/Replay Interview (5:23)

The lip-synch sounds really cool.

Wow at the lip-synch machinima. Can we control Elena, Chloe and Sully also ?

Thanks for posting the video. I understand what they meant by Cinema Mode now. Did they record the in-game events and use the engine to replay them ?
 
I know I've had my doubts about Uncharted 2 MP, but that was very impressive.

The average player probably won't make much use of the more advanced stuff, but still some cool things come from this.
 
Wow thats really some crazy stuff.

I just wished UC2's MP was a bit more fast paced on shooting [like Gears] with even weapon placements, that would help it to rise in competitive levels.
 
Finally finished Uncharted, so I'm ready for #2. Pretty good game, but didn't like the finale. That seems to be a common thing with me. For some reason games don't have final bosses like they used to. The weird thing is I'd put the game down for months and I was literally 30 minutes from the end of the game. If I'd known that I probably would have finished it a lot sooner.
 
Wow thats really some crazy stuff.

I just wished UC2's MP was a bit more fast paced on shooting [like Gears] with even weapon placements, that would help it to rise in competitive levels.

what? I always thought U2 beta was much faster paced than Gears........
 
Back
Top