Support NeoGAF

[benny_a]

Now, question: Should this even matter if nobody noticed in the first place? I'm sure there will be some people who now probably say they noticed this all along, or knew something was up, but come on, that's not really true now, is it? Nobody's going to believe that.

Nobody needs to believe anything. We have a post history and the "graphics" have been discussed in the context of multiplayer in KZSF threads and people were rightfully calling it downgraded compared to the singleplayer but came to the wrong conclusion what method was employed.

 

[HTupolev]

Now, question: Should this even matter if nobody noticed in the first place? I'm sure there will be some people who now probably say they noticed this all along, or knew something was up, but come on, that's not really true now, is it? Nobody's going to believe that.

What?

People have been complaining about the MP being all wonky since this machine launched.

If our local 1080 paparazzi truly thought that something was up, or even worth investigating with the game's resolution, a pixel count would have been done ages ago.

...Or had anyone done a pixel count, they would have found the resolution to be 1920x1080, because in terms of unique pixels represented within a particular frame that's exactly what it is.

This is a bit beyond the scope of usual pixel counting methods.

We now have an example with Killzone that's very close to a 720p resolution for the MPp

Qualitatively speaking, it looks astronomically different from 720p.

It really reminded me of how Halo Reach looked when things were in motion, although the blur appears more precise in KZ.

Interesting.

Reach actually uses temporal AA, but it shuts off for things detected to be in motion. Aside from the occasional miss causing ghosting artifacts, Reach's motion sloppiness is mostly sketchy motion blur blending at work.

 

[GoofsterStud]

Just shows 1080p is not a big deal at all. If the lower resolution was that noticeable this would have been figured out the day after release. 60 to 30 frame rate is noticeable instantly. 1080p to 720p ain't shit.

Where were you when we were saying that Single Player is more clear?

 

[Codeblew]

It is if you're playing it on a 1080p TV.

You realize that KZ3 and KZSF are two different games on two generationally different platforms, right?

You are right it is upscaled. My beef with upscaling is when it is an uneven upscale and the scaler has to make up information. This is an even upscale: 960x2 = 1920. So it doesn't have to make up missing pixels, it simply doubles the horizonal pixels. Whereas 720/900 upscaled to 1080p has to calculate the inbetween/shared pixels when upscaling which causes blur.

 

[EventHorizon]

Not really.

It might be worth noting that we already have an example of an XB1 game using temporal supersampling: Ryse.

True, but Ryse's final image isn't 1920 x 1080 either. It is 1600 x 900. So while it is true that the XB1 can use this technique, it doesn't appear that it can do it to produce a true 1080p image.

X0 can render and read from/to DDR3, there is no problem. (Unlike X360 in which you had to write to edram and couldn't read from it.)

I am no expert on graphic rendering so anyone feel free to correct me, but I believe the problem for the XB1 is that it's DDR3 memory isn't fast enough to be used as a frame buffer. So it's not a matter of if it can be done, but if it can be done quick enough. The way I understand it is that the frame buffer is generally rendered in ESRAM, although it does have the ability to partially render the frame in main memory. So once a frame was done, it'd have to be copied to the DDR3 main memory in order to save it. Then it has to be read back in order to calculate the next frame.

So my question is, is that extra read and write to the slower DDR3 too much to handle?

 

[coolasj19]

Reposted for new page because I agree with it. Why'd Senjutsu get banned? =/

That would not be correct, as "anamorphic" specifically means the image is stretched for final display, and The Order is not. A correct description would be "native 2.4:1 AR in a1080p frame".

"Killzone: Shadow Fall's multiplayer runs at 960x1080 temporally reprojected to 1080p, targeted at 60fps."

I know it's a mouthful; I know they'd have to explain what all that means. But that's their responsibility. The multiplayer isn't native 1080p, and just because it's simpler to say that doesn't make it less false.

Heck, the complexity--and apparent uniqueness--of this technical approach could even have been explained as a positive: "The technical wizards at Guerilla have come up with a new frame-by-frame rendering method that's better than traditional upscaling. It maintains most of the gorgeous sharpness and robust detail of the native 1080p single player campaign, while still allowing them to target 60fps during the most intense 24-player online combat!" This approach to explanation has the benefit of being true.

Nobody needs to believe anything. We have a post history and the "graphics" have been discussed in the context of multiplayer in KZSF threads and people were rightfully calling it downgraded compared to the singleplayer but came to the wrong conclusion what method was employed.

It seems a lot of the uproar here particularly, is because people assumed the graphics of the Multiplayer were the way they were because of Guerrilla's usage of AA. Because they said the MP was "Native 1080p". More interestingly to me at this point, is Leadbetter's casual dropping of this knowledge. As if everybody already knew this as fact. The article this came from was on a completely different, but relevant, topic. This is enough for an entire analysis all on it's own.

From a dev perspective, I want to learn the algorithm employed by GG to interpolate the frames. I'm not a graphics programmer but I'm interested on the approach and logic they took to make the game look good given its resolution.

Maybe when this is blown over to some capacity, you and Paz can go double team Guerrilla's engineers for us.

 

[Zephyx]

From a dev perspective, I want to learn the algorithm employed by GG to interpolate the frames. I'm not a graphics programmer but I'm interested on the approach and logic they took to make the game look good given its resolution.

 

[Shao Kahn Brewing a Stew]

From a dev perspective, I want to learn the algorithm employed by GG to interpolate the frames. I'm not a graphics programmer but I'm interested on the approach and logic they took to make the game look good given its resolution.

It's a very smart loophole if you want to call it that. Not sure if I want to commend them or not. They fooled everyone because we didn't have any numbers, but a lot of people know something was off about the multiplayer.

Regardless, this goes to know that if a developer makes a game they want to, it'll look amazing in the end if they put the right thought to it.

Of course, same cannot be said about multiplatform games that are usually gimped to be in one system and isnt the original vision they went in with.

Fix this shit, Guerilla?

 

[HTupolev]

You are right it is upscaled. My beef with upscaling is when it is an uneven upscale and the scaler has to make up information. This is an even upscale: 960x2 = 1920. So it doesn't have to make up missing pixels, it simply doubles the horizonal pixels. Whereas 720/900 upscaled to 1080p has to calculate the inbetween/shared pixels when upscaling which causes blur.

KZSF's multiplayer is usually a totally different approach, and isn't upscaled at all.

True, but Ryse's final image isn't 1920 x 1080 either. It is 1600 x 900. So while it is true that the XB1 can use this technique, it doesn't appear that it can do it to produce a true 1080p image.

Ryse's temporal supersampling takes place over the entire 1600x900 buffer. And I wouldn't be surprised at all if the game is overall doing more work for reprojection than KZSF MP.

Even if they had to bleed half a 1080p buffer into DDR3, that wouldn't be that big of a deal. A single pass over a 960x1080 buffer isn't likely to be very catastrophic for a 68GB/s memory pool.

 

[sono]

I get confused reading this thread. I think I am concluding it is 1080p/60fps in output to the tv, what is the debate is that how those frames are rendered to the frame buffer. Did I get it right ?

My other question is how can you reliably work arrive at that conclusion by looking at the output, or do these guys have access to the frame buffer some how ?

I have a couple of questions, when did resolution become a such a big deal? Did anyone really expect $400-500 consoles to output 1080p/60fps on a consistent basis give the specs of the consoles?

Resolution and frame rate are everything for this gen, this is the main differentiator from last gen that we waited 8 years for.

Price has nothing to to do with the question at hand.

 

[see5harp]

Now that I think about it, I clearly remember the IQ debate that spurred a comparison between Ryse's AA solution and SF's. People pinned the blame on the FXAA+TMAA combo solution utilized by Killzone as the culprit for blurry IQ.

http://www.neogaf.com/forum/showthread.php?t=709298

You need glasses. Go and watch the high bitrate footage.

It's absolutely nothing even resembling the word "blurry".

lols

 

[HTupolev]

I get confused reading this thread. I think I am concluding it is 1080p/60fps in output to the tv, what is the debate is that how those frames are rendered to the frame buffer. Did I get it right ?

Yes. It's all a question of how things get handled internally.

 

[RedAssedApe]

tl;dr

:O what'd he get banned for?

 

[Grinchy]

I wish this much discussion was taking place about how unplayably laggy the game is for US players. The word on how broken the MP is needs to get around so GG finally does something about it.

 

[Codeblew]

KZSF's multiplayer is usually a totally different approach, and isn't upscaled at all.

It is if you're playing it on a 1080p TV.

You realize that KZ3 and KZSF are two different games on two generationally different platforms, right?

I thought you were implying that it was upscaled and I was just admitting that it was in regards to doubling the horizontal pixels. It is kind of a grey area.

 

[justsomeguy]

Three major things jump out to me:

1) we were lead to believe that kzsfmp is native 1080p/constant 60 (i believe they said except when lots of grenades go off at once) - we were misled either intentionally or by omission

2) DF didn't pick this up? Seriously?

3) (as per the other thread) Resolution =/= IQ

 

[HTupolev]

I thought you were implying that it was upscaled and I was just admitting that it was in regards to doubling the horizontal pixels. It is kind of a grey area.

I was implying that KZ3 is upscaled, because KZ3 is a 720p game on Playstation 3.

KZSF MP isn't upscaled, and does not use horizontal pixel doubling. It generates a new 960x1080 field every 1/60th of a second; the missing data is not gained by upscaling, but by sampling information from the previous frame.
This isn't a grey area at all. It is not upscaling.

 

[pottuvoi]

I am no expert on graphic rendering so anyone feel free to correct me, but I believe the problem for the XB1 is that it's DDR3 memory isn't fast enough to be used as a frame buffer. So it's not a matter of if it can be done, but if it can be done quick enough. The way I understand it is that the frame buffer is generally rendered in ESRAM, although it does have the ability to partially render the frame in main memory. So once a frame was done, it'd have to be copied to the DDR3 main memory in order to save it. Then it has to be read back in order to calculate the next frame.

So my question is, is that extra read and write to the slower DDR3 too much to handle?

DDR3 has ~20GB/s more memory bandwidth than Ps3 had for whole system, so it should be enough for 'some' heavy lifting.

Also If one wants to use ESRAM to render the buffer and then copy it to DDR3 for later use, nothing is stopping you.

 

[Liabe Brave]

...Or had anyone done a pixel count, they would have found the resolution to be 1920x1080, because in terms of unique pixels represented within a particular frame that's exactly what it is.

This is a bit beyond the scope of usual pixel counting methods.

Actually, in this thread some GAF screenshots were posted, and my pixel counting did in fact show the 960x1080 resolution. However, these were on very small samples (Shadow Fall's geometry is complex enough that long straight lines are rare), and on player models. Those will generally be moving horizontally, so I think the displacement algorithm goes for efficiency and doesn't really add much extra detail. (Logically, it has to be doing this some of the time; if it always added back missing stairsteps then the image wouldn't be perceived as blurry!)

But, there was that shot which did count as 1920x1080, despite having combing artifacts here and there indicating that the temporal reprojection was running. So some of the time--essentially low-speed/still camera and objects--pixel counting would have problems.

 

[GameAddict411]

The upscaling technic is very smart, apart from the bad motion blur it actually not that bad. It also explains why people didn't notice it all this time. Though honesty considering the variable performance they should have made it a locked 30 FPS experience with a true native resolution.

 

[Unknown Soldier]

2) DF didn't pick this up? Seriously?

DF is proving this generation that they literally are no better than forum posters on GAF at picking things up. Their evaluation of Thief was so faulty as to essentially be worthless to anyone trying to decide on version of the game based on technical merits. DF pretends to be experts but in reality they seem to know much less than some posters here in GAF and they are more biased too.

 

[Kuro]

Now, question: Should this even matter if nobody noticed in the first place? I'm sure there will be some people who now probably say they noticed this all along, or knew something was up, but come on, that's not really true now, is it? Nobody's going to believe that.

If our local 1080 paparazzi truly thought that something was up, or even worth investigating with the game's resolution, a pixel count would have been done ages ago. That didn't happen, regardless of what people may be saying currently. So, what it all boils down to is that this is just one more example of a game (if this is all accurate and no mistakes were made) in which something that was supposed to be so effortlessly spotted and identified, was missed completely by just about everybody. I distinctly remember posts of people mocking developers and posters who sometimes said that it was tough to notice the difference between certain resolutions, and that it wasn't always as easy to spot as people claimed it was.

We now have an example with Killzone that's very close to a 720p resolution for the MP, an MP that many people, even myself, were impressed with visually, and nobody seemed to notice anything amiss with the resolution despite this. That just about tells you all you really need to know.

Nope, the guy is right. Everybody missed this. Since when has a developer telling people a game was 1080p stopped people from doing a pixel count anyway? I think I'm only being fair in saying that if the game was an Xbox One title, a resolution analysis would have almost certainly taken place no matter what the developer claimed, and I think everybody knows that. But the best part here is that if people thought the game's MP looked great before (lots of people did) then nothing should be different about today.

Ah and here he is

 

[Toparaman]

Good stuff.

 

[Fallout-NL]

Crazy no one noticed and he didn't test the MP when it came out.

Oh I fucking noticed. But people said I was wrong.


Here:

Odd, honestly, it looks nothing like that (to me, should I add that quantifier?) when I play multiplayer. The visuals take a pretty massive hit coming from singleplayer, to the point where I thought it had a lower resolution even.

Though looking at those shots, I might be somehow mistaken (because yes, these look good). Are they direct captures from someone running that game on an actual PS4?

Weird, I'm weirded out.

Haha, and it was dark10x himself who said I was wrong.

 

[HTupolev]

Actually, in this thread some GAF screenshots were posted, and my pixel counting did in fact show the 960x1080 resolution. However, these were on very small samples (Shadow Fall's geometry is complex enough that long straight lines are rare), and on player models. Those will generally be moving horizontally, so I think the displacement algorithm goes for efficiency and doesn't really add much extra detail. (Logically, it has to be doing this some of the time; if it always added back missing stairsteps then the image wouldn't be perceived as blurry!)

But, there was that shot which did count as 1920x1080, despite having combing artifacts here and there indicating that the temporal reprojection was running. So some of the time--essentially low-speed/still camera and objects--pixel counting would have problems.

Yeah, I guess it depends on where you're looking and what's going on.

This post on B3D shows the rather interesting behaviour, where on occasion it seems like it might be falling back to an upscale (?).

But overall frames do tend to show something that basically looks like 1920x1080. That's especially true when you're standing still, which is usually not a bad idea for pixel-counting images.

 

[sant]

I only played multiplayer and single player briefly but MP did look a little blurry. I thought I was imagining things

 

[T.O.P]

The res upscaling and interlacing is super obvious here. Like damn.

Since the interlacing occurs temporally, the artifacts appear some times as ghost dithered stuff from the previous frame. That must look really confusing in motion.

Jesus Christ

 

[coldfoot]

I was implying that KZ3 is upscaled, because KZ3 is a 720p game on Playstation 3.

KZSF MP isn't upscaled, and does not use horizontal pixel doubling. It generates a new 960x1080 field every 1/60th of a second; the missing data is not gained by upscaling, but by sampling information from the previous frame.
This isn't a grey area at all. It is not upscaling.

Well, it's temporal upscaling, generating an interlaced image and then deinterlacing it is definitely not rendering at native.
To be honest, they should have either just went with 720p60 or 1080p30 for MP instead of this

 

[Zephyx]

It's a very smart loophole if you want to call it that. Not sure if I want to commend them or not. They fooled everyone because we didn't have any numbers, but a lot of people know something was off about the multiplayer.

Regardless, this goes to know that if a developer makes a game they want to, it'll look amazing in the end if they put the right thought to it.

Of course, same cannot be said about multiplatform games that are usually gimped to be in one system and isnt the original vision they went in with.

Fix this shit, Guerilla?

It is a smart loophole indeed. Given the time constraint and pressure during launch, I can see they opted to do it as their last resort if they were pressured to push their game to 60fps when it's not their original vision. The only negative thing I can make out of this is keeping this under wraps for so long when people have a method to figure it out eventually. Being a dev is hard.

Maybe when this is blown over to some capacity, you and Paz can go double team Guerrilla's engineers for us.

I'm not a game developer but I work with back-end software a lot. If I ever venture into game development/engine coding, the algo by GG should come handy at some point. Algorithm development is fun.

 

[GameAddict411]

DDR3 has ~20GB/s more memory bandwidth than Ps3 had for whole system, so it should be enough for 'some' heavy lifting.

Also If one wants to use ESRAM to render the buffer and then copy it to DDR3 for later use, nothing is stopping you.

Its not that easy, the DDR3 ram speed is a huge bottleneck, so is the eSRAM amount. Comparing it to the PS3 memory total bandwidth is silly considering that the XO GPU is 8.5x more powerful and its going to need significantly more bandwidth to saturate the GPU.

 

[coolasj19]

DF is proving this generation that they literally are no better than forum posters on GAF at picking things up. Their evaluation of Thief was so faulty as to essentially be worthless to anyone trying to decide on version of the game based on technical merits. DF pretends to be experts but in reality they seem to know much less than some posters here in GAF and they are more biased too.

Oh, DF (Leadbetter) totally knew about this. When, or how, he knew about it is unclear. But the information we are talking about right now was just casually dropped in this article about choices between 1080p@30fps and 720p@60fps. The real question is how the F' did he not think this was newsworthy? It's seriously goddamn baffling.

In the single-player mode, the game runs at full 1080p with an unlocked frame-rate (though a 30fps cap has been introduced as an option in a recent patch), but it's a different story altogether with multiplayer. Here Guerrilla Games has opted for a 960x1080 framebuffer, in pursuit of a 60fps refresh. Across a range of clips, we see the game handing in a 50fps average on multiplayer. It makes a palpable difference, but it's probably not the sort of boost you might expect from halving fill-rate.

Now, there are some mitigating factors here. Shadow Fall uses a horizontal interlace, with every other column of pixels generated using a temporal upscale - in effect, information from previously rendered frames is used to plug the gaps. The fact that few have actually noticed that any upscale at all is in place speaks to its quality, and we can almost certainly assume that this effect is not cheap from a computational perspective. However, at the same time it also confirms that a massive reduction in fill-rate isn't a guaranteed dead cert for hitting 60fps. Indeed, Shadow Fall multiplayer has a noticeably variable frame-rate - even though the fill-rate gain and the temporal upscale are likely to give back and take away fixed amounts of GPU time. Whatever is stopping Killzone from reaching 60fps isn't down to pixel fill-rate, and based on what we learned from our trip to Amsterdam last year, we're pretty confident it's not the CPU in this case either.

Either Leadbetter doesn't know what article topics get clicks, didn't want to expose Guerrilla, didn't feel like doing the article/interview... actually. I have no idea why he didn't use this information sooner and in a different context than an opinion piece.

 

[Zen]

That's a really impressive technique, kudos to Guerilla, my respect for their tech skills grows every time we hear something new.

DF is proving this generation that they literally are no better than forum posters on GAF at picking things up. Their evaluation of Thief was so faulty as to essentially be worthless to anyone trying to decide on version of the game based on technical merits. DF pretends to be experts but in reality they seem to know much less than some posters here in GAF and they are more biased too.

They're actually worse. Their articles are only useful for Gaf to have a good starting point to pick apart all the things they get wrong.

 

[jimi_dini]

Seriously DF has been missing alot of stuff lately.

Shouldn't it have been detected pretty easily? You just need 2 frames right after each other in motion and then look at them.

 

[Valnen]

The res upscaling and interlacing is super obvious here. Like damn.

Since the interlacing occurs temporally, the artifacts appear some times as ghost dithered stuff from the previous frame. That must look really confusing in motion.

Damn, that shot looks like a PS3 game. Man...launch titles.

 

[Deleted member 125677]

Where's your 1080p GOD now, Shahid???

I hope this is more a launch period problem than a hardware limit problem. It's so weird to me that they can't maintain 1080p/60fps with the specs they have

 

[Kuro]

Where's your 1080p GOD now, Shahid???

I hope this is more a launch period problem than a hardware limit problem. It's so weird to me that they can't maintain 1080p/60fps with the specs they have

Look at Resistance 1 and then look at Killzone 3.

 

[Codeblew]

I was implying that KZ3 is upscaled, because KZ3 is a 720p game on Playstation 3.

KZSF MP isn't upscaled, and does not use horizontal pixel doubling. It generates a new 960x1080 field every 1/60th of a second; the missing data is not gained by upscaling, but by sampling information from the previous frame.
This isn't a grey area at all. It is not upscaling.

Who was talking about KZ3? Sure KZSF frame buffer may be 1080p but I wouldn't call it native 1080p. It is only rendering 960/1080 per frame and blurring in the previous frame.

 

[Durante]

Too bad I have no interest in this game, if I ever looked at it I might have noticed something
That screenshot Dictator93 posted is pretty terrible.

Yes, but only for stuff with high parallax; the concertina further on doesn't have them. Overall it's an interesting choice. There's that ghosting on thin, fast-moving objects, and textures are blurred, but edge aliasing is reduced.

That's pretty normal for any analytical single-sample AA technique. Except that here you get ghosting on subpixel aliasing in addition to the usual artifacts.

And that different information is not just the same pixels from the last frame. If it was, there would be vertical combing artifacts on every single object in motion, not just thin stuff that's crossing quickly, as seen above. So there's some algorithm in place which tries to interpolate things in an intelligent way (and fails on certain stuff in the ~20ms window it has).

All in all, there's no existing jargon to describe what's going on. The only accurate approach is a full description, like "It's 960x1080 temporally reprojected to 1080p." I know that's wordy and unsatisfying, but it's the best way. Talking about interlacing or upscaling is just not correct. (Though it's hard to get away from those words; I believe I may have used "interlaced" somewhere earlier in the thread myself.)

I'd describe it as an analytical AA technique with 1/2 a spatial plus 1/2 a temporal sample per pixel.

 

[HTupolev]

Well, it's temporal upscaling, generating an interlaced image and then deinterlacing it is definitely not rendering at native.

I hesitate to call it upscaling, because upscaling is usually applied to cases where the intermediate data is interpolated from known data. That's not what this temporal process is doing; it's producing the intermediate data by finding something that might very closely approximate it in an ever-so-slightly old buffer.

Its not that easy, the DDR3 ram speed is a huge bottleneck, so is the eSRAM amount. Comparing it to the PS3 memory total bandwidth is silly considering that the XO GPU is 8.5x more powerful and its going to need significantly more bandwidth to saturate the GPU.

Regardless of "what it takes to saturate the GPU" or "where the bottleneck lies", the point remains: it's an extremely small strain on a 68GB/s pool to do a single pass over a half-res "1080p" buffer 60 times per second.

Who was talking about KZ3?

You were (inadvertently I think), on the last page, lol.

 

[nelsonroyale]

Damn, that shot looks like a PS3 game. Man...launch titles.

Haha, play the game. SP visuals are pretty incredible. Also, that is a rather crap shot. It would be easy to get something that looks far better than that...He was using it to demonstrate the artefacts creating from the rendering procedure I guess.

 

[Gestault]

DF is proving this generation that they literally are no better than forum posters on GAF at picking things up. Their evaluation of Thief was so faulty as to essentially be worthless to anyone trying to decide on version of the game based on technical merits. DF pretends to be experts but in reality they seem to know much less than some posters here in GAF and they are more biased too.

Is this really the conclusion you're coming to when DF was the only place to put out this information with hard numbers and researched the technique, our resident pixel-counter missed it completely and is only now claiming he can tell it's not native 1080p, and DF not catching this in their original report would run contradictory to the bias you're assuming?

 

[Gestault]

Oh I fucking noticed. But people said I was wrong.


Here:



Haha, and it was dark10x himself who said I was wrong.

That's funny and frustrating all at the same time.

 

[boredandlazy]

Ah and here he is

If only he'd put as much effort into reading others' posts as he does in writing his own.

 

[Sendou]

Look at Resistance 1 and then look at Killzone 3.

But they were on PS3. I'm not an expert but everyone praises PS4 for being very easy machine to develop to. In other hand PS3 was a very hard machine to work with. Wouldn't it make sense the gap between early and late games on PS4 is smaller than it was on PS3?

On topic I don't think this is such a huge deal but why did they need to go out their way and lie about this?

 

[TGO]

Never really played the MP but Damn, no noticed, No one?
And DF did a technical on it previously too? WTF?

 

[MilkBeard]

It isn't 1080i, because a single frame of 1080i is 1920x540. They basically took the principle of 1080i and flipped it on its side, literally.

It's Witchcraft, that's what it is.

 

[Jack cw]

Never really played the MP but Damn, no noticed, No one?
And DF did a technical on it previously too? WTF?

Everybody that played the MP noticed the worse IQ compared to the SP.
And no, DF didnt even bother investigating why this has been the case.

 

[hodgy100]

Good man.

So for all intents and purposes, the output from the ps4 to your tv/monitor is 1080p.

So basically, nobody lied, they just didn't explain the methodologies at work.

Makes this thread all the more hilarious.

lets not beat around the bush here, it is absolutely not 1080p by the definition of the term. Sony and Geurilla claimed it was, it wasn't so much a lie as a bending of the truth, which is something we aren't giving Microsoft a pass for with their "the output is still 1080p", referring to up scaling. So we shouldn't give Sony a pass in this case.

 

[coldfoot]

I hesitate to call it upscaling, because upscaling is usually applied to cases where the intermediate data is interpolated from known data. That's not what this temporal process is doing; it's producing the intermediate data by finding something that might very closely approximate it in an ever-so-slightly old buffer.

You might say it's "extrapolating" in this case...similar to how old school sophisticated "line doublers" worked. Still it's generating pixels without going through the traditional rendering pipeline, so maybe not upscaling, but definitely "cheating".

Oh and ElTorro needs to make the Banderas gif for this thread where the final banderas is vertically interlaced.

 

[Liabe Brave]

...our resident pixel-counter missed it completely and is only now claiming he can tell it's not native 1080p...

Though plenty of other GAFfers are capable pixel counters, I believe you're referring to me. I'm not infallible, but you can quit with the character assassination. I couldn't tell before because in some situations this technique will defeat pixel counting. I can now tell, given new screenshots, but even with some of those the true resolution is not apparent.

I hope you can forgive me for not detecting an interpolation method that has apparently never been used before, just as I will forgive you for being a jerk in the name of winning an argument.