• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Digital Foundry: Final Fantasy 15: PS4 Pro Gameplay - First Look/Tech Analysis

https://www.youtube.com/watch?v=Ew0_5fF0WaM






giphy.gif


Digital foundery really needs to add CC on their videos man.
EDIT: wow ! it does have some---I'm late :)
edit#2: Wait it hardly makes any sense anyways. :(
 
Yes, if they can figure out what's causing it then it could be a quick fix that doesn't require extra hardware grunt to deal with. It's a bug, not a result of hardware not being up to the task.

Thanks for the answer. That actually makes me much more hopeful that they can patch this out,
 
Hmm... looks like another L for the ps4 pro to hold. Not surprising due to very weak CPU it's struggling to maintain steady fps. I wonder that CPU is not bottlenecking the GPU. But I suspect this will be story for pro version of games. Struggling to maintain fps. They should go for 1600p instead of 1800p. The hardware just isn't there IMO.
 

23qwerty

Member
Hmm... looks like another L for the ps4 pro to hold. Not surprising due to very weak CPU it's struggling to maintain steady fps. I wonder that CPU is not bottlenecking the GPU. But I suspect this will be story for pro version of games. Struggling to maintain fps. They should go for 1600p instead of 1800p. The hardware just isn't there IMO.

CPU is irrelevant to resolution, no?
 

HTupolev

Member
I still don't know what I'm supposed to be seeing when people say frame pacing issues.
Can anyone explain in detail what frame pacing issues are? I'd like to know more for what I'm looking for.
tbjbRTR.gif


Top square moves on a perfect 200ms timer. Bottom square usually does the same, but there's a 300ms frame in the middle followed by a 100ms frame, making it look like it freezes and skips.

Both top and bottom have ten unique frames over two seconds, so the framerate for both averages out to 5fps, but the top looks smoother because they're timed evenly.
 

Elios83

Member
Hmm... looks like another L for the ps4 pro to hold. Not surprising due to very weak CPU it's struggling to maintain steady fps. I wonder that CPU is not bottlenecking the GPU. But I suspect this will be story for pro version of games. Struggling to maintain fps. They should go for 1600p instead of 1800p. The hardware just isn't there IMO.

Frame pacing is a different thing from struggling to mantain fps.
Game is still 30fps, it is not dropping frames. It's the way the frames are presented that might cause some judder although how much you perceive that is totally subjective. Most people don't notice small frame drops let alone frame pacing issues.
The Pro is delivering lots of improvements here.
If Square can investigate the issue and fix it of course it will be better.
 

valkyre

Member
Hmm... looks like another L for the ps4 pro to hold. Not surprising due to very weak CPU it's struggling to maintain steady fps. I wonder that CPU is not bottlenecking the GPU. But I suspect this will be story for pro version of games. Struggling to maintain fps. They should go for 1600p instead of 1800p. The hardware just isn't there IMO.

The game is rock solid 30 fps....

You need to check out what framepacing is.
 

demigod

Member
Hey hey

In theory, at a higher framerate, frame pacing issues become less and less noticeable right? What would the cut-off point be?

The frame-pacing stuttering on the HQ side is so bad I thought the video encode was fucked.

I like how you both, especially jett, were calling out nxgamer in that other thread but are completely silent in this one. Can't admit you were wrong?

Nice to see DF also come to the same conclusion as NX. Maybe certain individuals will cut NX some slack...

I got you fam.
 
I agree the HDR is excellent on the pro. Also, I don't care if this game got frame pacing issues or not. Im still going to play for 100+ hours and enjoy it.

I got through FF9 battle loading times...I can play through some frame pacing issues.
 

Bebpo

Banned
Yeah, just tried the demo on Pro and HQ mode if they fix the framepacing will look great. But as it is, it feels like a 20fps AC Unity on PS4 type stuttering choppy mess. Totally unplayable. The LQ mode feels good, but there's no advantage over a regular PS4 so it makes the Pro kind of a waste. They need to either fix the framepacing in HQ or get the framerate to 60 in LQ so that there's some usefulness of the Pro.

I gotta wonder what's causing all the frame pacing that's been showing up this gen. Never really noticed it before Bloodborne and it's effected a bunch of games since. Like shouldn't the programmers, tech people making games be noticing and catching this stuff as they make the game? Is framepacing difficult to get rid of without sacrificing performance? The fact the HQ mode seems to have way more frame pacing than the LQ mode in this demo sure makes it seem like frame pacing is a performance issue like tearing since you'd think the HQ/LQ modes in the same game would be rendering the same way.
 

Shiz86

Member
DF, youre the only site on the net providing this form of analysis, at this volume and quality and what's astonishing is the time in which you deliver this. Thank you for your tireless effort to inform and educate us all. Id like to donate, just a one off sum, I don't want to sign up for recurring payments though.

If you like DF search out NX gamer. He actually beat DF to the FFXV demo analysis confirming the frame pacing issues and 1800p.
 

Izuna

Banned
I like how you both, especially jett, were calling out nxgamer in that other thread but are completely silent in this one. Can't admit you were wrong?

Jett, and to some extent myself, were saying how it didn't look too good. You know that checkerboard 1800p, and according to DF here it's low-quality effects, are causing the poorer IQ.

I think I can speak for Jett when I say, it doesn't what it's supposed to be if it looks blurry.

Now as for myself, of course, I'd be the first to admit I have a bias against NXGamer ~ and for that, I'll eat as much crow as possible. My disbelief was only amplified by the fact that the screens looked rough. Which, of course, I still stand by.

Titanfall 1 is 792p on Xbox One and has FAR better IQ than the FFXV demo. So I don't think it's checkerboard-ness that's causing it. I'm happy to hear DF to analyse what the problem is.

tl;dr Jett isn't wrong for having his opinions. And the fact that DF here has to explain why there's so much softness should tell you that this is not what the average person should expect from 1800p.
 
At 2:55, when the camera is panning up, doesn't it look like the framerate/pacing of the Lightweight mode is noticeably worse than HQ mode?

It looks stuttery to me. The opposite is clear a 3:55 for example, though.
 

NXGamer

Member
Nice to see DF also come to the same conclusion as NX. Maybe certain individuals will cut NX some slack...

Thanks man, I do feel some "bias" at times in criticsm but always happy to have discussion and opinion on things ;-)

I like how you both, especially jett, were calling out nxgamer in that other thread but are completely silent in this one. Can't admit you were wrong?



I got you fam.

Cheers fam! Respek!

If you like DF search out NX gamer. He actually beat DF to the FFXV demo analysis confirming the frame pacing issues and 1800p.

I do try my best, always tough as this is only a weekend thing 90% of the time. My full time job is a demanding mistress ;-). Thank you for the support, greatly appreciated.

Jett, and to some extent myself, were saying how it didn't look too good. You know that checkerboard 1800p, and according to DF here it's low-quality effects, are causing the poorer IQ.

I think I can speak for Jett when I say, it doesn't what it's supposed to be if it looks blurry.


Would love me some eyes that can discern pixel count like that ;-)

Now as for myself, of course, I'd be the first to admit I have a bias against NXGamer ~ and for that, I'll eat as much crow as possible. My disbelief was only amplified by the fact that the screens looked rough. Which, of course, I still stand by.

But when we spoke before you said that you were a fan of mine ;-( did I detect some passive aggressive delivery at that time? Or disingenuous statement? ;-)

In all seriousness that is fine, but me vs scenario is not what I am into and I welcome discussion and debate, but why the predetermined bias against me if I may ask?

Titanfall 1 is 792p on Xbox One and has FAR better IQ than the FFXV demo. So I don't think it's checkerboard-ness that's causing it. I'm happy to hear DF to analyse what the problem is.

tl;dr Jett isn't wrong for having his opinions. And the fact that DF here has to explain why there's so much softness should tell you that this is not what the average person should expect from 1800p.

I already stated this in the previous thread, The pixel count is only part of the conversation in what makes up a frame and Iq. I have been saying this for as long as my channel (and before) has been going.
 

Izuna

Banned
Would love me some eyes that can discern pixel count like that ;-)

Pixel counting aside, bad IQ is bad IQ ;). Generally it's why I don't like many TAA solutions.

But when we spoke before you said that you were a fan of mine ;-( did I detect some passive aggressive delivery at that time? Or disingenuous statement? ;-)

In all seriousness that is fine, but me vs scenario is not what I am into and I welcome discussion and debate, but why the predetermined bias against me if I may ask?

In the most light-hearted way to put it, you're the hero we need. Technical analysis is being trusted as "correct" ~ used to fuel console wars and is being used to educate would-be IQ enthusiasts. The majority of what I've seen from you is content I like. I don't mean to berate your efforts, but "ballparks" and "opinionated" descriptions is something I've noticed in a some DF videos (which are considered gospel) and that's something I personally dislike.

It's hard af and I couldn't do it, to discern what's going on in a game, but I don't see anything wrong with myself questioning the precision of tech analysis. I'm just a bedroom dweller also. [insert smiley here]

I already stated this in the previous thread, The pixel count is only part of the conversation in what makes up a frame and Iq. I have been saying this for as long as my channel (and before) has been going.

Well at the moment, it's to cross-reference between DF and yourself. I'm happy to eat crow with regard to my concerns if it means I'd no longer have to in the future.

As I said, I no problem admitting bias. So, good work with the FF vid. I'm happy you included the Xbox in the analysis, everyone else appeared to negate that.
 

Yoshichan

And they made him a Lord of Cinder. Not for virtue, but for might. Such is a lord, I suppose. But here I ask. Do we have a sodding chance?
So I have a 65" 4K HDR10 OLED with the OG PS4. Is it a waste? Can I still notice the HDR that is shown in the later parts of the video or is that capped to PS4 PRO?
 

artsi

Member
So I have a 65" 4K HDR10 OLED with the OG PS4. Is it a waste? Can I still notice the HDR that is shown in the later parts of the video or is that capped to PS4 PRO?

OG PS4 supports HDR, so you're good to go. The OG PS4 version still looks great.

But if you want to experience FFXV the absolute best way, Pro would be the choice :p
 

leeh

Member
To be fair 1800p checkerboarded is actually internally 1/2 that resolution.
I'm a bit puzzled, does it render at 1800p and checkerboard up to 2160p or is it rendered at 1440p or something and then checkerboard up to 1800p?

I thought they'd take whatever rendering resolution and checkerboard it up to 2160p?
 

Izuna

Banned
I'm a bit puzzled, does it render at 1800p and checkerboard up to 2160p or is it rendered at 1440p or something and then checkerboard up to 1800p?

I thought they'd take whatever rendering resolution and checkerboard it up to 2160p?

It's checkerboarded to 1800p, then merely upscaled to 4k. afaik
 
I feel like we should write checkerboard resolutions a different way. 1800p.c or something.

True 1800p looks fantastic. It's pretty much what the Surface Book uses as native <3 http://i.imgur.com/7IcJ0Nw.jpg <-- (and that's with FXAA)
Yeah, it would be a smart thing to do as it not being the same as that resolution visually and technically makes it a bit confusing I guess.
I'm a bit puzzled, does it render at 1800p and checkerboard up to 2160p or is it rendered at 1440p or something and then checkerboard up to 1800p?

I thought they'd take whatever rendering resolution and checkerboard it up to 2160p?

As far as I know, the game renders 2x 1600X900 in the checkboard style (so that amount of pixels, but with the appropriate gaps), interpolates the missing points up to 3200X1800, then this is upscaled to 3840X2160.

So it is internally half the resolution of 3200X1800. And since the post processing like motion blur is quarter or half resolution of that interal resolution, it looks really smudgy / pixelated in comparison to the final 3840X2160 output.
 

Caayn

Member
I feel like we should write checkerboard resolutions a different way. 1800p.c or something.

True 1800p looks fantastic. It's pretty much what the Surface Book uses as native <3 http://i.imgur.com/7IcJ0Nw.jpg <-- (and that's with FXAA)
Agree. It's important to differentiate between checkerboard and native resolution.
As far as I know, the game renders 2x 1600X900 in the checkboard style (so that amount of pixels, but with the appropriate gaps), interpolates the missing points up to 3200X1800, then this is upscaled to 3840X2160.

So it is internally half the resolution of 3200X1800. And since the post processing like motion blur is quarter or half resolution of that interal resolution, it looks really smudgy / pixelated in comparison to the final 3840X2160 output.
Shouldn't 3200*1800 checkerboard use +/- 2272*1278 as the base resolution?

1600*900 will result in only 1/4 of the pixels actually being shaded in a 3200*1800 frame.
 
Shouldn't 3200*1800 checkerboard use +/- 2272*1278 as the base resolution?

1600*900 will result in only 1/4 of the pixels actually being shaded in a 3200*1800 frame.

It's 2x1600x900, so a half of the pixels in a 3200x1800 frame. Technically it's 3200x1800 with every second grid of 2x2 being reconstructed
 
Shouldn't 3200*1800 checkerboard use +/- 2272*1278 as the base resolution?

1600*900 will result in only 1/4 of the pixels actually being shaded in a 3200*1800 frame.

2x 1600 X 900 was a silly way to write it, yeah lol. Woops. I sould say 2x the pixels of 1600X900. So half of the total pixels of 3200X1800.

So 2.880.000 pixels are "real" of those 5.760.000 pixels which are then scaled up to 8.294.400 pixels.
 

Caayn

Member
It's 2x1600x900, so a half of the pixels in a 3200x1800 frame. Technically it's 3200x1800 with every second grid of 2x2 being reconstructed

2x 1600 X 900 was a silly way to write it, yeah lol. Woops. I sould say 2x the pixels of 1600X900. So half of the total pixels of 3200X1800.

So 2.880.000 pixels are "real" of those 5.760.000 pixels which are then scaled up to 8.294.400 pixels.
Thanks. So just for clarification: An 1800p frame being constructed with checkerboarding uses 2x 1600*1900 + information from the previous frame?

If that's the case, shouldn't that be depended on the checkerboard solution used? I believe that Rainbow Six Siege actually uses checkerboard with 1/4 resolution with 2xMSAA to reconstruct the final image.
 
Saw this yesterday. Aside from frame pacing issues, this is shaping up to be a graphically amazing game.

Geez, this and KH are gonna be amazing with HDR.
 

jett

D-Member
I like how you both, especially jett, were calling out nxgamer in that other thread but are completely silent in this one. Can't admit you were wrong?



I got you fam.

Well jeez fine, but FFXV does look extremely blurry when displayed in 4K. Either I'm not familiar with 1800p checkerboard or it's just not that great of a resolution. It's not like I was the only one, almost everyone in the demo thread assumed the game was running at just 1080p.
 

Guymelef

Member
Don't put your faith in a patch to fix the frame pacing...
They had all this time since Platinum demo and it isn't fixed at lauch...
 

dark10x

Digital Foundry pixel pusher
Well jeez fine, but FFXV does look extremely blurry when displayed in 4K. Either I'm not familiar with 1800p checkerboard or it's just not that great of a resolution. It's not like I was the only one, almost everyone in the demo thread assumed the game was running at just 1080p.
It's way blurrier than 1800p should be but the steps don't lie. I think it's the TAA + super low res motion blur
 
Top Bottom