The Framerate Analysis Thread part 2

[Neb]

With 60fps they mean 60Hz display output of a console (which still would be 60Hz even if the game would render at 20fps for example).

Here's another link from Digital Foundry:

http://www.digitalfoundry.org/blog/?p=694

Yeah but the video needs to be 60fps to capture all the unique frames on a 60hz monitor/TV. I assume PS360's source video is 60fps as the framerate bar shows it getting upwards 34fps in some places. However the final video is 30fps and uploaded for the masses. However it doesn't mather as it got the framerate info bar for each second to view framerate. But it seems cutscene framerate info might be wrong, say how about the rest of his analysis as his other video seems correct vs Eurogamer (GTAIV) which also on 360 runs above 30fps and has tearing.

Another question would be if tearing is the same becouse it is massive in Gamersyde and PS360 analysis and would probably result in a large perfomance drop if vsync was to be enabled. is it same for you?

 

[MazingerDUDE]

ISP going down. And I base it on PS360 video as hes shown his method lines up with Eurogamers analysis videos (GTAIV). But your then claiming his tools/measurment is wrong?

Maybe you should check out this video

http://zoome.jp/ps360/diary/224/

and here's a quote from grandmaster in his face-off 20 article regarding Bionic Commando's frame rate

There is the sense that the PS3 version is smoother (in actual fact, frame-rate analysis shows a plus/minus variance of around ten per cent in favour of either platform at any given point)

Yes I'm claiming his measurement is inaccurate based on my first-hand analysis with the actual game. Heck, if any of you have actually played the game, you'd have known that there's something wrong with the video.

EDIT: Done, your's look smoother and perhasp less tearing to(?), clearly over 30fps but then how come Gamersydes and PS360 video shows stuttering that is clearly visible and PS360 analysis is showing ~18-20fps?

In a 30 fps video, you'd see a torn frame twice as long as you'd in the actual game. That's why you shouldn never judge a game based on some internet video

 

[Neb]

Maybe you should check out this video

http://zoome.jp/ps360/diary/224/

and here's a quote from grandmaster in his face-off 20 article regarding Bionic Commando's frame rate.

So roughly ~10% inacurracy rate for Eurogamers videos/PS360's(?)?

Yes I'm claiming his measurement is inaccurate based on my first-hand analysis with the actual game. Heck, if any of you have actually played the game, you'd have known that there's something wrong with the video.

So care to share the video with framerate counter like the regular Digital Foundry videos or as PS360 analysis videos?

In a 30 fps video, you'd see a torn frame twice as long as you'd in the actual game. That's why you shouldn never judge a game based on some internet video

Thats why I quoted the numbers from his analysis in the video showing ~50% torn frames and the 18-20/25-34fps ranges. But then according to you the Eurogamer videos are worthless aswell as all other?

Eurogamer numbers are wrong to?

 

[MazingerDUDE]

So roughly ~10% inacurracy rate for Eurogamers videos/PS360's(?)?

In plain English, he meant that the frame rate is pretty much equal for both (unlike in ps360's video )

So care to share the video with framerate counter like the regular Digital Foundry videos or as PS360 analysis videos?

Why? Is my 60 fps HD video and 60 indivisually captured frame buffers not enough for you to see, that it's not 18-20 fps as seen in ps360's video?

I think I provided you enough, you'll see the frame rate analysis video probably later date in DF

Thats why I quoted the numbers from his analysis in the video showing ~50% torn frames and the 18-20/25-34fps ranges. But then according to you the Eurogamer videos are worthless aswell as all other?

Eurogamer numbers are wrong to?

Unfortuantely his numbers are wrong to begin with. And how does that make grandmaster's video to be wrong as well? And I was talking about your take on Gamersyde's 30fps video, and how you 'felt' the video was stuttering and running at 20 fps ,while you'd clearly see it's well over 30 in my 60fps video.

I do think anyone's opinion about a game based on some internet video (including Eurogamer's) not actual game itself is pretty much worthless

 

[Neb]

Why? Is my 60 fps HD video and 60 indivisually captured frame buffers not enough for you to see, that it's not 18-20 fps as seen in ps360's video?

I think I provided you enough, you'll see the frame rate analysis video probably later date in DF

But I did not see any framerate counter nor torn frames counter despite whatever the framerate guessing hints.

Unfortuantely his numbers are wrong to begin with. And how does that make grandmaster's video to be wrong as well? And I was talking about your take on Gamersyde's 30fps video, and how you 'felt' the video was stuttering and running at 20 fps ,while you'd clearly see it's well over 30 in my 60fps video.

Wrong for cutscene or wrong for whole video?

I do think anyone's opinion about a game based on some internet video (including Eurogamer's) not actual game itself is pretty much worthless

So then you answered your own question...

...And how does that make grandmaster's video to be wrong as well?...

All are worthless including those with software to measure framerate and torn frames?

So has this all been some kind of form to harvest hits on sites for Eurogamer and that Zoom site?

Becouse that is what this thread is about right? Solid numbers for tearing and framerate, right? Not how it feels as for some 15fps is perfectly smooth (huh) and for others awful.

 

[MazingerDUDE]

All are worthless including those with software to measure framerate and torn frames?

So has this all been some kind of form to harvest hits on sites for Eurogamer and that Zoom site?

Becouse that is what this thread is about right? Solid numbers for tearing and framerate, right? Not how it feels as for some 15fps is perfectly smooth (huh) and for others awful.

I think this discussion is going nowhere (if there ever was any) I believe I've provided enough source materials to back up my point. I'll end my part here

 

[Silent_Buddha]

Maybe you should check out this video

http://zoome.jp/ps360/diary/224/

and here's a quote from grandmaster in his face-off 20 article regarding Bionic Commando's frame rate



Yes I'm claiming his measurement is inaccurate based on my first-hand analysis with the actual game. Heck, if any of you have actually played the game, you'd have known that there's something wrong with the video.

So, is PS360's framerate analysis sometimes biased towards the PS3 whether through accident, capture tool configuration, analysis of online videos only or otherwise?

Regards,
SB

 

[MazingerDUDE]

So, is PS360's framerate analysis sometimes biased towards the PS3 whether through accident, capture tool configuration, analysis of online videos only or otherwise?

Regards,
SB

I think there's some problem with his frame rate measuring algorithm, as the numbers are way off in some games (the ones with lots of torn frames)

And it's got nothing to do with his tools as some people'd like to accuse me of criticizing. (In fact I use the same Intensity board myself )

 

[ps360]

Tearing Judgment
Brief description
/http://gamess.web.fc2.com/image/coj/tearing.PNG
It is actually complex a little more.

http://www3.telus.net/public/dhwag/COJ.zip
framerate-001.jpg XOR framerate-002.jpg


My ScreenShot
X0002.PNG XOR X0003.PNG



Tearing Frame --> No Frame Count

 

[nightshade]

http://www3.telus.net/public/dhwag/COJ.mp4

Here's the entire opening cut-scene encoded in 60 fps. If you still feel that the game's running at 20 fps, maybe you should have your eyes examined

Honestly...I earlier thought that the game ran at 40 something FPS.
But I was shy to state it here...thinking I would be wrong as the analysis by PS360 seemed otherwise

 

[Gubbi]

Tearing Frame --> No Frame Count

So, if every frame tears in a 60 fps feed, your framerate is zero ?

Cheers

 

[rekator]

I just downloaded the Gamersyde 10min gameplay of ~500MB to see the intro for reference. Seems to me it stutters and have massive tearing much like PS360's video hints. 20fps seems plausible. Also visible is the massive amount of tearing and what seems to be similar framerates as PS360's analysis. Say how's the tools working with such amounts of tearing?

Nebula on the french part of Gamersyde they indicated that there more tearing in their video than in-game. I remember that Grandmaster explained this by the fact of a 30fps video grow the tearing effect more than reality.
But to analyse something (this is for all the analyses), you need tools more precis than that you want measured. For a 30fps video in 1080p the best is a 60fps 2160p tools, but we don't have easily access for that so a 60fps 1080p is the best choice.
More your tools are precis more you analyse need less interpretation and less brainstorming for know if your results are good!

 

[grandmaster]

Disclaimer: I only got the game today and have only captured the cut-scene that Mazinger has captured.

What I have noticed however:

CoJ is a 1VBL game. You can get one torn frame following another. In fact you do, most of the time! This makes accurate tracking of torn frames very difficult, but actual frame rate using the methodology set down by Quaz51 in the original FPS thread gives me a reading above 40fps. And Mazinger is right in that looking at the cut-scene, it is clearly running more smoothly (torn frames excepted) than a 30fps game.

Assuming this continues into the gameplay, my contention is that the developers have gone for as much response from the controls as possible at the expense of visual consistency. V-sync off, as many frames as possible, the lowest latency from the controller as is possible.

 

[grandmaster]

In a 30 fps video, you'd see a torn frame twice as long as you'd in the actual game. That's why you shouldn never judge a game based on some internet video

Here's how it works. In a 2VBL game, typically every other frame in a 60Hz output has the potential to be torn. So when an internet video decimates that 60Hz down into the standard 30fps, one of these things happens at any given point:

1. The torn frame is on-screen for twice as long as it is in "real life"
2. The untorn frame before it or after it is on-screen for twice as long as it is in "real life"

In a 1VBL game like CoJ, the amount of torn frames has a strong possibility of being well over 50%. So there is far more chance of 1 happening than 2, making the game look worse than something like Red Faction Guerilla.

In short, as Mazinger says, never judge the extent of tearing by internet video, except to tell you whether it tears or not, if you see what I mean

 

[ps360]

So, if every frame tears in a 60 fps feed, your framerate is zero ?

The basis is so.

Bsasic tearing -> 0 Frame
http://zoome.jp/ps360/diary/202/

For 60fps Game tearing -> 0 - 0.9 frame (Calculation from site of incidence)
http://zoome.jp/ps360/diary/198/

 

[user542745831]

This makes accurate tracking of torn frames very difficult, but actual frame rate using the methodology set down by Quaz51 in the original FPS thread gives me a reading above 40fps.

So will you still be able to analyse Xbox 360 and PS3 version of the game with Digital Foundry performance tool and release a feature showing an accurate Digital Foundry framerate analysis graph?

In a 2VBL game, typically every other frame in a 60Hz output has the potential to be torn.

In a 1VBL game like CoJ, the amount of torn frames has a strong possibility of being well over 50%.

What does '1VBL' and '2VBL' stand for?

 

[Neb]

Nebula on the french part of Gamersyde they indicated that there more tearing in their video than in-game. I remember that Grandmaster explained this by the fact of a 30fps video grow the tearing effect more than reality.
But to analyse something (this is for all the analyses), you need tools more precis than that you want measured. For a 30fps video in 1080p the best is a 60fps 2160p tools, but we don't have easily access for that so a 60fps 1080p is the best choice.
More your tools are precis more you analyse need less interpretation and less brainstorming for know if your results are good!

Good to know although I was not reffering to amount of tearing by visual view but rather PS360's torn frame counter which shows roughly on avg 50% torn frames for his gameplay session. ~50% is a massive amount no mather what assuming the torn frame counter for his analsysis is correct. :smile:

 

[grandmaster]

Yes, that's an interesting way to calculate things. And that's where frame rate measurement deviations will occur between the methodologies: we measure the physical amount of times a set part of the screen refreshes (ie a single line) and call it "one frame". ps360's tool gauges a "percentage" based on the location of the tear.

The only time there would be disparity would be in 1VBL games like this one, which are thankfully very rare. Mandatory v-sync for next gen, please!

 

[ps360]

1VBL=60fps? 2VBL=30fps?

 

[grandmaster]

Effectively yes.