The Framerate Analysis Thread part 2

[E2K]

Ninja Gaiden Sigma 2 can't be analysed properly (by me, any way) because unless I've made a big mistake, it spends a lot of time updating faster than 60fps. There appears to be no frame rate cap. Faster than 60fps might sound great but in reality it means more torn frames.

No frame rate cap means that FPS averages are incorrect when the game runs at 60fps+ as we only have a sample of 60fps to work with. It's the theoretical maximum.

No frame rate cap means that torn frames can be entirely unique making them pretty much impossible to find programmatically (unless ps360 has solved this one!).

Download this: http://www.megaupload.com/?d=SKML6S5Z

In the NGS-60plus sequence you see that the torn frames show no common information from one to the next. Frame rate analysis will show these to be unique frames, not torn ones.

In the NGS-sub60 sequence we see the game running under 60fps... in a couple of frames you will see tearing that has video data in common with the frame before/after it. This is how we locate where the tear is.

My tool says that NGS2 has around 10% torn frames, just as ps360's does. However, my eyes tell me it's much more than that and I guess this is the reason why.

Can you compare torn frames / fps (even if they are different to our eyes, as you perfectly explained) in sigma 2 for the 'attract mode' which pops up when you don't press start after the developers title screens?
I want to know the impact of the 1080p software upscale.

 

[grandmaster]

The attract mode is an FMV movie captured and edited from someone playing the game, it is not actual gameplay. So it will be exactly the same between 720p and 1080p and will be just as impossible to analyse.

 

[E2K]

The attract mode is an FMV movie captured and edited from someone playing the game, it is not actual gameplay. So it will be exactly the same between 720p and 1080p and will be just as impossible to analyse.

Ah, okay, because of the massive tearing I expected it to be realtime. Companies mostly cleanup tearing, I read MS even has it as a requirement if you want to publish a game on XBL.

 

[Quaz51]

Ninja Gaiden Sigma 2 can't be analysed properly (by me, any way) because unless I've made a big mistake, it spends a lot of time updating faster than 60fps. There appears to be no frame rate cap. Faster than 60fps might sound great but in reality it means more torn frames.

No frame rate cap means that FPS averages are incorrect when the game runs at 60fps+ as we only have a sample of 60fps to work with. It's the theoretical maximum.

No frame rate cap means that torn frames can be entirely unique making them pretty much impossible to find programmatically (unless ps360 has solved this one!).

Download this: http://www.megaupload.com/?d=SKML6S5Z

In the NGS-60plus sequence you see that the torn frames show no common information from one to the next. Frame rate analysis will show these to be unique frames, not torn ones.

In the NGS-sub60 sequence we see the game running under 60fps... in a couple of frames you will see tearing that has video data in common with the frame before/after it. This is how we locate where the tear is.

My tool says that NGS2 has around 10% torn frames, just as ps360's does. However, my eyes tell me it's much more than that and I guess this is the reason why.

i don't understand. this screenshot sequence is all sub60fps and there are common data between all torn frames. the more difficult situation is the two first screenshot because there are just 3 common lines between the frames but it's here and it's sub60fps. one frame + 3 lines is 59.75fps

 

[grandmaster]

It is a requirement but they don't seem to enforce it any more now that the guy who came up with the requirement has left MS - hence Alan Wake looking a bit wobbly in the last vid I saw. Very, very few people do clean-up - probably because unless you automate the process it takes a HUGE amount of time.

 

[grandmaster]

i don't understand. this screenshot sequence is all sub60fps and there are common data between all torn frames. the more difficult situation is the two first screenshot because there are just 3 common lines between the frames but it's here and it's sub60fps

*smacks head*

Yes I see it now. There's a small % dither that is throwing off the tool.

 

[user542745831]

Ninja Gaiden Sigma 2 can't be analysed properly (by me, any way) because unless I've made a big mistake, it spends a lot of time updating faster than 60fps. There appears to be no frame rate cap. Faster than 60fps might sound great but in reality it means more torn frames.

I found with WipEout 720p that in certain stages it seems to exceed 60fps. You see completely unique torn frames with no residual data shared with the previous frame or the next. The frame rate analyser cannot lock on to these frames, and short of the capture tool dropping lots of frames (which I would have a visual indicator of), the only other explanation can be that there is no limit to frame rate.

What about WipEout HD / WipEout HD Fury?

 

[grandmaster]

I will check it when I (eventually) run the WipEout HD analysis but there is no dither on duplicate frames on WipEout, plus others I've shown affected clips to have concurred that there is no duplication of video data so for now I'm sticking with that.

 

[user542745831]

Thanks for your reply.

Maybe you could ask:

http://www.eurogamer.net/articles/digitalfoundry-wipeout-hd-fury-interview

 

[SG79]

With regards Ninja Gaiden Sigma 2, its frame rate is uncapped. I'm seeing torn frames that are entirely unique, suggesting 60fps+. A bit surprised that they didn't cap at 60 to be honest.

I've been told that the review and retail codes show far less tearing, so it's possible the capped it. I'd say it's surprising seeing how the NGII demo seemed smoother to me when the main reason why NGS2 seems to have slowdown is because of the lack of a cap.

Have you tried gathering crowds? Ignoring enemies, and they'll follow you into another area so the next encounter will be bigger. The area underneath the cherry blossoms is notorious for a noticeable drop.

 

[ps360]

Cabela's BGH 2010 Demo - XBOX360

Part1 http://zoome.jp/ps360/diary/392/
Part2 http://zoome.jp/ps360/diary/393/

NFS SHIFT - XBOX360
http://zoome.jp/ps360/diary/394

 

[Rotmm]

NFS SHIFT - XBOX360
http://zoome.jp/ps360/diary/394

Impressively steady framerate and almost no tearing. Game actually looks pretty good too.

 

[ps360]

Brutal Legend Demo - XBOX360

Part1 http://zoome.jp/ps360/diary/395/
PArt2 http://zoome.jp/ps360/diary/396/

 

[assurdum]

I've been told that the review and retail codes show far less tearing, so it's possible the capped it. I'd say it's surprising seeing how the NGII demo seemed smoother to me when the main reason why NGS2 seems to have slowdown is because of the lack of a cap.

Have you tried gathering crowds? Ignoring enemies, and they'll follow you into another area so the next encounter will be bigger. The area underneath the cherry blossoms is notorious for a noticeable drop.

Update : I'm sorry but I suspect your source had wrong: reading ign review seems the tearing problems not change so much in the review code and ign hardly notice a slight tearing. Slowdowns confirmed too.

 

[SG79]

Update : I'm sorry but I suspect your source had wrong: reading ign review seems the tearing problems not change so much in the review code and ign hardly notice a slight tearing. Slowdowns confirmed too.

Whatever slowdown is there isn't like the one in the demo build, and tearing remains. We're going off topic as it is, so I'll just say no, he was not mistaken at all.

The game is out in less than a week so see for yourself.

 

[ps360]

Forza Motorsport 3 DEMO

http://zoome.jp/ps360/diary/401/

 

[nightshade]

Forza Motorsport 3 DEMO

http://zoome.jp/ps360/diary/401/

Is there a problem in that video ?
Why does the yellow indicator stays between 40s to 50s while the white numbers stay at 60 ?

And during the start of video why is it as if its capped to 30 ?

 

[makattack]

Is there a problem in that video ?
Why does the yellow indicator stays between 40s to 50s while the white numbers stay at 60 ?

And during the start of video why is it as if its capped to 30 ?

One is the average (yellow text) and the other is the real-time rate indicator. The FMV is around 30 fps, hence the changing average as the game itself is 60 fps.

 

[2real4tv]

Played LP2(ps3) framerate seems to flucuate is it even locked at 30?

 

[ps360]

LOST PLANET 2 DEMO CO-OP - PS3

Mission1
http://zoome.jp/ps360/diary/402/

Mission2
http://zoome.jp/ps360/diary/403/

Mission3
http://zoome.jp/ps360/diary/404/
http://zoome.jp/ps360/diary/405/