Console Games and Determine Frame Rate (FPS)

[CrimsonRayne]

Hi there, I'm, new

I've googled around for a solution and can't quite get the answer I'm looking for. Determining FPS in a PC game is easy enough, but I can't find anyway to determine the FPS in a console game.

Obviously, if things get busy in a game (let's say Uncharted 3 for instance) you can feel the FPS 'chug' but chug isn't an accurate example of how many FPS is actually being displayed per second.

Does anyone know how to measure the FPS of a console game? I have a PVR HD for capture, is that enough or do I need something more?

Thanks, Paul

 

[orangpelupa]

you need to feed the video to fps analyzer software. i quickly googled but cant find one

maybe make it yourself on avisnth?

 

[Shifty Geezer]

If you can record on a PVR with frame advance, you can count how many unique frames there are in a second for any particular clip as a brute-force solution, but it's not very exact as that method doesn't factor in partial screens rendered with tear. Some years ago there was discussion on this board about how to do exactly what you ask on an exact level, with looked into measuring unique frames and tears. The result of that discussion is used by Richard Leadbetter in his Digital Foundry articles on Eurogamer.

So basically, for an exact answer check out DF's faceoffs. For home investigation, you can just divide number of unique frames by amount of time.

 

[Cjail]

Yeah use DF face-off.
Also on YouTube there's a channel called "ps360 frme" that also does has frame-rate analysis and comparisons and to my knowledge it is reliable.
Not every title is covered though.

 

[CrimsonRayne]

If you can record on a PVR with frame advance, you can count how many unique frames there are in a second for any particular clip as a brute-force solution, but it's not very exact as that method doesn't factor in partial screens rendered with tear. Some years ago there was discussion on this board about how to do exactly what you ask on an exact level, with looked into measuring unique frames and tears. The result of that discussion is used by Richard Leadbetter in his Digital Foundry articles on Eurogamer.

So basically, for an exact answer check out DF's faceoffs. For home investigation, you can just divide number of unique frames by amount of time.

I'm trying to do my own analysis though, as I'd like to cover various titles myself and figure out the testing methods used.



Hmmm....

 

[Shifty Geezer]

If you search the forum for the original discussion, with 'grandmaster' as a poster, you should be able to find the discussion. You'll need some software though to test every individual pixel, otherwise you can only count frames different with your own observational skills. Oh, and compressed video streams inherently introduce frame changes that cause false positives. You need to evaluate the pure video stream.

 

[CrimsonRayne]

If you search the forum for the original discussion, with 'grandmaster' as a poster, you should be able to find the discussion. You'll need some software though to test every individual pixel, otherwise you can only count frames different with your own observational skills. Oh, and compressed video streams inherently introduce frame changes that cause false positives. You need to evaluate the pure video stream.

Hi there - sorry to be a pain in the butt, I've found this:

http://forum.beyond3d.com/showpost.php?p=1160288&postcount=53

which says:

"Methodology then. The fps detector counts the number of duplicated frames in its sample. The actual amount of screen being measured can be preset. To measure a precise fps, only one line of 1280 pixels from the centre of the screen is sampled.

To measure the amount of torn frames, the whole screen is sampled. Dupe frames from this test minus the dupe frames from the one line test are worked out as a percentage of the whole sample....*snip*"

It doesn't mention what the 'fps detector' is though - would it be avisnyth?

 

[Davros]

Aren't partial frames caused by frame rate exceeding screen refresh ?
Is that possible on a console ?

 

[Arwin]

Aren't partial frames caused by frame rate exceeding screen refresh ?
Is that possible on a console ?

Consoles are basically PCs that you can program at a much lower level. So expect everything a PC GPU can do and worse. The original xbox was a tearing monster here in Europe, some games like Ninja Gaiden had tearing constantly right in the middle of the sreen.

 

[Shifty Geezer]

HIt doesn't mention what the 'fps detector' is though - would it be avisnyth?

I'm pretty sure Richard got some proprietary software written. Does it not mention that elsewhere in the thread?

Aren't partial frames caused by frame rate exceeding screen refresh ?
Is that possible on a console ?

You can have two causes. One is frame overrun as you say. That's rare on consoles that struggle to hit the refresh rate of the display. The other is when a frame takes longer to render than one refresh of the screen. In these cases, if you don't lock refresh to the next vertical sync the new screen is rendered to the display when the previous screen is being redrawn. Illustrating with numbers, let's take a 30 fps game with a 33ms refresh. Every 33 ms, a screen is drawn. There's a frame being shown for 33 ms while the next one is rendered. If one screen takes 45 ms to render though, the previous screen is drawn twice. It's drawn for one whole frame, and then for another 12 ms while the new frame is still being rendered. When finished, that one is swapped in, and we have the old screen for the top 12/33 of the display, and the new screen for the other 21/33 of the display, and a tear between them.

Being a power-PC user, this concept of a frame taking longer to render than a screen refresh is probably very alien to you. the worst culprit I remember for tearing was actually Never Winter Nights. With v-sync off, it tore horrifically, 4 or 5 tears a screen sometimes, despite being 'smooth'. But then switching on v-sync cost the smooth framerate. It was a lose-lose with that game and a reason why console gaming was so good, because back then we had a focus on higher framerates and locked refresh.

 

[Davros]

Being a power-PC user,

You sweet talker you

 

[Arwin]

Yeah ... Watchdogs at the PS4 reveal was a nice example of lots of tearing as well. H wait, that was actually running on a high end PC, not the PS4.

 

[StSonic]

I am looking to do this exact same thing and am completely stuck of how to go about it. CrimsonRayne managed it as he was doing console analasys on his YouTube channel RedGamingTech.
My current capture card is the Elgato HD60 which may be a problem as it does not allow capturing any video with 3rd party software. Any more help on this subject will be greatly appreciated.

 

[Shifty Geezer]

I can only point to the same reference given CrimsonRayne. Read the DF discussion on how to do this, which, AFAIK, needs proprietary software. At least it did for Richard Leadbetter. Maybe CrimsonRayne figured out a solution using existing tools?

 

[StSonic]

Thanks for the quick reply shifty.

I don't know how crimson achieved this, he uses a standard frame counter similar to fraps as the overlay. I have been wondering if it's possible to record game play with a variable frame rate which matches the source.
VirtualDub has an option called 'integral framerate' which is what Nvidia uses for their FCAT analasys. I can't test it at the moment as the elgato does not work with their software.