I own both PS3-OE and PS2-JP versions but I tend to run the game on PC/FBA because it’s easier to get started at any given time, I can’t tell the difference between game speeds from one version to another but I’m definitely interested in hearing which emulator is the most accurate.
Just finished testing testing the 360/360 training mode and dreamcast.
3S Lag Results (60fps) / (120fps):
CPS3 - 4.18 / 7.24
PS2 - 4.98 / 9.42
PS3 (filters off) - 4.10 / 7.58
PS3 (filters on) - 4.04 / 7.64
360 (filters off) - 4.44 / 7.30
360 (training mode) - 4.50 / 7.52
Dreamcast - 6.36 / 11.36
As you can see, don’t play on dreamcast.
We plan on testing ps2 with a stock controller, emulators, and cps3 on an arcade monitor soon.
So 360 laggier than PS3? interesting
Yeah I asked about that too. Mr Isotopez said it doesn’t really mean it’s laggier than ps3. maybe he can explain in detail for the uninitiated such as myself.
Isotopez, do you know what the error/uncertainty on your measurements are? Did you perform repeats of the test?
This might tell us whether PS3 and 360 are equivalent or there is some small delay.
I think the difference between ps3 and 360 has to do with how the 120fps camera cuts the gameplay frame in two. For ps2 and both ps3 tests, the cut was somewhere above chuns head, but for both 360 tests, it was down by her knees (all of which is coincidental). This in turn affected how I counted the frames. I’m going to redo the tests for 360 and dreamcast so that each frame gets divided in close to the same place, which should make the results a little more consistent with each other.
So are the ps3/360 ports decent now after the patch?
I’m doing 1 video of 50 trials for each console.
I want to try to get more consistent results in regards to where this line:
http://i.imgur.com/6GS5Tzx.png
shows up on each console’s test. There’s some uncertainty introduced in how I am counting the frames as a result of that line being in a different spot between consoles.
I should attempt to do an uncertainty analysis, but I don’t want to get into that until the recorded video is more consistent.
Isn’t that line more based on the electricity itself? How exactly are you going to gauge that using homemade tests? A multimeter on the PSU? TORQUE ON TORQUE BITCHES
It’s probably the best way to have accurate and truly comparable results…
I also think it may be a better idea to have like 3 videos of 15 tries than only one with 50. I can’t explain why exactly and it may be a coincidence when i tried, but i think i had better results (less cut images) like that.
Also, could you verify on PS2 and DC how does the bars/sprites/etc change on hit like i did on the frame data thread few weeks ago?
I’m curious to know if the difference between OE and CPS3 (FBA, not directly arcade though) was also in DC and or PS2 port.
I’d bet for an OE exclusive thing…
Won’t the image always be cut like that if I’m recording a 60fps game with a 120fps camera? The camera will always be capturing one part of two different gameplay frames.
I’ll check it out and post my findings in the frame data thread.
I’ve finished rerecording and recounting footage for the 360 and dreamcast. I decided to only count the frames at 120fps and then divide those numbers by 2, which gives the lag at 60fps with half a frame of precision.
The videos for the consoles aren’t consistent with cps3 because they refresh at a different rate, so the console averages are likely skewed a little compared to cps3.
These numbers do not mean that 360 lags more than the ps3 or even cps3. Both consoles’ averages fall within the same range of error.
3S Lag Results:
CPS3 - 3.6
PS2 - 4.7
PS3 (filters off) - 3.8
PS3 (filters on) - 3.8
360 (filters off) - 4.1
360 (training mode) - 3.9
Dreamcast - 6.1
*All these averages should be +/- 0.5 frames of the actual lag.
Edited sent PM
let’s have a discussion on this, if you guys are interested.
so I think we can probably assume that 360 and PS3 both lag the same, though it is interesting that 360 averaged a bit higher consistently.
of course you can’t really lag 1/3 of a frame. a higher average would just mean that you’re getting a few more 5 frame delays than the other platforms. or it could just be “that’s how the screen was refreshing this time around” and mean nothing at all.
finding a way to be accurate about it is more difficult than initially expected. what I’m taking from this is that CPS3/360/PS3 all have the same delay offline, as does OE training mode. and that PS2 is delayed a frame, and that Dreamcast is delayed two frames. but with the kinda large window for precision, it’s hard to say any of that with 100% certainty.
there’s still the emulators to test, which I think is the one we’ll have to regard with the most skepticism - all we can really say is how much the emulators lag on one specific setup, within a half frame of accuracy.
Here’s the results for fba and shmupmame. Also, we don’t have a crt, so we had to use an lcd monitor, which comes with lag. We did these tests on a Dell U2312HM monitor, which is listed as having 9.3ms of lag here. I’m not sure how accurate that number is, but I posted lag numbers with the extra monitor lag and without it. Also, lcd monitors don’t show the image as being cut in two on the camera, but they come with ghosting, and with aero off there’s screen tearing, both of which possibly affected the lag count a little.
3S Lag Results (Excluding Monitor Lag)
FBA (aero on) - 6.0
FBA (aero off) - 4.6
Shmupmame (aero on) - 5.4
Shmupmame (aero off) - 3.8
Shmupmame (fullscreen aero off) - 3.7
3S Lag Results (Including Monitor Lag)
FBA (aero on) - 6.5
FBA (aero off) - 5.1
Shmupmame (aero on) - 6.0
Shmupmame (aero off) - 4.4
Shmupmame (fullscreen aero off) - 4.3
edit: We’re going to see if we can also test groovymame, but that may take a while to get up and running.
I’m more interested in game speed now!
me too. I think we could save your more accurate input lag testing for later. we have ballpark numbers now anyway, and I think these are mostly not a surprise to anyone, it’s just nice to have confirmation of what people have been saying for years.
onto the speed tests!
So Shmupmame is possibly the most accurate lag wise then? CPS3 being 3.6 and PS3 being 3.8, if we assume that the monitor response time, tearing and what-not didn’t mess with the final result.
Any error from screen tearing or ghosting would inflate the lag numbers since the move comes out, but it’s a little hard to see it happening, so I might count the next frame as the startup frame. Shmupmame’s purpose is to reduce lag, so it should be better than fba or mame in that regard. Groovymame is supposed to be another good alternative to fba or mame, but it’s harder to set up.
Our next goal is to test for speed differences and input timing differences. My plan for the input timing tests is to make a programmable controller. We’re thinking of testing necro’s elbow cannon juggles on urien (1 frame link), cancelling normals into stuff with different cancel timings, and super input windows. Let me know if there’s anything specific you guys want to know about, and we might be able to fit it in.
and people will still be using fba on shady lcd models and sub-optimal usb configuration. And still believing to play 3s.