Aight then, sounds kind of funny that you get the same results no matter what settings, for me the results vary if I fiddle with the options, but then again I have a shitty comp.
If it would be of any help, Iām able to make a lossless direct recording at a 60fps for 360, PS3, and FBA. If anything, there would be less variables when using a direct recording.
that would be cool. the more info we have, the better.
if we get the same results and make a video out of it, people will probably prefer watching a video of a direct capture anyway.
I recorded the same situation on FBA, 360, and PS3. Ryu vs Ryu on Ryuās stage, holding up for constant neutral jumping. No filters used on any system, and all done in the same standard aspect ratio. All three were recorded using the same HDMI capture card at 720p 60fps, and I did two runs of each just in case there was some sort of discrepancy from the same system.
As expected from other tests, the clock in FBA treated 54 frames as one second, and the clock in PS3/360 treated 60 frames as one second. However, the actual movement of the characters stayed in perfect sync up until FBA got a time over. It was the same case in both runs. If FBA is moving faster in actual gameplay than OE, itās by some amount less than a frame.
Looking at some videos recorded from a cab (After hunting for a decent chunk of time where no super freezes were present), it seems as if the clock runs slightly faster than real time on the actual cab as well. Of course, this is in no way conclusive as thereās plenty of unknowns which could cause the video to be inaccurate, and I lack a cab to test for myself.
Something weird about Auto Frameskip; the program and the ggpofba.ini file give conflicting ideas. If Auto Frameskip is set to on in FBA, it sets a value of 0 in the ggpofba.ini file, and if itās off, a value of 1. However, in the ggpofba.ini file, it states āIf non-zero, skip frames when needed to keep the emulation running at full speedā which would mean that 0 is off, 1 is on. I havenāt tested it extensively, but it seems like you want Auto Frameskip set to āonā in FBA in order for it to actually be off.
Huh, thatās interesting. Not what I would expect for sure. Each animation should be a set number of frames, right? So if FBA is running faster you would think these animations would resolve slightly faster as well.
So what does this test tell us then? That each animation occurs over a set number of frames, and that those frames are identical speed between all versions?
So what would differ then if theyāre all running at different speeds? Just the input windows, or other things too?
This test tells us that the game itself seems to run at 60fps on each system without skipping any frames, so at the base level, they appear to be at the same speed (Aside from the clock). Was it ever confirmed that they actually ran at different speeds, or was it left up to feeling? Even with this result, itās still very possible for the different versions to feel different.
There is still a lot that could account for perceived differences, as the design of this test eliminated input delay of controller, display, and system itself as factors. Even in my case where the same stick and monitor are used across all three cases, I still canāt help but feel the timing is slightly different on FBA than PS3/360; not the speed of the game, but the timing of inputs and such.
Hardware aside, thereās also the consideration that the code for OE is significantly different than the original game. Even in the case of a good port, slight coding differences could result in slight differences in how inputs and such are handled, which ends up being felt on an intuitive level in a game were frame specific timing exists.
You could probably PM someone who can record and has access to a 3S cab to get a recording of it.
Just tell them exactly what you need recorded (Ryu vs Ryu on Ryu stage, 60 FPS video, etc)
how did people not know that the timer is not real time? It isnāt in any fighting game? I think people have serious ADD if they see the timer and think those are real seconds. o.0;
Iāll also run the tests on my end for verification, but it seems pretty clear. This is more complex than I expected going in. For something that pretty much everyone agrees on (that the console ports and FBA are not correct timing wise) this is kind of tricky.
Yeah we could could do that. Given what we know so far though Iām not even sure if it will matter. If PS2, OE, and FBA all match on the jumping test, I dunno if we should expect different results from the arcade hardware.
Theyāre real seconds on PS2 and OE.
I think the next step is to check the input lag of each, since weāre not seeing any difference at a level where input lag is not a factor. Even if itās something as small as FBA having one frame less input lag than OE, thatās something that would cause a perceived difference in timing.
Did someone check PS2? Iād be veeeeery surprised if PS2 is in fact not faster than othersā¦
(you wrote PS2 several times above but i thought it was a mistake)
yeah, while PS2 having the same clock speed still blows my mind, we now know that clock speed does vary independently of gameplay, as shown by FBA running less frames per in-game second and therefore timing out first despite the jump tests matching.
actually, since FBA and the arcade have the same data, couldnāt we extrapolate what we currently know to assume the speed of the arcade version? for example, time a round from the arcade version and if it times out in unison with OE, then we know arcade is slower than OE and FBA. if it times out with FBA, then arcade is the same speed as OE and FBA. shinryu already did this, but he didnāt find it conclusive, but if we can find enough videos that donāt have super freezes, then i think we can come to a tentative conclusion without having to record a new arcade video.
edit: maybe thatās a little too optimistic. the probability of finding a 99-second time over without any freezes in an existing video is borderline zero.
Maybe you should also try doing stuff with the game when you test the speed, such as moving the characters around the screen, doing special and super moves, hitting the characters, etc. to sort of put stress on the system and to see if that creates irregularities between the platforms. You could also time all the stages to see if thereās some kind of ST bullshit going on with it, eg. different stages have different timers.
I went ahead and picked up the PS2 version today because the store down the street had it. After spending 30 minutes trying to figure out what dark corner of the universe in which my US PS2 was hiding, I did the same jump test with it. As a PS2 game, I donāt have the option of recording 720p 60fps over HDMI like with the other systems, so I settled for component 480i, separating the fields to produce 60fps.
The clock did line up exactly with OE at 60 frames each second, just as IglooBob reported. The jump test also matched exactly with OE, frame for frame. I was hoping for some sort of different resultā¦ It would be nice if we had some specific situation known/believed to be an issue between versions that we could check/test.
wow. well there goes 8 years of speculation, unless speed differences apply only to certain things and not others. youāre right, a different result wouldāve been interesting since it would confirm what we thought must be true. i was thinking about using turbo/rapid-fire to do some other move repeatedly, but those things would probably not offer flawless results.
Fucking unnerd this shit for people who actually play 3rd Strike and not Nerd Strike.
Iām super fucking tired and nothing is registering on anything anyways T_T
i have access to a arcade setup. let me know if you need anything. bewareā¦i am a noob with video shit.
extrapolate is a neat word.
QUOTED FO TRUTHINESS
exacerbate sounds dirty
Was this tested on different stages as well? In ST different stages run at different speeds.