P4A's Training Mode (Frame Data Related)

[TheRealBobMan]

I've noticed that when using the Recording function, the game displays the frames as you're recording as well as playing back what you've recorded.

Also, the game seems to advance slowly when pausing and unpausing. To the point where accurate timing your start button presses can make the game not advance at all or advance 1 frame.

Couldn't we use this to gather most frame data with relative ease instead of waiting or using recording devices and editing software? Just use the record function to make sure you're accurate about pressing the start button. Start the move several frames after you start recording, and just adjust your numbers off of when it started. Should be able to see how long a move takes to go active, how much stun, etc.

Right? Unless I'm missing something. Delete the thread if I'm wrong.

[Soriphen]

Holy shit, that never crossed my mind LOL. BB had the same thing. Though, it won't be accurate...and how would you know when the active frames begin and the recovery frames end? It would be much easier if the game had a hitbox viewer though.

[4r5]

This is assuming the recording clock count is based off the game's FPS and not some arbitrary count. To verify you would still need to go through the standard methods, or you would need a move whose data is known. (Though I think the 'known' data we have is still under question.)

Or, you could start a new set of data that is based off the recording clock, instead of the game's FPS. Lol.

Edited August 14, 2012 by 4r5

[TheRealBobMan]

I don't have an answer for getting the active frames, or even measuring hit-pause with perfect accuracy, but I checked it against the given data and some data that someone had put together through conventional means and it seems to add up exactly so far, albeit with only Slab's 5A and AoA tested. It's difficult to gauge exactly when block-stun ends because the animations are fluid, but if you stick someone directly in front of an attacking character and have him/her block, you can see the exact frame they go from idle to attacking, and then the frame that the move connects. It seems that instant blocking reduces the recovery animation time by 2 frames, so that seems accurate with what people were predicting before, right?

So assuming that this isn't an arbitrary timer, we can at least get startup time and possibly static difference. I'm still trying to figure out the in between. I think I can tell when the hit-pause ends, but I need to fiddle with this some more. I played around with it for a little bit last night when I posted this, and I'm on my lunch at work right now. If someone wants to cross verify this, it'd be much appreciated.

Or we could get some people to just gather the frame data. It's just that this works for those of us that want to put in the effort but don't have the equipment.

Hey, it might be a load of work to do every move and compile the data that way, but how much work is it to record the recording timer and verify that it's going off of the console's FPS? Shouldn't take anywhere near as long to do. If someone doesn't want to go through all the trouble for the full set of data, but wants to spend 20 minutes doing that for us so we can do the rest ourselves, that'd be pretty awesome. : )

[FunkyP]

I did some test with my capture equipment on PS3, recording the game at 60fps with the recording timer, and then slowing the footage down to 1fps i can confirm that the recording timer moves at the same framerate as the game, each frame per second corresponded to one count on the timer. One thing i noticed though is that testing AoA's with established frame data, using elizabeth's as an example her overhead was only 33 frames instead of the published 35.

I'd post the footage but youtube downgrades the framerate to 30fps so i dunno how effective i'd be

here's the video with a 60fps timecode

http://www.youtube.com/watch?v=Efwc3ylZUMo&feature=youtu.be

Edited August 14, 2012 by FunkyP

[Veteru]

The original poster's observation is correct. The in game record timer counts the number of frames that have passed. And yes it is possible to repause the game after advancing 1f. If you have the motivation to research frame data this way, more power to you.

[shtkn]

good to know! maybe we can even use this to figure out things like hitstop or attack levels and whatnot. maybe if i'm bored enough i can do this later.

if you've got really good recording equipment, i bet you could use the hitstun meter on the side to figure out how much hitstun each attack actually has!

[WolfCrimson]

Does this work for BB as well?

[c_nul]

I did some test with my capture equipment on PS3, recording the game at 60fps with the recording timer, and then slowing the footage down to 1fps i can confirm that the recording timer moves at the same framerate as the game, each frame per second corresponded to one count on the timer. One thing i noticed though is that testing AoA's with established frame data, using elizabeth's as an example her overhead was only 33 frames instead of the published 35.

I'd post the footage but youtube downgrades the framerate to 30fps so i dunno how effective i'd be

here's the video with a 60fps timecode

http://www.youtube.com/watch?v=Efwc3ylZUMo&feature=youtu.be

When i watch your video (nice video by the way) i think it could be 35 frames of startup if you watch the animation instead of the input. I know a lot of games where the input data have a lag display. So when you input something it could be possible the game show this only after several frames. In SF4 it's one frame after but maybe in Persona 4 it's 2 frames after

[FunkyP]

When i watch your video (nice video by the way) i think it could be 35 frames of startup if you watch the animation instead of the input. I know a lot of games where the input data have a lag display. So when you input something it could be possible the game show this only after several frames. In SF4 it's one frame after but maybe in Persona 4 it's 2 frames after

After reviewing the video you may be right, she starts up the stance a couple frames before the input is displayed, and the blocking animation starts up after the IB flash making it 35 frames realistically.

[Veteru]

When i watch your video (nice video by the way) i think it could be 35 frames of startup if you watch the animation instead of the input. I know a lot of games where the input data have a lag display. So when you input something it could be possible the game show this only after several frames. In SF4 it's one frame after but maybe in Persona 4 it's 2 frames after

The input display is definitely delayed. Arc System Works games has never been good about having the input display show up accurately.

In GG it would actually combine seperate button presses if you had pressed them closely enough.

In P4U the input usually shows up 1f after it begins to take place. I wouldn't be surprised if there was some variation to it though.

[Krackatoa]

After reviewing the video you may be right, she starts up the stance a couple frames before the input is displayed, and the blocking animation starts up after the IB flash making it 35 frames realistically.

I can confirm this as someone with capture equipment. The input display in Training Mode has a 2-frame delay on it. The Replay function also has one frame of startup.

[Xie]

Interesting! I've been acting under the assumption that the difference was 1F instead of 2F, so if that's true, it would change some of the data that I've been testing.

[Krackatoa]

Well, by 2F I mean that you input a move, the move starts on the next frame, and then the input comes out. So it could technically be construed as 1F, I guess.

tl;dr - Forget I ever posted.

Edited August 15, 2012 by Krackatoa

[Xie]

I see, I was recently using the way input appears on screen to determine throw break windows, so because of that, the numbers in my head are pretty fuzzy.

[shtkn]

http://www.dustloop.com/forums/showthread.php?14493-Mitsuru-Kirijo-General-Discussion&p=1378400&viewfull=1#post1378400

tested mitsuru startups for most attacks. How would i test supers or frame advantage?

[c_nul]

For frame advantage make both character jump (mirror match) after a move on hit or block and compare them during the jump to see who jump first and how many time before the other.

Becareful to not use this with jump cancel move, try another thing like crouching or moving backward though usually jump is the best option to figure out the end of a move

[GirugaMarc]

Did startup frame data on non-counter DPs using this.

Labrys- 25

Mitsuru- 9

Aigis- 8

Yikiko- 22

Yu- 9

Ahikiko - 8

Slabby- 14

Ground techs take 36 frames for everyone. Great, now I don't have to worry about safejumps not working on certain characters :p

Edited August 16, 2012 by GirugaMarc

[shtkn]

Short hop testing:

43F total

9F landing recovery

fastest attack input: 17F. it might be possible to do it faster? I dunno, don't know how to use the turbo on my stick.