Frame Data (2014/11/2 huge update)

[Doval]

Mega Buster
Arm cannon hit*: 7
First buster shot: 8
Second buster shot: 20
Third buster shot: 32
IASA with 1 shot: 36
IASA with 2 shots: 48
IASA with 3 shots: 60

* While idle and grounded.

It's probable that the arm cannon has a hit box on frames 19 and 31, will test further. The buster shots are clearly 12 frames apart, and there's 29/28 frames of recovery (depending on whether you count the arm hit or not) regardless of the number of shots fired.

Mega Upper (Strong Up Attack a.k.a. Up Tilt)
Invincibility: 5-8
Hit: 6
Airborne on: 9
Land on*: 50
Landing lag: 20
IASA*: 71

* Assumes it was done on a flat surface.

Slide (Strong Down Attack a.k.a. Down Tilt)
Lowest height*: 3
Hit: 4
IASA: 47

* Although the slide reaches its lowest point on frame 3, he already starts in a sliding position from frame 1. His head is a few pixels higher during frames 1 and 2.

Charge Shot (Side Smash Attack)
Hit: 19
IASA: 70

Spark Shock (Up Smash Attack)
Side hits: 9
Top hit: 11
IASA: 71

Flame Blast (Down Smash Attack)
Hit: 19
IASA: 79

Top Spin (Dash Attack)
First hit: 8
IASA: 60

Flame Sword (Forward Air Attack)
Hit: 9
IASA: 41

Slash Claw (Back Air Attack)
First Hit: 4
Second Hit: 7
Third Hit: 10
IASA: 43
Landing Lag: 19

Air Shooter (Up Air Attack)
Hit: 11
IASA: 54

Hard Knuckle (Down Air Attack)
Hit: 23
IASA: 65
Landing Lag: 24

Metal Blade (Standard Special 1)
Hit: 16
IASA: 43

Rush Coil (Up Special 1)
Rush appears on: 1
Invincibility: 6-11
Movement starts on: 7
Rush begins to leave on: 117
IASA: 41
Landing lag:

Crash Bomber (Side Special 1)
Hit: 20
IASA: 46
Detonation: 119 frames after sticking (2 seconds)

Leaf Shield (Down Special 1)
Leaves appear on*: 13
Leaves rotate on: 50
Leaves hit on: 52
Act on**: 60
IASA/Leaves disappear on: 146
Total leaf active time: 94 frames (1.73333... seconds)
Total actionable time***: 86 frames (1.56666... seconds)

* If the action is interrupted after the leaves appear (e.g. by landing or Rush or getting hit), they'll still become solid and begin to rotate.
** Can shield/move/jump/grab/throw on this frame.
*** Assumes the action isn't interrupted.

Jump
Crouch duration: 4
Airborne on: 5

Air Dodge:
Invincibility: 2-29

The Fine Print

The data is collected by recording videos at 60 FPS, splitting them into individual frames and manually inspecting the images. Due to camera motion, blur, and synchronization/ghosting issues, there may be slight (1 frame) errors here and there.

Hit data shows the first frame of the hit only. It's difficult to determine on which frame an attack ends.

IASA (Interruptible As Soon As) frames were collected by holding shield (on the ground) or rapidly pressing the jump button (in the air) during each move and observing when the action occurred. That's the first frame in which you're able to act, which is often slightly sooner than the move's animation ends.

Invincibility frames are based on the character flashing with the same white tint that's visible during rolls.

 

[Doval]

Reserved for future use.

 

[Z1GMA]

I tried to collect Frame Data for the first time yesterday.
What I found was that his Jab shows 10 different pictures before it hits (Arm Hit) on the 11'th Picture, meaning it hits on frame 11.
However, that seems way to slow to be correct, so I must be doing something wrong.
Still, there are 10 pics before the hit-pic comes out, so...

Before you add more moves, Bair showed 4 pictures before it hit on the 5'th, but that feels way too slow as well.
It feels more like it hits on frame 3.

 

[Doval]

I tried to collect Frame Data for the first time yesterday.
What I found was that his Jab shows 10 different pictures before it hits (Arm Hit) on the 11'th Picture, meaning it hits on frame 11.
However, that seems way to slow to be correct, so I must be doing something wrong.
Still, there are 10 pics before the hit-pic comes out, so...

Before you add more moves, Bair showed 4 pictures before it hit on the 5'th, but that feels way too slow as well.
It feels more like it hits on frame 3.

What are you recording with and what's the frame rate on the movie file (as seen in the file properties?)

EDIT: More importantly, how are you looking at the individual frames? I wouldn't trust a movie player to let you advance frame by frame correctly.

I haven't gotten to writing it up yet but my footage shows that b-air comes out on frame 4 (which is very quick, mind you).

Also I suspect that there's no arm hit, and that the buster shots start out inside the arm canon (and thus aren't visible on the very first frame they hit). I don't think Mega Man suffered hit lag in the footage, which he would've if he's hitting with his body. I need to double check. I'll test the second and third shots against hittable props like the tendon in Brinstar to get around the problem of the first shot pushing victims away.

I'll try to get less blurry footage of Mega Upper to see if I can make out where the invincibility ends too.

 

[Z1GMA]

What are you recording with and what's the frame rate on the movie file (as seen in the file properties?)

EDIT: More importantly, how are you looking at the individual frames? I wouldn't trust a movie player to let you advance frame by frame correctly.

I used Mega Man in training mode at 1/4 speed while recording with a digital camera.
I then moved the footage to my computer and simply played it, then slowed it down further so that I was able to see each single picture of the animation(s).

Jab showed me 10 pictures and then hit on picture 11.
Jab is clearly not that slow, (I have a pretty good sense for "feeling" amounts of frames, roughly) so I don't know why it shows me 10 pictures before the hit-picture.

The problem is probably my lack of experience when it comes to collecting Frame data, hah.

Looking forward to getting correct stuff from you, though.

 

[Doval]

I used Mega Man in training mode at 1/4 speed while recording with a digital camera.
I then moved the footage to my computer and simply played it, then slowed it down further so that I was able to see each single picture of the animation(s).

Jab showed me 10 pictures and then hit on picture 11.
Jab is clearly not that slow, (I have a pretty good sense for "feeling" amounts of frames, roughly) so I don't know why it shows me 10 pictures before the hit-picture.

The problem is probably my lack of experience when it comes to collecting Frame data, hah.

Looking forward to getting correct stuff from you, though.

There's a couple of things at play here:

• I'll bet that the game's engine can vary the number of in-between animation frames according to the speed you're playing at to make everything still look smooth; if it simply played the same number of animation frames repeated 4 times it'd look really jerky, and it'd look jerkier still at 2/3 speed.
• I'll bet whatever device you're using records at 29.97 FPS (that's what my phone does, and my camera does that by default - 60 FPS recording is a special mode it has for recording high-speed things), so you only get half the frames.
• The movie player probably doesn't let you stop at every specific frame. Videos aren't stored as a sequence of images; since most frames are very similar to their predecessor, videos only store a full image every so often and the rest of the frames simply encode what changed from the previous frame.

The 3rd point is probably skewing your data the most. I doubt the video player lets you skip to every 1/30th of a second in the video. I'm using ffmpeg to split up the video into individual images. If you want to try it:

1. Download it
2. Unzip it
3. Put the video file inside the bin folder
4. Run Command Prompt
   
5. Run this command: <full path to ffmpeg\bin\ffmpeg.exe> -i <full path to the movie file>
   -y -f image2 -c:v mjpeg %06d.jpg

You can make things a bit simpler by putting the video in the bin folder and then doing cd <full path to ffmpeg\bin>. Then the command from step 5 reduces to ffmpeg.exe -i movie_name.mov -y ...

 

[fromundaman]

Thank you very much for this!

Can I make a request though?
For the jab, can you do a count for IASA frames on grounded commands (I assume that's what you did here) and one for how many frames it takes to cancel a jab into a jump (This is visibly less than what it takes to cancel a jab into a grounded command like shield).

 

[Doval]

Thank you very much for this!

Can I make a request though?
For the jab, can you do a count for IASA frames on grounded commands (I assume that's what you did here) and one for how many frames it takes to cancel a jab into a jump (This is visibly less than what it takes to cancel a jab into a grounded command like shield).

You're welcome. I'm not sure if I understand your request. You can't cancel the buster shots (other than by firing less shots) and you can ground jump/move at any point in the move.

 

[Locke 06]

I'll also put in a request.

Could you do phantom landing lag? That'd be utilt off the edge and then landing normally from FH and SH separately, as well as the FH/SH landing lag after jumping off of Rush and cancelling the normal lag with NAir.

 

[Doval]

I'll also put in a request.

Could you do phantom landing lag? That'd be utilt off the edge and then landing normally from FH and SH separately, as well as the FH/SH landing lag after jumping off of Rush and cancelling the normal lag with NAir.

I can check after I finish going through the videos I recorded on Friday. I thought phantom landing lag is the same lag, just at a later time?

Added Rush Coil and air dodge just now, that'll be my last addition for today. It's a bit incomplete (I didn't record Rush Coil's landing lag nor a full air dodge duration) so I'll have to go back to those later.

 

[Locke 06]

I assume it is, but I was just hoping we could confirm. Also, short hopping the phantom lag, I think, makes the lag shorter.

edit: At your convenience, of course.

 

[SBjunkie]

I could really use this information. thanx dude. I knew mega upper had invincibility frames because all shoryukens do. Pretty good move to challenge other moves with.

 

[Z1GMA]

@ D Doval , would it be possible for you to test how many frames you're allowed to hold jump and still shorthop?
Like, at what frame does the game read it as a fullhop?

Could you then also test it while in the middle of Jab/Ftilt?
(I have a feeling it's harder to shorthop while spraying pellets)

 

[Doval]

@ D Doval , would it be possible for you to test how many frames you're allowed to hold jump and still shorthop?
Like, at what frame does the game read it as a fullhop?

Could you then also test it while in the middle of Jab/Ftilt?
(I have a feeling it's harder to shorthop while spraying pellets)

It's impossible for me to test this because I rely on analyzing video after-the-fact. To know that I'd need to be able to play on a frame by frame basis. In all the previous games you short hop if you're not holding jump on the frame you're airborne. If you play around at 1/4 speed in Training, especially with a slow jump character, that still appears to be true.

I haven't had a chance to look at the footage for jumps during buster shots, but I suspect the difficulty is in the logistics of going from attack to jump to attack, which changes the way you hit the jump button.

 

[Z1GMA]

It's impossible for me to test this because I rely on analyzing video after-the-fact. To know that I'd need to be able to play on a frame by frame basis. In all the previous games you short hop if you're not holding jump on the frame you're airborne. If you play around at 1/4 speed in Training, especially with a slow jump character, that still appears to be true.

I haven't had a chance to look at the footage for jumps during buster shots, but I suspect the difficulty is in the logistics of going from attack to jump to attack, which changes the way you hit the jump button.

Well, I believe Mega Man "gets airborne" faster when spraying pellets, then when not spraying pellets.
That could be what's causing it.

Imma test it in 1/4 later and see if I can feel any difference.

 

[Fenrir VII]

So Thinkaman is compiling frame data from a script dump by Dantarion for 1.0.4, and there seems to be some differences from what we have.
Link to data: http://smashboards.com/threads/frame-data-and-hitbox-values-for-every-character-wip.378789/

Currently no IASA data at that link... not sure if he'll be able to add it.


Jab1 - Hits on frame: 2.0 < 5 frames faster compared to Doval's 1.0.3 data
Nair - Hits on frame: 2.0 < 5 frames faster compared to Doval's 1.0.3 data
Dash Attack - Hits on frame: 8.0 < No change
U-tilt - Hits on frame: 6.0 < No change
D-tilt - Hits on frame: 5.0 < 1 frame slower compared to Doval's 1.0.3 data
U-smash release - Hits on frame: 8.0 < 1 frame faster compared to Doval's 1.0.3 data
D-smash release - Hits on frame: 17.0 < 2 frames faster compared to Doval's 1.0.3 data
Fair - Hits on frame: 9.0 < No change
Bair - Hits on frame: 4.0 < No change

Nair/jab change confirmed false... just a difference in what's being measured I suspect. No change from old