Total frames of a move depends on its interaction with the opposing character

I posted earlier on Twitter how the total frame of Akira's on block is different from what is listed in the Total Frame column on VFDC command lists. The total frames in the command lists is based on the VF5FS Master Guide. It appears (might have been mentioned somewhere in the book) that the total frame of the move is for when the move does not make contact with the opposing character (whiffing).

Circular moves that are guarded have a unique animation. So I assumed for a long time there is a small subset of moves where the total frames of that move will changed based on its interaction with the opposing character. But while working on a video on fastest timing/execution of moves, I started to question whether changes to the total frame of a move is a common feature of VF5US/VF5FS game system.

Every move that I have tested so far-10-has some changes to the total frames dependent on its interaction with the opposing character.

Method to determine Total Frames of a Move by Identifying its Input Buffer Window
Anyone can determine the total frames of a move by having the "input display" toggled on in Training Mode. Specifically, the method shown here determines the beginning of the 12-frame input buffer window. Example I will use is .

1. Turn on "Input Display."
2. Set the CPU on Neutral, Normal Hit, Counter Hit, or Guard based on which interaction you want to determine the total frames of. To determine total frames for a move that did not make contact against opponent set CPU on Neutral and move as far away from the opponent.
3. Input the move you want to determine total frames for (for this example, ) and immediately input as fast as you can. For moves that have additional input options, you will want to press before inputting . will "reset" the game's interpretation of your inputs.
4. Make note of the amount of frames that have passed since the initial press of to the initial press of under two categories.
   1. If standing does not execute, you have entered before the input buffer window.
      
   2. If standing does execute, you have entered inside or after the input buffer window.
5. Repeat step 4, adjusting the timing of inputting .
   1. Keep only the highest frame number (N) for when your does not execute. You can erase or ignore the lower frame numbers.
   2. Keep only the lowest frame number for when your does execute. You can erase or ignore the higher frame numbers.
6. Repeating enough times, you will reach a point where will execute when inputted at frame N+1. At this point, you have determined the beginning of the input buffer window for the move.
7. The total frames of the move will be N+12.

Results
Pai, Akira, and Taka's Low Punch

• When does not make contact against opponent or is guarded by opponent: Total Frames 33
  • does not execute, outside input buffer window: 21
    
  • does execute, inside input buffer window: 22
• When normal hits or counter hits opponent: Total Frames 27
  • does not execute, outside input buffer window: 15
    
  • does execute, inside input buffer window: 16

Pai, Akira, and Taka's Standing Punch

• When does not make contact against opponent or is guarded by opponent: Total Frames 27
  • The second does not execute, outside input buffer window: 15
    
  • The second does execute, inside input buffer window: 16
• When normal hits or counter hits opponent: Total Frames 24
  • The second does not execute, outside input buffer window: 12
  • The second does execute, inside input buffer window: 13

Pai's and

• When or does not make contact against opponent: Total Frames 36
  • does not execute, outside input buffer window: 24
  • does execute, inside input buffer window: 25
• When or is guarded, normal hits or counter hits opponent: Total Frames 38
  • does not execute, outside input buffer window: 26
  • does execute, inside input buffer window: 27

Conclusion
So from the small sample size, it does seem to be fairly common for moves to have different total frames depending on whether it makes no contact, it is guarded, it normal hits, or it counter hits. There does not seem to be any specific pattern. Pai's and takes more frames to recover when making contact with the opponent than when it does not touch opponent. But Pai, Taka, and Akira's and (I assume all characters' version of these moves share the same stat) take less frames to recover on normal hit and counter hit.

Take home message? If you have no issues with fastest timing of moves, then this does not really have any importance to your game play and just take this as some esoteric information about frame data!

For those still struggling with hit checking (like me!), this may help explain some of the issues we have. For example, hit checking . There is a 6 frame shift to a 12-frame input buffer window from guard compared to a hit. The time you have to hit confirm from roughly ~1/3 of a second from guard to ~1/4 of a second on NH and CH is also significant, in my opinion. Hit checking on guard is already tough at ~1/4 of a second to hit check, but to hit check on NH and CH to get fastest time is even less ~1/5 of a second.

 

Color Key

• Yellow = Move's Execution Frames
• Red = Move's Active Frames
• Blue = Move's Recovery Frames
• Green = Opponent's Recovery Frames
• White Letters = 12-frame input buffer window

Made this after some discussion on the "speed" of Pai's with Chanchai. If the figure above does not make sense or that you think it is incorrect, please post and let us have a constructive discussion!

1. Guarding against an opponent's low punch will put your character in a guard animation - ~16 recovery frames
2. Guarding against Pai's will put your character in a guard animation - ~19 recovery frames
3. If you can guard confirm low punch and input moves to come out at fastest timing, it would be safe to assume guard confirm Pai's would be as difficult or easier to do.

Determination of Guard, NH, CH recovery frames is a combination of the method used in the original post and the frame data on VFDC's command lists.

 

I think this is where we agree to disagree during the stream on Sunday.

If you lined up Akira's diagrams to the respective Pai's diagrams, the input buffer windows of the opponent recieving the attack have a large overlap. Depending on how you aligned it, their input buffer windows have a 10 to 11 frame overlap. Thus, if you use the same timing against both Akira's and Pai's , the likelihood is your action should come out consistently the same.



Which brings back to the perception of what you referred to as the "stun". I assumed during our discussion, this is the guard, normal hit, or counter hit "stun" recovery frames. Which I listed to the left in the diagrams. In that sense, I think the "stun" recovery frames is more similar than different between Akira's and Pai's . My guess is that it will likely be similar for majority if not all safe fast mid attack category of moves (another topic of discussion...). It will need more empirical testing of other moves to prove though.

Or the "stun" you are referring to is reacting to what you see or are focusing on screen? For example, maybe with Akira's you recognized it by frame "15-16", but against Pai's you recognized the move by frame 20-21? Which would explain the "feeling" of Pai's having a faster stun recovery compared to other characters?

===

While not the reason I made the diagrams above, take notice of the input buffer window overlap (or lack of) when you defend or get CH by Akira's or Pai's .... There is only 1 frame of overlap between the two 12-frame input buffer windows. So a pretty significant difference to input timing your action.