Third Vid. I am holding the down button until the end of each 1. There seems to be a 1 frame window where you will get a standing attack instead of a ducking attack. Sometimes I can get the timing right over and over but this is all I was able to record. You should get the idea.
The way computer interaction works and that includes consoles and games is, that all input command get put on an input stack which can be realized in software or even hardware but that doesn't matter. At a regular interval a dedicated process/threat reads the stack, interprets the commands and hands them over to the main threat. This bug is not about precision and execution but about timing. It depends on the collection interval and the commands being separated in subsequent stacks, that's why you cant reproduce it reliably even though one frame executions can be done with ease on a hitbox. Pinpoint just frame execution is not enough, you also need to be in correct evaluation cycle of the engine because you can input multiple commands in one frame.
There are four commands involved in this whole ordeal:
- Press down
- Release down
- Press one
- Release one
All depends on the order of these commands and whenever all of them are in the same stack or get split up in two different stacks. I can't say what the bug is about if there is one because no scientific or engineered approach has yet been applied to it but I can guess. Per your video and description it only happens in the following sequence:
- Press Down
- Press 1
- Release 1
- Wait
- Release Down
We can ignore the "wait" and "release down" command because the "bug" has already happened at that point. So this leaves us with four possible scenarios if my guess is correct (s1 = stack 1, ... ):
s1 - Press Down
s1 - Collect commands
s2 - Press 1
s2 - Collect commands
s3 - Release 1
s1 - Press Down
s1 - Collect commands
s2 - Press 1
s2 - Release 1
s1 - Press Down
s1 - Press 1
s1 - Collect commands
s2 - Release 1
s1 - Press Down
s1 - Press 1
s1 - Release 1
s1 - Collect commands
I can't say which one is more likely to produce a problem because it comes down to the specific programming of the engine and only proper debugging can reveal the truth.
We now need to factor in the "negative edge" system which interprets key up commands also as key down command. This could explain why instant down1 works because there are two commands in short order that have the same impact so if one gets swallowed, there is always the negative edge input to save the day:
- Press Down (gets swallowed by the game)
- Press 1
- Release Down (also counts as Press Down due to negative edge)
- Release 1
My guess is that if the input bug exists, it might happen far more often than we can see but due to the negative edge the command still works. I should get my "special hardware" today and see if I can get some "machine assisted" results on this because after all is said and done, none of us know what's up and all there is are wild speculations (such as this).