For an optimal flight, you need to jump 20 frames before releasing "A" and starting to fly. Also, alternating "A" every other frame like I've done in the TAS (turbo) is optimal.
Edit: a slide-kick entering a new zone is optimal as long as you don't get 5.5 speed before entering the zone, but at least get 6.5 speed 2 frames before getting back your 6.0 speed entering the room.
Here is a (not very self explanatory) graph about the speed for SF vs. SF+blow (as seen here: http://www.youtube.com/watch?v=zC3yMvOtf7g)
Average speed for SF+blow is obviously superior, but doesn't mean it should be used always. For very short distance (such as reaching the end of a stage) you better use SF only, if it covers the distance.
If anyone is interested, the current Kirby 64 TAS is very improvable. survivewr came up with this technique (which I just call "boost") that consists of going right for 2 frames, letting go 6 frames, right for 2 frames, letting go 8 frames, etc. It can be found in a few very tight parts of the run where I had to use it, but without knowing lua or hex editing at the time, it made the TASing so tedious that I didn't really use it. Here's a graph:
You can clearly see the difference. That's only for 67 frames. There are well over 15k frames where this technique could be used and it could end up saving 1 minute in the long run, with a few other timesavers.
Also, this graph shows how it goes when you don't have start-up:
Don't mix-up the terms "optimal" and "optimized".
With that being said, the "120 stars" TAS is certainly not close to perfection as new tricks were quickly found just after we finished it.
The upcoming "0 stars" submission though will certainly be the most optimized N64 movie to date.
What new tricks were found? And where are you with 0 star?
Most known improvements are listed in the 120 stars submission text, but a new red coins route + new RR100 route/big house in the sky. Plus, many little improvements that were omitted because of luck manipulation or were just too hard to pull-off or required too much testing at the time.
As for the 0 stars, I'm not working on it, but the team is still trying to manipulate Bowser 1 to do the "dance". They should be done by the end of the month, I'm not sure.
Don't mix-up the terms "optimal" and "optimized".
Of course the MM TAS isn't "optimal" since new routes have been found and stuff, but it is certainly very "optimized" as everything is done on the first possible frame and the angles are straight and neat and the luck manipulation is well-done.
With that being said, the "120 stars" TAS is certainly not close to perfection as new tricks were quickly found just after we finished it.
The upcoming "0 stars" submission though will certainly be the most optimized N64 movie to date.
TAS OF THE YEAR, BABY!
Edit: you guys worked so hard on the entertainment part... it's insane.
I highly respect you guys, we need more devoted TASers like yourselves. Really great work!
I'm highly positive that this run can be improved using fox.
I am currently at the Kirby team part and 10 seconds ahead (using fox).
I'll surely lose time on Metal mario (unless I can manipulate him well), but will save a lot on the "race to the finish" part.
Sadly, I got a desync by loading a bad savestate on the mario and luigi part, but it was still a good training for me.
I would hold off publishing this (unless someone really wants to) before I get done with my run.
Average speed for SF+blow is obviously superior, but doesn't mean it should be used always. For very short distance (such as reaching the end of a stage) you better use SF only, if it covers the distance.
You can clearly see the difference. That's only for 67 frames. There are well over 15k frames where this technique could be used and it could end up saving 1 minute in the long run, with a few other timesavers.
Also, this graph shows how it goes when you don't have start-up:
