Before you get excited, it's not notable. Adjusting for RTA timing, the TAS gets 4:22, and a high one at that. Just like my original run.
It has brought into question whether 4:21 is possible, though. I've learned some things about this game whilst creating this TAS:
- Jumping, in and of itself, is faster.
- Falling is slower.
- Some parts of the current RTA route are actually slower, by frames.
There is one element to this game that ties all 3 of these points together, and is not detectable in a run: Lag reduction.
I had a NIGHTMARE fixing desyncs due to the way lag works in this game. For practical purposes, it's random. The frame that you perform specific actions will influence how many lag frames you get thereafter. Even an action such as jumping up to a platform will give variable results depending on when you jump.
The most surprising thing this taught me was that taking the third turtle in Stage 3 is SLOWER. How is this possible? It's because the speed the turtle moves is only marginally faster than taking the bottom path, but the lag generated by the higher path, including being forced to collect the heart up there, generates so much lag that it actually loses more frames than it saves.
I think if I went back and completely redid this TAS from the beginning with lag reduction in mind, it could be a 4:21 equivalent, but it's a lot of frames to save. About 50 of them. Will there ever be an RTA 4:21? Don't hold your breath. Will record and upload soon.
Did you also look into movement speed ? LosvikeN also did a TAS, got 4:21 iirc and also found out how movement speed works and it's pretty dumb. Basically as you start moving it increases until you get to a certain value, where it iterates between 2 different values, but those values can be different depending on how you move and jump/fall from platforms. So if you aren't paying attention to the values on RAM watch, you can actually lose a lot of time when walking, just because you don't have optimal movement speed.
I did see that Los did a TAS, but I have no idea what emulator it was performed on and thus how accurate it is.
The timing discrepancies between our two TASes are 2 seconds by TAS timing, 1 second by RTA timing, so there's a big difference somewhere.
Lag is a nightmare to work with in this game, and this is basically the result of a couple of hours of work; this was done more for the sake of calculating which RTA strats would be faster, but I may start over with lag reduction in mind. I wasn't aware of the movement speed however, so that's something else I'll keep an eye out for next time.
