Alyosha's comments:
SuperSqank, CoolKirby and I improve Madagascar thanks to better optimization and new time saving tricks.
Despite adding the BIOS and over 700 lag frames, the run is now shorter than the original and under 28 minutes.
This run is console verified as in the temp encode.
SuperSqank's comments:
I decided to run this game RTA in 2023 and throughout my time running, I found new uses for exploits and skips. I knew the TAS could go lower and when Alyosha gave me the opportunity to improve the TAS, I gladly took the opportunity. It's hard to say how much faster this run is due to the emulation differences but I believe it is around half a minute faster. A majority of the improvements in the run were done on my part with Alyosha improving certain levels/syncing up the run and CoolKirby returning to optimise the final level. Around half the levels are more or less the same but the other half are faster with some levels being multiple seconds faster.
The main exploit which is more heavily utilised compared to the previous TAS is heavier usage of double jump storage. It is possible to store Alex's/Melman's double jump by getting them in the air (through jumping or falling off a platform) and then switching characters. You can then do a running jump with Marty and then double jump with Alex and/or Melman, keeping the extra speed and height from Marty's jump. This was utilised sparingly in the old TAS but it is used in many more areas in this TAS and it can allow for new skips/routing, most notably in Lemur Liberation and Wild World.
It is possible for the run to be further optimised. Lag reduction throughout the run is definitely something which could potentially be done and some areas could have faster strats. The most notable place where time could be saved however is the Final Level. We had to reset twice in order to get favourable RNG in this level. Saving time earlier in the run can mess up the RNG in the Final Level and none of us were able to figure out how the RNG worked exactly or any alternative ways to manipulate it. If we could get the same RNG or better in the final level without having to reset (and without having to lose significant time earlier in the run), that would save a good couple of seconds at least.
Overall however, I am happy with how the improvement turned out. We managed to break the 28 barrier, even with the emulation differences and the new strats should help to make the run a bit more interesting. This is a run I quite like and as the RTA WR holder, I am glad I was able to contribute to this TAS. I want to give a special thanks to:
YoungJ for making the original TAS.
CoolKirby for making the original TAS and improving this TAS.
Alyosha for bringing this project into fruition, giving me and CoolKirby the opportunity to improve the run and console verifying the run.
CoolKirby's comments:
It's been several years since the original run, and sometimes I would wonder if new tricks would allow me to come back and improve it. Thanks to SuperSqank's expertise and Alyosha's unbreakable determination, a new run was born, and I had the chance to contribute. Though I learned about the run late and only had the hardware I brought on vacation, which runs GBAHawk at ~25 FPS, I was still able to contribute a timesave at the end of the run.
There's always the potential for better optimization, but this is a nice improvement by my co-authors on a great emulator that Alyosha worked hard to make console-accurate. Between GBA and DS, only the GBA has this level of precision at this time, which makes me glad that YoungJ1997lol and I chose to TAS the GBA version of the game. I'm also happy to see the run go below 28 minutes.