Rayman Advance is a port of the original Rayman to the GBA. It consists of various platforming challenges, bosses, and powerups. In order to unlock the path to the last boss, you must break all the cages hidden throughout the game, 6 in each level.
The improvement here is mainly due to a death abuse strategy on one of the bosses that I saw in the RTA run while I was researching the game and resyncing the vba file. This strategy saves over a minute.
Another big time save is in the very last boss, where I found you can hit him much earlier than in the original run, this saves several seconds. Additionally, I saved some frames with better grappling hook movement.
This run is also console verified, as in the temp encode. Console verifying this run was quite challenging as it uses EEPROM for saving. Even though it only actually makes a save file at the start of the game and then never saves after that, you can still get desyncs on console all the way at the last boss if timing is off by only 100 cycles between write timing on emulator and console. Additionally, the expected timing on my particular cart was ~1000 cycles different than the setting I used for Donkey Kong Country, so it took quite some time to find the right value. Timing between attempts is also a bit variable, which also caused desyncs. It works about 50% of the time.
Apparently you can skip the last boss entirely if you start from an existing save file that already beat him in a previous run. This is used in RTA but not available when starting from clean saves.
Finally, I found (and fixed) a serious save state issue in GBAHawk while making this run, so working on this game was all around very productive.
feos: Claiming for judging.
feos: There were questions in the thread about using new techniques found for this gam
We believe in iterative and collaborative improvements, not in restricting movies until they're perfect. Because if we aimed for the latter, we'd never get perfect runs, since they can only be perfect after - exactly - iterative and collaborative improvements, published or not. Unless the new submission is sloppy and slower than the current publication in a whole lot of places, it's considered decently done, and can serve as an interim version until someone wants to redo the whole thing again.
One-minute improvement may not be to much over the course of an hour, but it's still much better than, say, couple frames. Console verification is a good thing too.
