I finally had a chance to look into this and... I hate to say it, but a problem immediately jumped out at me. Namely, the boss battles: You bring up the phantom hit glitch in the submission text, but it seems you didn't use it to its full potential in every applicable boss fight. It appears to always be the fastest way to deal damage to bosses. I did some light testing on several bosses with this glitch and was able to save almost 15 seconds across 5 different boss fights (1, 3, 6, 7 and 8), meaning there is still quite a lot more time to be saved over this submission, much more so when factoring in the final two bosses. Given the amount of health they have, it is possible this run could go sub-18.
Since syncing is difficult and my tests were everywhere, I've opted to upload a userfile of just my attempt at Sucker Puncher. It's the best example of how effective the phantom hit can be at saving huge amounts of time on bosses, as just this boss alone is nearly 7 seconds faster than in the submission. My input begins around frame 50000:
User movie #74102476863902801
410 frames saved. This guy's kind of a pain, he doesn't particularly like to be phantom hit, but it is always possible to do so, it just requires some good positioning and some pattern manipulation. I use normal punches to change what he's going to do next for the most part: Mixing in some normal punches might also help for damage routing on every boss, not just in terms of influencing their pattern, but also helping to get them below 140 health to use a final scream without wasting time on slower attacks. There's also a glitch(?) I found involving grabbing and instantly releasing enemies by pressing X a few frames after initiating a grab. This can help with positioning and pattern manipulation on presumably any boss that can be grabbed. Definitely play around more with this.
I'll also include a list of my other improved bosses and some improvement times to (at least) match or (hopefully) beat, along with some notes on what I did or what could be done better in general. These are NOT in the above userfile, but I can provide individual userfiles of these fights if wanted -
Boss 1 (Zink): 45 frames saved
This avoids using the scream at the beginning of the fight and mostly focuses on phantom hits. There's almost certainly a better strategy than what I did, though.
Boss 3 (Lava Boss): 160 frames saved
Honestly one of the easiest patterns to deal with in regards to chaining phantom hits. The scream I used to finish the boss is heavily delayed for some reason, I imagine this can be avoided with some fine-tuning to save even more time, but even with the delay there's still a significant improvement.
Boss 6 (Frank): 85 frames saved
This is honestly something that should have been avoided altogether: You use an extra phantom hit before the scream when the scream alone was enough to kill him. Memory address
$13D4 (2 byte, unsigned) is, I believe, the value for the first spawned enemy on-screen, which means it's always the health value for bosses. I'd highly recommend adding this to your RAM watch and keeping track of it for boss fights.
Boss 7 (Zink 2): 190 frames saved
Same thing as the first stage, just with more health and thus more phantom hits, and thus more time saved.
Keep in mind these are all first passes at these bosses: There is absolutely more time that can be saved with better damage routing. The intention here is that all of these times are improved upon further, or at the very least matched.
As for the actual stages... The game's pretty janky so it's hard to tell for sure, but they honestly look fine to me. Good enough on their own for a first publication of this game, I'd say. Assuming the game is sync stable, a final pass over each boss with updated strategies should definitely be worthy of publication. As I said, this could possibly be sub-18, perhaps even a low 17:5X with every boss being heavily optimized in damage routing. I do mean
each boss as well. I skipped over 2, 4 and 5 because they all looked passable, and the only thing I changed in 6 was the final hit, so it's entirely possible that all of these fights have further improvements of their own.
I'm sorry I couldn't get to this earlier (the huge changes to TASvideos have been keeping me either too busy to work on judgements or in dire need of some time to myself), I know it's rough to have put so much time and effort into this run only to be told that once again there's still a lot more time to come off of it, but boss optimization should HOPEFULLY be the final thing that gets this game through.
