warmCabin

MM2 Zipless Progress Update 1

Nov 8th, 2020 (edited)
484
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 8.00 KB | None | 0 0
  1. Quick:
  2.  
  3. Savings: 8 | 8
  4.  
  5. I painstakingly optimized the laser section. I saved some subpixels here and there, but I ultimately saved no time. The room with the sniper armors synced with the old run.
  6.  
  7. The fight itself had a major improvement. Using quantum shots, I was able to get 4 20s (nice) down to 19s, which was JUST ENOUGH to break the framerule!
  8. I did try these in the old run, but I kept running into the 3 shot on-screen limit. However, I found that I could fire my lemons sooner to make them disappear sooner, which gave me more room for manipulation. I also had to be a little bit persistent and try random stuff to make some of them actually work.
  9.  
  10. Old:
  11. 1: 26
  12. 2: 20
  13. 3: 19
  14. 4: 19
  15. 5: 20
  16. 6: 19
  17. 7: 19
  18. 8: 20
  19. 9: 19
  20. 10: 19
  21. 11: 20
  22. 12: 19
  23. 13: 19
  24. 14: 20
  25. GREAT JOB!
  26. Total delay: 278
  27.  
  28. New:
  29. 1: 26
  30. 2: 19
  31. 3: 19
  32. 4: 20
  33. 5: 19
  34. 6: 19
  35. 7: 19
  36. 8: 19
  37. 9: 19
  38. 10: 19
  39. 11: 19
  40. 12: 19
  41. 13: 19
  42. 14: 19
  43. HE DIDN'T STAND A CHANCE!
  44. Total delay: 274
  45.  
  46.  
  47. Metal:
  48. Pure Quick strats turned out to be faster!
  49. It's actually slower in terms of in-game time, but switching at the very beginning of the stage is a free 13 frames, meaning this strat just barely
  50. comes out ahead, by (X) frames. It's enough to save 8 (?) coming into the fight.
  51. I had given up on this strat because I couldn't shoot the drill on the fast belt, meaning I had to jump over it and lose a bunch of time (7 frames!!)
  52. It turns out that by manipulating the Item 1 values with extra shots, the salient boomerangs would have better velocities or trajectories or something,
  53. and the drill could be killed. The other TASers and I still have no idea how exactly each weapon uses these values, so we kind of treat them like RNG...
  54.  
  55. Shinryuu's Metal fight somehow saves a lag frame on the way out.
  56.  
  57.  
  58. Bubble:
  59. Saved 1 frame (or 2?) due to shorter hops on the 3rd frog. Especially by removing a 12-framer
  60. Saved 1 frame by KILLING THE HERMIT CRAB!
  61. Hilariously, I'm falling so fast that I outpace my Metal Blades, so it's optimal to shoot them a bit later.
  62. The subpixels in this section line up PERFECTLY. It's actually better to do a little short hop at the bottom to preserve them.
  63.  
  64. With extreme optimization that involved intentional extra lag, I was able to crack the framerule in the boss fight.
  65. It is now a perfect fight, with 19 frame deltas (+lag) between all the hits.
  66. It turns out that if a projectile would hit on an underwater lag frame, it will instead hit on the next frame, meaning a 19-frame shot can be impossible.
  67. By creating true lag frames, you can shift the underwater pattern and hit the shots as soon as possible...relative to the frame counter.
  68. It actually hits on the same frame, but the frame count ($1C) is 1 less, so you're further ahead relative to the framerule.
  69. This means that I only saved (X) frames instead of the full 8, but saving the full 8 is impossible.
  70.  
  71.  
  72. Flash:
  73. For some reason I could never hope to understand, the persistent throwing animation decided not to bother me this time around, saving me a frame for free.
  74. It was not enough to crack the boss entry framerule.
  75. I was able to save 2 frames in the boss fight, making it 1 frame off from beating a framerule!! However, the fight itself is perfect. It cannot be improved.
  76. We'd need a substantial difference in the stage itself to see any improvements.
  77. This stage is great for manipulating Item 1 values.
  78. On the boss order, and how it doesn't matter. Or...maybe Flash can beat a framerule?!?!?!
  79.  
  80. Heat:
  81. There was a subpixel save before the first ladder thanks to Maru, but it saved no time. This stage is pretty much airtight.
  82. Once again, I could save some time by attempting the yoku block section instead of slowing down to sync Shin's inputs, but I would much need to save 8 frames
  83. to actually see the difference. I only slowed down by ~2 frames, so I'm deeming this impossible.
  84.  
  85.  
  86. Air:
  87. Shinryuu found a nice optimization in the first bird section. If you slow down by jumping a little higher off the final Item 1, your subpixels line up such that
  88. that you can kill all three eggs without jumping. This saves subpixels (0 frames), but the Air Tiki jumps get misaligned, and it's slower overall after you fix them.
  89. Just because I like it so much, I decided to keep this optimization in and slow down a bit on the jump down to sync things up!
  90.  
  91. In my previous submission, I discovered* the "agressive despawn" technique, which saved time despite causing extra lag. However, this time around, I got unlucky:
  92. Bad RNG going into the Air fight means I had to waste 4 frames to get a pattern that didn't blow. Since I was throwing away that time anyway, I thought I might
  93. switch back to the classic egg kill strat to reduce lag. Figuring out the correct choice was a framerule fiasco.
  94.  
  95. My options were:
  96. 1. Make do with the shitty pattern
  97. 2. Wait 4 frames for better RNG (remember, I can only start boss fights every 4 frames)
  98. 3. Do the classic egg kill and start the fight 8 frames later
  99.  
  100. Option 1:
  101. There are some things not even TAS can do. Airman only gives us two viable patterns; the others will invariably result in suffering and disappointment, no matter
  102. how hard you try.
  103. Option 2:
  104. At first, it seemed like the 4 frames was going to be enough to lose the 8 on the way out. That would mean option 3 is the best, since that has less lag.
  105. But with aggressive optimization of the boss fight, I was able to save just enough time to beat the framerule, making this option the fastest!
  106. Option 3:
  107. This fight is one frame off from beating a framerule and being the optimal choice. However, the fight is perfect, and cannot be improved.
  108.  
  109.  
  110.  
  111. Tech stuff:
  112. Throw timers:
  113. Thanks to RTA runner cyghfer, a couple of confusing behaviors are now understood.
  114. I didn't mention this in my previous submission text, but sometimes, throwing animations seem to last a little longer if you're in the air. We TASers took
  115. this in stride: "Just try different jumps and see which is better. I probably miscounted." But cyghfer was not satisfied. He investigated it, and realized
  116. that *the timer does not decrease when you are moving upwards.* I was able to locate the timer at address 0x36. I believe this is a remnant from a special
  117. falling animation that got abandoned--think Mario's hat in Super Mario World.
  118. This had the additional effect of demystifying the ladder fidget technique. In my last submission, I noted that you gain height even though you actually
  119. fall downwards, describing it as "very mysterious." As it turns, this timer actually decrements by 2 every time you grab onto a ladder. So what this technique
  120. essentially does is cause that as many times as possible to run the timer down faster. This means I can empirically decide when to stop fidgeting instead of
  121. adding more until it stops working--you should stop at X frames. To be perfectly honest...this is still a bit mysterious to me.
  122. This had the *additional* additional effect of demystifying the off-screen ladder grab gods! Turns out that you get that 16-pixel boost if and only if
  123. the throw timer is 0.
  124.  
  125. Boss entry RNG:
  126. I didn't touch on this in my last submission text, but I should have.
  127. RNG is affected by your X subpixel, which usually gives you 16 options on any given frame. However, when you touch the boss door, your subpixel is reset (like any other wall),
  128. which limits your options. Depending on whether you walk, jump, or tiptoe into the fight, you will begin the fight at a different position, giving you 3 different options to
  129. work with. Sometimes that's just not enough, and you have to waste 4 frames to get the RNG you want.
  130.  
  131. Air Man pattern-spefic subpixel-specific timesave:
  132.  
  133. Mention Item 2 and 3 under fractional energy:
  134.  
  135.  
  136.  
  137.  
  138.  
  139.  
  140.  
  141.  
  142.  
Advertisement
Add Comment
Please, Sign In to add comment