Submission #1007: Megafrost's GBA Metroid Fusion "0%" in 1:16:14.23

Console Game Boy Advance Emulator VBA
Game Version USA Frame Count 274454
ROM Filename 0690 - Metroid Fusion (U).vbm Frame Rate 60
Branch 0% Rerecord Count 31109
Unknown Authors Megafrost
Game Metroid Fusion
Submitted by Megafrost on 3/10/2006 11:30:12 PM

Submission Comments
  • VisualBoyAdvance - Rerecording 9b for windows
  • As fast as possible
  • Collects 0% of the items
  • Takes damage to save time
  • Manipulates luck
  • Abuses programming errors in game
Metroid Fusion is by far the most linear of the Metroid games, even more so than Metroid 2. All attempts to break sequence end in utter failure -- even getting an early powerbomb tank will do you no good until you actually download powerbomb from the Sector 5 (ARC) data room. It's still a good game for speedrunning, however.
My final in-game time is 0:42, 8 minutes faster than the currently fastest run on SDA.
Ordinarily there is a vine in Sector 3 (PYR) that forces you to take a route with an unavoidable missile tank, but the vine can be destroyed by a shinespark. However, the manuever to pull off that shinespark is long, complicated, and has a margin of error of only 7 frames. No human has ever pulled it off and I have no reason to believe it will ever happen.
The results of randomness for boss doors and enemy spawns are decided the moment you enter the room, while the color of X parasites and the actions of Yakuza and Zazabi are decided the moment they spawn. The randomness is based purely on the frame counter. Unfortunately some of the delays for luck are quite noticeable, but I hope to fix that in a future version.
It's worth noting that the flare in front of the charge beam when it fires will deal damage, so point blank shots deal more damage. This means I should be in the face of any bosses I'm fighting when I fire. Against Ridley the amount of additional damage is minimal, and against SA-X's first form the flare does not actually damage it.
Quite a few bosses are killed faster with missiles than charge beam, but I still blazed through them. Firing a missile leaves you with a looong delay before you can fire any other weapon, but you can still charge during that time. So against bosses that require missiles to kill quickly, I fire missiles in between charge beam shots. Ridley is actually killed faster with charge beam than missiles, but I only delay my charge beam shot 3 frames if I fire missiles in between. So unless the charge beam is worth 20 missiles against Ridley(which it isn't), using missiles in between shots is faster.
I waste as few shots as possible. There are a lot of times when I fire a seemingly random shot, but each of those shots actually has a purpose. Turning around is faster if you shoot 1 frame after pushing the opposite direction, and I can't come out of a flip after doing a wall jump unless I shoot. There are times when I shoot so that I spend as little time making the annoying charging beam as possible, but there are also times where I charge early in order for there not to be a stray shot. I can't explain my inconsistency here.
By far my favorite manuever is, while speed-boosting, to jump without going into a flip. This requires frame-perfection: Let go of the direction for one frame, push jump the next frame, then hold the direction after that. By not going into a flip, I allow myself to kneel on the ground when I hit it(which also requires frame-perfection).
As good as my run is, there is a lot of room for improvement. Around when I was working on my Nightmare fight, someone discovered that it is actually possible to hit Core-Xs with two missiles at once anytime during the fight, not just when it spawns. I wasn't about to restart the whole run then, but for consistency I decided not to use the trick. It might also be possible to better manipulate the Core-Xs so that they are killed sooner and float less. Apart from a few frames here or there, I don't see any other ways to improve this run.
I'm not certain exactly how much faster than my trial run this is, but my in-game time for the trial run was 0:43, compared to the 0:42 in this run.

Last Edited by DeHackEd on 3/12/2006 4:31:13 AM
Page History Latest diff List Referrers