• Recorded with Mupen64 0.5
  • Time Trial mode
The Tetris and Tetris Attack runner did it again, this time on the Nintendo 64.
Tetrisphere isn't exactly like Tetris, but it's still a puzzle game involving a sphere. The main objective is to dig through several block layers (on the sphere) to reach the core, by dropping blocks to align 3 of the same type so they disappear.
I decided to do Time Trial mode, for a few reasons:
  • Other modes like Rescue, Puzzle and Hide&Seek have many levels and could get long and repetitive
  • Time Trial is the only mode where scoring matters, since you only get 5 minutes to score as much as possible, so I can show off instead of repeatedly rushing to the core
  • It's the shortest mode, yet still long enough to show off
I found 2 glitches while doing the run:
  • Input is accepted every frame but is only handled (by sum) about every 5 frames (varies), when the game status updates. I can abuse this when sliding a block to cut around corners, by moving it through another between two updates, the same way a character in another game could go through a wall by moving too fast.
  • If I move a block where it can fall, but hold another to slide right after, the block will stay up in air. This was useful a few times to setup gravity combos faster.
There's no luck manipulation in this run. The block randomization can be manipulated, but only 3 blocks in advance and I haven't checked exactly how. It wouldn't be too useful here anyway, as it's hard to plan 3 blocks in advance, and a large part of the scoring was done using the wildcard block to slide everything.
The hard part about making this run was that I had to plan every single move (what to do next, where can I do another gravity combo, how to do it, etc.), which was hard sometimes when I couldn't get a good view of the level.
I also had to do everything with perfect timing, which again had to be tested on every action (move, hold/release, drop) because there was a varying delay (sometimes 10-15 frames) caused by update lag.

Summary

Level 1

I max out the bonus multiplier early (20x, there's 2 ways to boost it and each must be done 10 times), then go all out to destroy blocks for the rest of the level.
Only one core piece (out of 100) needs to be cleared to finish the level, but since it waits for the combo to finish and I go too fast for that to ever happen, I clear the entire level. Well, almost ... a few blocks were left, all of the same type while I was stuck with the opposing type with nowhere to drop it at.
1001 combo, 93% cleared
  • Score: 45 639 900
  • Time: 0:03:27

Level 2

Again, I concentrate on clearing the entire level fast, and this time succeeded, leaving no block at all.
1038 combo, 100% cleared
  • Score: 125 483 900
  • Time: 0:02:13

Level 3

Unlike the first 2 levels with 3 layers, this one has 4, so it took longer. About halfway through, I started going for gravity combos (causing chain reactions when blocks fall) for bonus points instead of pure destructive speed, as each one gives a bonus equal to the current combo score. From there, the score takes a sudden jump ...
1158 combo, 100% cleared
  • Score: 276 804 300
  • Time: 0:00:27

Level 4

I start this one with a fast combo score, then time runs out ... or does it?
Since the game doesn't cut me off until there's no blocks disappearing or falling, I did about 1:17 of overtime to clear the entire level again with a high combo score and a lot of gravity combos, almost doubling the score in the process!
965 combo, 100% cleared
  • Score: 569 319 300
  • Time: 0:00:00

Final stats

  • Total score: 569 319 300
  • Highest combo score: 50 804 000
  • Biggest combo: 1158
  • X count: 227 (total gravity/fuse combos)
  • Total pieces taken: 4162
  • Total drops: 927
  • Cores uncovered: 393
  • Levels passed: 3
This could be improved with a better scoring strategy, as shown by the much faster scoring at the end, but I did a few different tricks and still scored a whole lot more than anyone could ever dream to get ... for reference, my realtime best is just a little over 15 million.

Bisqwit: Accepting, because it looks impressive, it's entertaining, and we want more N64 movies. :)
(Unfortunately, I can't encode this movie because of bad mupen64 plugins and/or bad opengl settings.)
Phil: Encoding...