Submission #6754: Archanfel & Baxter's NES Tetris "Mode B" in 00:36.89

Nintendo Entertainment System
Mode B
(Submitted: Mode B)
(Submitted: Tetris (U) [!].nes USA)
BizHawk 2.4.1
2217
60.0988138974405
25666
Unknown
Submitted by Archanfel on 5/20/2020 1:04 PM
Submission Comments
Hello TASVideos comяades! This is improvement of 114 frames over the previous Tetris "Mode B" TAS.

Game objectives

  • Emulator used: FCEUX-2.2.3 / BizHawk 2.4.1
  • Uses hardest difficulty
  • B-TYPE
  • Completes the game without any blocks left
  • Aims for shortest input time
  • Manipulates luck
  • Genre: Puzzle

Improvements

Archanfel

My humble improvement is new strategy for endgame. To finish the input earlier i brought the playing field at end to a specific configuration that the last two tetraminos could finish the game by free fall from the default spawn position.
No rotations + No movements = No input

Blazephlozard

Looking over the input I discovered that this game's soft drop mechanics were not being maxed out. At Level 19, a piece naturally goes down a row every 2 frames. After 3 frames of holding Down, your piece will go down 1 row, and reset that timer to 2 frames. So if you time that third Down frame right after a natural drop, you can go down 2 rows on 2 consecutive frames. But holding Down from there doesn't help, you have to restart a new 3-frame hold.
This leads to a pattern of 3 frames on, 2 off being the fastest soft drop I could find. Those 2 off frames are a great place to put horizontal movement without losing any time!
Dropping pieces faster allowed for new earlier RNG options, and facilitated an entire rebuild of the board. However, I still ended up aiming for the same shapes as Baxter did; I did make an effort for a new build, but it did not work out. There's some more info of my efforts in the forum thread.

Other comments

Initially i used 2/3 Baxter's old movie, it was extremely good optimised, but Blazephlozard with new soft drops tech made it even faster. However Baxter is still listed as coautor since his build was partially recreated.

Possible improvements

For sure it still is possible to ipmrove.
Game B is very hard to optimise. As was estimated by Masterjun in total there are about 32767 possible different seeds. However in practice only first 100-200 seeds are worth to be tested. If you will wait longer even luckiest pattent became poitless since you wasted to much time to wait it. However even "only" 200 cases is a lot of work. For example i gave up after testing only 4 cases. I wish good luck to hero of the future who will be able to test them all and find optimal.
By the way common mistake that i've read in discussion at Tetris NES forum is to exclude garbage patterns with odd starting blocks. It is wrong. Actually it is possible to clear all blocks in B-mode (a so-called "bravo") with almost any number of garbage pieces. If you build really high is possible to "cut" Tetramino(4 blocks) by top of screen and transform it into Tromino(3 blocks) and so change global parity.

Suggested screenshot

2298

Memory: Judging
Memory: Updating file with 91 frame improvement and updating emulator accordingly
Memory: Optimization looks good.
The internet appears to be a huge fan of quick tetris action and this TAS is no exception.
Accepting to moons to obsolete [925] NES Tetris "Mode B" by Baxter in 00:38.85.
feos: Pub.
Last Edited by adelikat on 11/4/2023 1:44 PM
Page History Latest diff List referrers