Submission #7806: ktwo's NES Castelian in 13:16.85

Console Nintendo Entertainment System Emulator BizHawk 2.8
Game Version JPN Frame Count 47890 (Cycle Count 4278557487)
ROM Filename Kyoro-chan Land (Japan) Frame Rate 60.09891149261369
Branch Rerecord Count 17091
PowerOn Authors ktwo
Game Castelian
Submitted by ktwo on 11/12/2022 6:28:33 PM

Submission Comments

Game objectives

  • Emulator used: Bizhawk 2.8
  • Primary objective: speed
  • Plays on the fastest combination of version (JP) and difficulty (Hero)

Game mechanics


(At the time of writing this) pirohiko recently published a real-time speedrun that included alternative door exits. Some of them leading to shortcuts and significant time saves.
To enter an opening, you normally press up on the d-pad. But it turns out that one opening in every tower (except tower 7, which doesn't have any door openings) can be entered by pressing down instead. The exit changes if the opening was entered by pressing up or down.
It hasn't been investigated why some doors have the special property with two exit destinations. Remains from the development process that were forgotten to be removed or intentional shortcuts? The manual only mentions up as the method for entering a door.


As a general comment, TASing this game in bizhawk 2.8 (instead of 1.13.1) introduced quite a bit of additional lag. I've taken the liberty to ignore this fact in most of the level-by-level comparisons below. But the actual improvement is therefore slightly bigger than what the number of frames shows.

Tower 1

New shortcut from entering the door at f1883 by pressing down. 153f saved in this level.

Tower 2

New shortcut from entering the door at f6362 by pressing down. 470f saved in this level.

Tower 3

New shortcut from entering the door at f10014 by pressing down. 546f saved in this level.

Tower 4

New shortcut from entering the door at f18106 by pressing down. 169f saved in this level.

Tower 5

No route change in this tower, but 15f lost from random lag (emulation differences). There is a door with an alternative exit in this tower as well, but it's time-neutral. By testing both door exits, the new exit introduced additional lag. I assume this was random lag, which in this case just turned out to be slower. But if someone were to work on an improvement, the alternative exit should be tested again.

Tower 6

New shortcut from entering the door at f32416 by pressing down. 1130f saved in this level.

Tower 7

There are no openings in this tower, so no alternative exit to consider. Two game frames (= 3 actual frames each) were still saved by jumping earlier when climbing the steps at the first molecule encounter. Overall, this level was 4f slower than in the previous submission (random lag from emulation differences).

Tower 8

The alternative exit in this tower drops you on top of an enemy and is therefore not useful. By turning around on the second to last elevator before activating it, the enemy pattern became more favorable for a 54f gain. Overall, the level was 26f faster than the previous submission when including losses from random lag (emulation differences).


pirohiko for being first to publish the alternative door exits

Samsara: Claiming for judging. The file has also been replaced with a bugfixed version.
Samsara: There's some interesting discussion to be had about the alternative exits, namely the fact that it could have been a debug feature that was never removed. However, I don't feel as though this is the case. It's too deliberate. It's one doorway per level and not all of them are even useful, with one of them even punishing the player. That one in particular feels in line with the design of the game itself, so I feel that much more confident that these are intentionally placed secrets and fair game under our rules.
Accepting as an improvement to the published run.

despoa: Processing...

Last Edited by ktwo 2 days ago
Page History Latest diff