Submission #6848: arukAdo's GBC Castlevania: The Adventure in 14:01.02

Game Boy Color
(Submitted: Konami GB Collection Vol. 1: Castlevania: The Adventure)
baseline
BizHawk 2.4.2
50241
59.73796198201476
5979
Unknown
Submitted by arukAdo on 8/19/2020 9:13:24 PM
Submission Comments
This is an improvement of 3682 frames (3287 lag) over the published movie (1612M)
I could go on and on for hours to explain the things I've discovered while making this run(s), I'm not sure it would be very interesting or that I can remember everything but I've used all at my disposal and research on the grey cart to produce this, I will try to update later.
The major new tricks are:
  • using pause to grab/ungrab the rope, the game use a check every 4 frames to see if you should catch the rope (0502-player state), by adjusting the pause you can "skip" the rope.
  • pausing to despawn monsters, candles or platforms, address range 0600-0700-0800 for "objects", a block is 12 bytes (don't quote me), the spawning timer is at 0009, if you pause the frame before a monster would spawn (and the timer reach a multiply of 8 or 16) the game will just not try to spawn it.
  • you can also press backward to manipulate the spawn, from research done by ThunderAxe we know now that the game try to spawn the monsters but for some obscure reason it decide to not do it (unlike pause where he doesn't even try, this decision is only worth a couple of cycle and hence why there's no frame difference by itself).
  • note that if the scroller is on the left side (and unlike on right side), you cannot press L+R to despawn something, instead you have to L-R-L
  • monsters usually has to be despawn 2 times (timer with multiply of 8, so every 8 frames it try again if you are in range of the monster), but by generating more lag with candles, its possible to backward or pause once instead of twice (this isn't a theory but definitely working), yet, we haven't found the timers related to your position that determine where the checks happen (monster range), but in most cases the delay is 8 frames which is very short so its unlikely you can despawn further/faster than whats presented here.
  • autoscroller; if you don't press UP all the time but rather in short burst, you generate near zero lag, this is a lot more effective on grey version, but still save over 500 frames here too.
  • manipulate luck : unlike previous runs, pausing for 1 frame (pause-blank-pause) doesnt work to manipulate the rng, instead you have to pause for 2 frames or more (pause-blank-blank-pause), this is how you manipulate bats flypath, direction of fireball projectiles or stage2 boss holes spawn.
To improve further:
  • invulnerability gives you a speed boost, I didn't try to measure it with subpixel but its just obvious while watching anyway, I'm not sure if there's more on candles I didn't hit.
  • you can go and test all candles, but provided you don't use pause, it is almost always slower to hit candles, except like before climbing a rope for example.
  • I couldn't figure how to replicate the Dracula fight, I'm blaming emulation for that but I could be wrong.
  • fix eventual small mistakes

Samsara: Claiming for judging.
Samsara: Well, first off - Welcome back, arukAdo! I could turn that into a Welcome Back, Kotter reference, but I'm afraid of people actually getting it. Anyway, welcome back, arukOtter.
The optimization looks solid. Naturally, the big draw is the enemy despawning, but the new rope trick adds a lot to that as well, so there's still a clear improvement that isn't pure lag reduction. Great work overall!
I love that this run exists. It's a crazy technical achievement, being able to despawn almost every single enemy in the game, and I have to give props to you, Mazzin and ThunderAxe for pushing the game to its limit through thorough refinement of the Eldritch technique known as "pausing the game". Pretend that didn't come off as sarcastic. I feel like this is one of those runs I'm going to show people in the future for the kinds of things that a TAS can do. It may not be anywhere near as exciting as other runs, but it's clear throughout the entire run that something is happening that's making nothing happen, and honestly... That's my jam, but to take this analogy in an entirely different direction, the community seems to be more of a fan of jelly.
I asked for feedback on the site and on Discord, and for the most part the consensus was "GAME BORING". The voting since I asked for feedback has been more telling as well: Accounting for the fact that there is a single misvote (one Yes should be a No), the voting went from 13/1/1 (90%) to 14/5/4 (72%). Having watched the run several times by now, I'm inclined to agree: Game boring. (quietly googles which belmont this is) Christopher moves very slowly, the sweet tunes are constantly interrupted by the sounds of pausing, and the novelty of there not being any enemies wears off partway through. It's less Castlevania: The Adventure and more Castlevania: Chris P. Belmont's Lazy Afternoon Zen Experience, which I'll be honest sounds like a much better game, but still not a very interesting TAS.
I'm accepting this as an improvement to the published run. It's staying in Vault for now, but remember it could always be knocked up to Moons if the publication voting goes well.
fsvgm777: Processing.
Last Edited by adelikat on 11/5/2023 3:15 PM
Page History Latest diff List referrers