Has it been determined if any emulators are console-accurate for the credits warp glitch? I've been playing around with it on BizHawk but I don't know if any findings would actually translate to an NES. While messing around, I also found that it's viable to mix in the third room instead of the first one since it allows for better movement and lag reduction in the first two rooms: https://tasvideos.org/UserFiles/Info/638712787127905239 (uses same ROM as the current publication). Mixing in the first room might still be faster so I'm going to look into that next.
Not sure about the emulators, but no one has console verified any TASes for this game yet. I have yet to try the newest "no abilities" one (which uses Bizhawk 2.8).
I have however tried [3318] NES Kirby's Adventure "game end glitch" by TASeditor, MESHUGGAH, CoolKirby, Masterjun, MUGG & 0xwas in 00:35.76 quite a few times on hardware. From what I recall, Kirby would always end up doing something differently than what the TAS expected. Like not jumping at the exact right moment. I've never tested with RAM initialization though (don't have the hardware for that), so maybe that would make a difference.
Kirby not jumping at the right time is something I experienced when using any other version of the ROM, so be sure you're using the exact one that the TAS uses. A desync like that sounds more like lag differences between versions, whereas the game not warping to the credits might be affected by RAM initialization. In my experience, the game would always desync in the second room at the top of the stairs before I found the correct game version.
Has it been determined if any emulators are console-accurate for the credits warp glitch? I've been playing around with it on BizHawk but I don't know if any findings would actually translate to an NES. While messing around, I also found that it's viable to mix in the third room instead of the first one since it allows for better movement and lag reduction in the first two rooms: https://tasvideos.org/UserFiles/Info/638712787127905239 (uses same ROM as the current publication). Mixing in the first room might still be faster so I'm going to look into that next.
Good job - I've been working on it as well and was able to get it to 2070 frames. Not sure if I'll be able to squeeze out any more at this point:
https://tasvideos.org/UserFiles/Info/638717266965499595
If you try beating this TAS, let me know if you're able to do so.
I also tried optimizing the room 1 mix route. I was about 27 frames ahead of the current publication halfway through room 1 but lost most of those frames to lag and only ended up 8 frames ahead (it doesn't go to the credits, but it's slower than the other route anyway). I doubt this route is worth pursuing any further since room 3 mix seems much faster:
https://tasvideos.org/UserFiles/Info/638717262545190640
For some reason Bizhawk allows for a hard reset on frame 3 while FCEUX doesn't seem to allow one until frame 5 (I had a hard time even getting FCEUX to reset, so it's possible I just messed it up). I needed to use a different button combination on the last frame to get to the credits with FCEUX, but it doesn't seem like this was caused by the slower reset, as I tried inserting 2 frames at the start of the Bizhawk movie and nothing changed (besides being 2f slower, obviously). I wonder if one emulator is more console-accurate than the other or if they're both inaccurate.
Bizhawk movie: https://tasvideos.org/UserFiles/Info/638718771605159400
FCEUX movie: https://tasvideos.org/UserFiles/Info/638718773099137070
This 1-1 is 29 frames faster than the current non-credits warp movie: https://tasvideos.org/UserFiles/Info/638745497760453505
I didn't find optimizing U.F.O. movement to be very fun, so I don't plan on continuing with the category, but if anyone wants to add pause glitches into the run, feel free to use this as your starting point.
Joined: 1/7/2023
Posts: 119
Location: Somewhere on planet earth probably
DarkRiolu27 wrote:
This 1-1 is 29 frames faster than the current non-credits warp movie: https://tasvideos.org/UserFiles/Info/638745497760453505
I didn't find optimizing U.F.O. movement to be very fun, so I don't plan on continuing with the category, but if anyone wants to add pause glitches into the run, feel free to use this as your starting point.
I'm not interested in TASing it, but I've always had this question: How is optimizing UFO hard? Isn't it just holding forward most of the time?
"He who is tired of TASing is tired of life" Something Homer Simpson definitely said
I'm not interested in TASing it, but I've always had this question: How is optimizing UFO hard? Isn't it just holding forward most of the time?
It's not hard, it's just incredibly tedious and time-consuming. Kirby's Adventure already has the problem of slight adjustments to movement significantly changing lag, but when you're using U.F.O., optimizing for lag is basically the entire process. Moving up for one extra frame might save 3 frames of lag. It might be better to move up a few frames at a time instead of all at once. Add to that the different combinations of enemies that can be destroyed and different timings for shooting them and you have a nearly infinite combination of options with no way to tell when you're getting close to the most optimal amount of lag.
Joined: 1/7/2023
Posts: 119
Location: Somewhere on planet earth probably
DarkRiolu27 wrote:
Mikewillplays wrote:
I'm not interested in TASing it, but I've always had this question: How is optimizing UFO hard? Isn't it just holding forward most of the time?
It's not hard, it's just incredibly tedious and time-consuming. Kirby's Adventure already has the problem of slight adjustments to movement significantly changing lag, but when you're using U.F.O., optimizing for lag is basically the entire process. Moving up for one extra frame might save 3 frames of lag. It might be better to move up a few frames at a time instead of all at once. Add to that the different combinations of enemies that can be destroyed and different timings for shooting them and you have a nearly infinite combination of options with no way to tell when you're getting close to the most optimal amount of lag.
Ohhhhhhh, yeah, that sounds like hell. Isn't there a way to bot it or something? I remember trying to TAS a game with a similar issue and that was my idea
"He who is tired of TASing is tired of life" Something Homer Simpson definitely said
It is time to reveal some discoveries that I have kept supressed for almost 2 years!
On September 14, 2024, keyword submitted their run #9286: keyword's NES Kirby's Adventure "no abilities" in 43:20.92 which was a surprising run since it incorporated a number of brandnew, never-before-seen clipping bugs at the time. One of them was Corner Clip.
keyword wrote:
Corner Clips
If you fly into corner in a peculiar way you can clip into walls. Currently, I've only discovered a setup for clipping into two-high walls, since this clip seems particularly precise. Hopefully one day someone finds setups for all sized walls.
Used in 7-1, 7-6.
I was intrigued to test if this clip could be done not just by jumping up a 2 tile wall but also in other ways. I had a small PM convo with keyword to reveal my discoveries to them. Here are the most important of the messages I sent to them:
Hello keyword,
since you said, corner clip is currently possible only on 2 tile tall ledges, I investigated and I managed to make it work with 1 tile tall ledges.
It should be possible on all right corners now. I have not investigated left corners. (...)
I found this clip because I looked at your 2 tile corner clip in 7-1, noted down your input and positionings each frame.
The frame right before "<^B", Kirby has a y pos of ~.464
By testing I found that ~.46 to at least ~.48 and a y speed of -300 or -333 are sufficient for the clip to work, using the same input sequence as in your 7-1.
Hello,
I made changes to my luascript.
https://pastebin.com/aAUh2Wtf
It fixes that lag was counted continuously while the emulator was paused, and position and speed values are not shifting back and forth anymore.
I tried clipping in the level select room but level doors are not loaded until the door is actually unlocked.
I'd like to check for more improvements and research corner clip a bit more. Would it work on left corners or ceilings?
I looked for some more timesavers.
1-3 (67F faster)
2-Boss (22F)
3-1 (3F)
3-3 (265F)
I hope this helps.
6-3 (144F faster, saves 1 HP)
6-4 (212F faster, saves 1 HP)
6-5 (79F faster)
7-4 corner clips look possible in the four circled rooms
7-6 corner clip looks improvable in the last room. Also, corner clip may be faster in the cave room in this level.
Apologies...
The reason I didn't reveal these tricks here yet is to give keyword a chance at making a new run and even though they said in May 2025 that they were not working on a new run and I could reveal the tricks if I wanted, I decided not to. Sorry about that. I hope it doesn't destroy any runs that anyone might have been working on.
I will post some input files soon.
Really cool! Always fun to see new tech in this game. Like SM64 ABC, developments in an Alternative category push the speedgame further and can eventually improve any% too.
Is it not possible to move horizontally through that 6-5 wall, in the room where you fall to the bottom?
1-3 (67F faster)
2-Boss (22F)
3-1 (3F)
3-3 (265F) I hope this helps.
6-3 (144F faster, saves 1 HP)
6-4 (212F faster, saves 1 HP)
6-5 (79F faster) 7-4 corner clips look possible in the four circled rooms 7-6 corner clip looks improvable in the last room. Also, corner clip may be faster in the cave room in this level.
