5-2 "366" Framerule Wasn't Actually Saved
3 years ago
United States

I just spent hours trying to figure out how Lekukie saved the mid 366 framerule in 5-2, and finally found out that he actually didn't.

When playing through the level normally, and doing the typical slowdown for devils spell (nothing fancy, just holding left), the first frame you can jump on for the clip is Frame 068. Jumping on this frame, and then doing a perfect devils spell with 3 frame perfect jumps will get you going down the pipe with Frame = 109 and R = 06 at the top of the Pellsson rom.

After entering the warp zone, the X sockfolder value in the HUD will be E. Here is a video with the typical HUD values (note that this devils spell is 2 frames slower than perfect, so all of the values will reflect that):

This is what you get when you start the level normally. Starting the level normally has the frame counter = 024. This holds true even if you restart the level with up+select. BUT, if you load a savestate (using left+select) after doing up+select, the HUD values will all change depending on what frame you did up+select.

In Lekukie's Twitch broadcast, he did just that. Here you can see him up+select to return to the beginning of the level, and it gives him the regular 024 frame number. But after running through it once, he loads back to a savestate he made at the beginning, and now the frame number is 152:

This actually messes up all of the frame counting HUD values, including the X sockfolder value and the R value. I tested this and everything lines up perfectly for it to be true. I got the frame count to be 152 myself at the beginning of the level, and here's a devils spell I did from there:

If you compare this devils spell to the first one of mine I linked above, you'll see they're the exact same speed. And yet, this one gets G into the warp zone instead of C. At this point, everything finally made sense to me. Looking at this, I could see 2 frames I could possibly save. I could jump 1 frame earlier after Mario comes to a stop, and I lost 1 frame on one of the jumps inside the pipe (the one where Mario turns around). So overall this devils spell is 2 frames slow.

Now compare this to Lekukie's: https://www.twitch.tv/lekukie/clip/BlindingCarelessChoughYee

You'll notice that everything is offset by exactly 2 frames! It all matches up perfectly! Lekukie jumps on frame 196 (1 frame earlier compared to the 197 in my video) and has all frame perfect jumps inside the pipe (Mario faces left for the entire clip). He gets "I" value upon entering the warp zone, which is 2 frames faster than G.

All of the HUD values are offset by 4 when comparing to regular HUD values that aren't messed up by loading a savestate. This means that Lekukie's X = I and R = 10 are actually the same speed as X = E and R = 6. When he finishes the level and gets R = 15, it's actually the same speed as R = 11, meaning it's 4 frames shy of saving the framerule. It's nothing to shake a stick at- his devils spell is the fastest you can go without doing a different slowdown method or a different clip method. But it unfortunately did not save the framerule. You can save 1 frame in the warp zone, and, according to Kriller, you can save 2 frames by doing a lower initial jump into the devils spell pipe and doing 4 frame perfect jumps inside the pipe instead of 3. All of that together would still come up 1 frame shy of saving the framerule. Drat! Not that it matters too much, since at that point it would be easier to just do the left side clip and save more time with it anyway.

Case closed! Be careful with your HUD values!

Edited by the author 3 years ago
Tecate, Logan_Sacrey and 10 others like this
British Columbia, Canada

Woah I never knew this. I'll keep this in mind when I use pellsson rom

Sweden

Well that's pretty neat, great work Kosmic!

iBall1 likes this
Game stats
Followers
1,001
Runs
1,763
Players
269
Latest news
FCEUX Ban

Following investigations regarding the accuracy and integrity of FCEUX, the SMB1/2J moderation team has decided that FCEUX will not be permitted for future submissions, effective September 17th, 2023. The accuracy of FCEUX has come into question following 108Pi's discovery that the current SMB1 Warp

6 months ago
Latest threads
Posted 3 years ago
0 replies
Posted 5 years ago
3 replies
Posted 6 years ago
0 replies