It's a hard trick that, just like flagpole glitch, requires multiple frame-perfect inputs just to save 0.35 seconds. (Not to mention the movement after the pipe clip, which requires a good deal of precision to not lose too much time and miss the framerule.)

So, don't worry about it yet, just go over the top of the level like normal and you can come back to this (much) later down the line once you start running out of other possible time saves.

I think you underestimate how much of a "slow down" the FFPG setups involve. Here's one example:

And for reference, each fast accel saves a couple of frames at most. (You only get to do 1 with that setup above, because you're skipping the underground shortcut too)

At the moment the new version appears to still be in-development, so we're waiting a little bit for it to either be finished and published onto the main branch, or seemingly abandoned (which would also not be surprising given Nicalis' track record).

Most likely however, the new version will end up being separated into its own subcategory. There will be a formal announcement in the coming days (and I also want to poll the community on a couple of other topics), so join the Discord server for more updates!

There is no FFPG setup for 1-1 that saves time

It's in the rules. "If you do not see a specific platform on this list, then it is not permitted." FCEUX is not in that list.

For the FPG: You are not on the right subpixel. (In fact, it doesn't look like you attempted any kind of subpixel manipulation at all.) Knowing what the "right subpixel" is would take more words to explain than I feel like typing out right now, so just know that if you delete the B input on frame 1226 in your TAS (this is the frame that you land on the third-to-last step), and then extend your jump off of the stairs to be a full jump (keep the timing the same), then you'll be set up perfectly to get the FPG. In fact, this is what RTA runners do to get the FPG in this level, so if you didn't know that there was an extra setup there then I would recommend reviewing an any% tutorial to get up to speed on that.

For the underground: You are holding right way too late. Play around with the timing until you find the earliest frame that you can start holding right without bonking into the wall. Also, you should avoid bonking into the pipe at the end by holding left as much as you need to so that you land on the ground before/at the same time you touch the pipe.

I don't recommend using the framerule plugin because the split times will be inaccurate. (Use the autosplitter instead if you play on emulator and don't trust your manual splitting.)

That said, if you want to use it anyways, then extract the DLL to the Components folder of your LiveSplit folder, and then under Edit Layout, click the plus button and add "Super Mario Bros. Plugin" (under the Control section) to your layout.

This is one of those things where it's better to just learn how to swim from the bottom. It's not that hard once you know the timings for when to push the swim button, and works 100% of the time regardless of how the firebars are.

If it was possible to clip into the ground while falling, then we would be doing FFPG in 8-3 (and other levels)

The main thing is that you want to do a small jump off the pipe to land as early as possible. Then it's just a matter of not overshooting the perfect pixel by too much

This is the recommended way to enable the autosplitter these days

Hey everyone,

We're considering a number of leaderboard rule changes and would like your input on them to help us decide how to proceed. If you could fill out the survey linked below, that would be much appreciated!

Here's the short version of what we're considering (these are explained in much more detail within the survey):

• Allowing a couple of (non-game altering) quality-of-life mods (namely, increasing the window size and allowing the game to run when unfocused)
• A mod to automatically advance text by simply holding the button so that you don't need to mash (with a delay to make it similar to human text mashing speed)
• Allowing Wine for runners that use Linux or Mac (although, Wine may be slightly faster than Windows; it's complicated)

For further discussion, we have threads about these topics in the Discord server (https://discord.gg/ksNyRY9), or you can also talk in here (or write your thoughts in the survey) if you don't want to join that for whatever reason.

Link to survey: https://forms.gle/8S14vc99bePCYBkX6

lol apparently it's actually fast enough (for a TAS using L+R)

(not optimized so there's probably at least a few frames that can be squeezed out of that)

It should be R08 with a perfect fast accel (R09 if you add a second backwards jump to get to running speed faster)

You in fact need to push something in the frame between the left press and the R+A input. It can be B (the most natural choice), up, down, or even select...there just has to be something there because otherwise Mario will retain some of the leftward speed from the left press and you'll get a bad accel. (Don't ask me why this is the case.)

It's much slower because you remain at 0 speed the entire time you're clipping into the coin block, whereas with the bump method you start accelerating back up to speed almost immediately after hitting the side of the block.

At 0:25, it's faster to jump so that you're already falling by the time you pass the edge of the platform (if you just run off, you'll have no vertical speed at that moment and have to accelerate from 0).

2:20 has a similar reason on top of the fact that you also have to turn around, and in this case, hitting the wall to instantly go to 0 speed so you can start moving left immediately (while facing backwards, which accelerates faster than facing forwards) happens to be a bit faster than trying to slide off the edge. There is a way to make the slowing-down-at-the-edge strategy a couple of frames faster, but the inputs are a lot more precise and it's also not needed to save this framerule, so there's no reason to go for it.

Edit: Here's a comparison of the wall strat vs. simply holding left to slow down and fall off the edge. When both are done optimally, it's actually a lot closer than I thought it would be, but if you're even just a little bit late on the left press for the latter strategy, your momentum will send you flying off to the right when you fall off and it'll be even slower than what you see here:

Currently, there isn't one (to my knowledge). I once looked into trying to get my autosplitter (the one that's enabled through LiveSplit's Edit Splits window) to support Bizhawk, but it's much less straightforward than Nestopia or Mesen, due to...reasons that I can't fully articulate. Basically, I'd need a completely different method of finding the RAM addresses that I need than what works for the other emulators, and I neither have a good idea for that right now nor am particularly motivated to try to find one. But if you (or anybody else reading this) think you can help, please hit me up and I'd be happy to chat about it with you!

Well, the TAS also saves an extra framerule, so...

There are a lot of different strategies that can get the 2-bump/blazit framerule, because the lightning framerule already has a lot of leeway for a TAS, so any strat that doesn't lose, like...40 frames over the fastest TAS strat should be able to at least get the blazit framerule.

I messed around briefly with TASing the pipe clip that you suggested, and yes it gets the blazit framerule easily (of course), but it seems to be a few frames short of getting the lightning framerule without L+R: