The pipes don't lead to anything special. And going into that SXF0 section the first column behaves like a wine, the second one like blocks and when going past that you get warped to the first part of the SXF0 section. It's weird. Don't ask me how it works.

If pipeless = mario can't go in any pipes, that's not possible. Look at 8-4 for example

I can reproduce this.

I also just found something similar with the big pipe blocking the way diappearing when breaking some blocks at the section where big mario only fits through when ducking. That causes one to be able to get to the end, but interestingly enough the pipe in the ground before the exit pipe is coin blocks instead of a pipe. Weird.

If you go on the top row of bricks by walljumping somewhere early you can also see the full level without the big pipe, but it's different as the pipes the no cheats run does pipe jumps on are sometimes there and sometimes replaced by coin block rows.

EDIT: what the hell https://imgur.com/a/wX7sQAO

Here's a minus world ATEOPT run: (8-1 and 8-2 in any% are hard, that's why I have not gotten a run there)

My idea for a warpless equivalent (as a possible category on the seperate leaderboard we might make, not for this site right now): Since 3-3 is impossible, play up to and including 3-2 (maybe even start 3-3). Then reset or game over and take the warp zone in 1-2 going to 4-1. Then play warpless to the end. That only leaves out 2 levels out of which at least one is impossible. It's sort of a max amout of levels% category.

Also I can help with the leaderboard if that's wanted

I just "completed" a any% run using savestates. It's good practise for warpless with fire as there are so many plants to kill (similar to regular 6-2). Looks quite playable and fun. 8-4 is great as one can skip the boring first room and does not need the first frame perfect jump (or fast accel) in the second one.

Is there a setup for the 8-2 clip? I'm really inconsistent at it

How to do 1-1 walljump: Hold L+B, when at the left wall do a good fast accel, hold right and jump the frame after you land. Release jump a short while (=a few frames) after. That should hit the walljump pixel at the second block.

And a random thing I found which is probably known: It's possible to jump the first frame possible in 8-4 and actually jump in air. Great for practising timing level beginning fast accels.

In any%: It could theoretically happen, I remember seeing taven go for 1-1 mavpole after a slow underground. That could have saved the optimal rule and allow WR. Taven has also done pl8-1 in a run, after which could have completed 8-2 through 8-4 if he didn't let mario die. There have been chances but only a handful of people even go for FPG in runs and I don't remember anyone getting a mavpole in a PB run. It's possible that it has been done in a run that was completed but had a death or massive slowdown somewhere.

If you mean any category, some mavpoles have been done in beat 5-4 recently. (minus world -1 full fpg is also a mavpole as it's a FPG without setup but I don't know if that's what you mean. If so, it has been done in a lot of runs and is quite consistent)

Smb1 goals for end of 2020:

• very low 5 or sub5 any%
• 44.7xx or better 8-4 IL
• complete a 1 warp run
• learn the few stages I can't do yet and then complete a warpless run
• do a segmented any% 4:55
• get low 4:56 pace to start of 4-2 (doing 1-1 and 4-1 FPGs) just because it feels good
• have fun

Current PBs and consistencies so I can compare to this when I (hopefully) view this again in 1 year:

• 5:13 any% (and had 5:05 to 8-4, 4:58 to 4-2)
• 44.92 8-4 IL
• 13 out of the last 100 attempts at optimal 8-1 without FPG
• 4% 1-1 FPG from savestate after underground
• I wanted to have FPG in a run as 2020 goal but I acidentally did that last year without even trying hard, on the third dead run after learning FPG

The vine method indeed has the higher ingame time counter, but for speedruns the time spent playing is measured by a clock unrelated to ingame time. The reason your wine climb way is slower in runs is because the animation of the wine appearing and mario climbing it takes much longer than the time it takes to go down a pipe.

And yes, people know this, it is even mentioned in the video about 4-2 speedrun history.

And in 99.99% of the cases such a simple thing already has been checked in a game this optimized. If you think you really found a timesave, record yourself or your TAS doing it and then play that video along the human or TAS WR for that level. Then it's really easy to see if it's faster, and if it really is it's a quite convincing proof assuming no cheats.

The 8-4 timing works now, thanks. However there now is another possibly related bug, which is not that significant but still not intended behaviour. After landing on the axe, the time was shown and mario kept running/walking, fell off the axe block to the left and got pushed through the wall below the axe as the screen scrolled. He also keeps the walking animation while usually standing while the messages appear. Screenshot: https://imgur.com/a/oN9DwJX

It only happens sometimes, I had it happen 2 out of 3 8-4 completions so far.

Where is the timer shown for 8-4 (or is it even at all)? When completing levels like 1-1 or 1-2 the time was shown to me during level transition. However after loading 8-4 from the menu and completing it I see the "Uuugh no Idea what to write here" screen but can't find the level time. Could also be useful for 8-4IL category to see it directly

Additionally to not being able to see the time after the level, it appears to not even get saved. When exiting to title and then selecting records, I'm able to see the times of other levels I completed (so it's not an general error with persistant save), but not the 8-4 time.

@Lul_ecks_dee When I did my estimate of 1 year, I assumed a sufficiently high bounty was existing and there were multiple people wanting to get it. Those factors make it happen much more quickly than without. Since runs have luck involved as noone is or will ever be 100% consistent at everything (even when excluding coin toss), it is always possible to do the 4:54 run in the first attempt, or after multiple 10000 more than needed on average. So it could take quite a bit longer. But assuming I know that I will have 3 years before anyone else gets 4:54, I would just extend everything proportionally. But I really think that if someone (that is sub5) has 1 year of 100% free time, a lot of motivation and some talent, there is a realistic chance of 4:54, even more so with a few additional months of attempts, which don‘t need to use up all the time that person has.

More time on 1-2g: I don‘t think so, my plan includes 1 hour per day for 6 months plus 1 month of 4 hours a day for 1-2, which also includes the clip. And if that‘s not enough it can also be practised in month 7. And it‘s early in the run compared to 8-1 and 8-2, so resets don‘t cost a lot of time. But I don‘t practise 1-2g so I don‘t really know 8-1 and 8-2: 8-1 FPG is included in FPG practise, fast accel is in fast accel practise so the only other thing needed to save the framerule is subpixel manipulation, which on its own is not that hard compared to fpg. For 8-2, noone really knows yet how easy or hard it is to get more consistent at it, but the fast accel part is practised, the 1st pixel bbg is 2 frame perfect inputs and the pipe jump is practised for 4 hours a day for 1 month (the rest of the level is so easy it doesn‘t need much practise).

Mindset and efficient practise are important, and I was assuming these to be already developed in my estimate. I could see spending some time on these while working on easier levels (8-3, 4-1), which don‘t need that much practise, being worth it.

First to get it is almost impossible for me (I don‘t even have a sub5, although my sum of best is a high 4:57, but I‘m really inconsistent at the fast strats without hard tricks except bbg), even if I grind the game for a long amount of time every day, the best runners will already be good at 8-1 and 8-2 when I only have high 4:55 strats down at best. Like that they will always be ahead, so grinding for 4:54 is not really worth it for me. To escape this, I would have to really grind a lot, more than any other person grinding smb1.

If I would actually decide to go for 4:54, here is what I would do:

• A job takes up too much time, so the bounty has to cover that
• With that out of the way, 10 hours per day for 300 days/year is possible -> 3000 hours of practise per year
• Other people will go for 4:54, even more so with a bounty offered, so I have to be quick. 1 Year needs to be enough -> 3000 hours time
• First 6 months: Practise only, no attempts. Spend 3 hours per day doing FPGs, as that‘s what most top level runs fail at currently. Then 1 hour for 1-2 and 1 hour for 4-2 clips, 1 on fast accels in world 8 and 4 on a specific level, one per month (ignoring 1-1 and 4-1)
• 7th month: Put everything together. Practise pairs of levels and full worlds, improve final weaknesses
• last 5 months: Attempts. If each one takes 3 minutes on average, doing 20 per hour, 200 per day, 6000 per month, 30000 in 5 months is possible.

This is obiviously very time consuming and I can imagine it to be quite boring, so someone would have to pay me quite a lot to make me do it. I‘m thinking about somewhere around 500.000€ (Or $, doesn‘t really matter).

Nice spreadsheet, I'll definitely be using this once I get into the required time range.

One question: For 638, I jump onto the first koopa (after the first bullet bill launcher), then go bottom on the first set of two bros without any killing, top on the second set of two (no avoiding of hammers needed) and then jump over all of the last four bros (very early jump or under hammer for 3rd). I think that is easier compared to your way because jumping onto the koopa is much easier than the double kill, at least for me. What do you think? (It's also possible that I'm too dumb to set the framerule correctly and above is all wrong, I don't have much experience with framerules and patterns)

The first method theoretically should always be as consistent or more consistent than the normal method.

Start by trying to jump on the last possible frame. Never risk jumping early. If you get the jump, you will hit the walljump pixel. If you don't get it, you fall off the pipe and can do the normal walljump. So if you never get the jump, you still have the consistency of the normal method. But if you get it sometimes, some attempts don't rely on the chance of the normal jump working, which increases the consistency (unless you already are at 100%). The only way you are screwed is if you jump too early, so don't do that. In that case, you could try to do a almost-full jump and try to hit the bottom walljump pixel if you are crazy, but that will lose time.

To confirm this really works (at least for me), I did 100 walljump attempts (2 frame setup) with each method:

• only normal method: 42/100
• only my new method: 38/100
• both combined: 53/100 (11% improvement over the normal method)

I am bad at frame-perfect tricks (as you can tell by my username), so getting a second chance probably helps me more than most other people. But since this walljump is so far into the run, any bit of consistency could be important here.

First method: This one uses the same location as the normal walljump, but has an easier way to get to the walljump pixel. Instead of dropping down after the pipe and jumping from the ground, you du a full jump from the edge of the pipe (last possible pixel I think). This works with 2 frame and 1 frame setup and is as fast as the regular setup (even a bit faster for me because I walk less on the pipe using this, but that may be just me doing something slightly different)

Second method (not as fast): As above, but do a full jump slightly earlier than in the first method. You will hit the walljump pixel below the usual one. This one is more difficult to do (no precise indication when to jump) and usually not as fast. I also don't know if it works with the 1 frame setup, so this is only useful if you are not going for anything super fast and find this setup easier than any other one.

These are most likely already known, but I could not find them here. Does someone know if the first one is really as fast as the normal method (using 1 frame setup on both)?