The only pixel stop that you don't really have control over is the final one on the vertically-moving platform where you morph into the shell (like in 8-4 as you said). Andre, Wah, and I agree that the second part is harder than the first part. Study Andre's written guide and our videos very carefully to see exactly what we do and how tight we make our movement. You really do have to make your movement as tight as possible; probably only a handful of frames away from optimal movement or else you get squished. If you're still having trouble with the second section, I can try to go into additional detail to help.

Good to hear that you will be recording. Once you have a PB video, I can look through it and calculate how much of a difference in loading times the R4 card has, if any. Again, I would prefer you play on a real cartridge just to be safe, but it's okay for now.

Good luck with runs! If you want me to send you resources for learning Any% or have any questions about the run just message me on Discord, Twitch, or YouTube.

We don't currently have a rule against using an R4 card, but we would prefer you play on an actual cartridge (they're not that expensive anyway).

Recording yoour DSi with a webcam or camera is perfectly acceptable. If you don't already have one of those, you can buy a cheap USB webcam on Amazon.

Desmume is fine, but since you already have a DS we would prefer if you used it.

Thanks!

I finally finished calculating the loading time advantage for VC and emu for 100%! Here are the final results:

• VC 100% times will be adjusted by 290s (4m50s)
• emu 100% times will be adjusted by 267s (4m27s)

When v1bertizm finishes calculating the loading times for Cannonless using his method, we might change the time adjustment for that category. But for now, the rule changes are done.

I finished my method for timing Cannonless, and I got that VC adds 1m29s and emu adds 1m21s. I'll wait until vibe is done timing by his method to change the rules.

yeah iOS emulators are allowed. Just remember the rules for recording runs.

awesome. Do you know the optimal overworld RNG movements for world 5?

Some updates:

Thank you everyone for voting on the poll! About 80% of people voted to adjust for loading times in all categories, so that's what I will do. I have been working on the loading time calculations these past few days with help from v1bertizm. We have been making progress, but it's going slow since we need to go through a couple hours of raw footage frame by frame to accurately calculate loads. Here are some updates/findings/rule changes:

• for Any%, VC and emu times will now be adjusted by 46 and 40 seconds respectively rather than 45 and 38 seconds
• for Low%, VC and emu times will now be adjusted by 60 and 55 seconds respectively

I should have the adjusted times for Cannonless soon. When I do, I will edit this post. 100% will take longer; it may be a few days.

Awesome! These two new things make the trick much easier.

This one isn't as bad, but I thought I'd share it because it was funny:

Accidental frame-perfect inputs Kreygasm

The problem is that a lot of people record using a webcam (or don't record at all), so the load remover wouldn't work for everyone unfortunately.

First, some background:

NSMB can be played on DS, Wii U Virtual Console (VC), and emulator (emu). VC and emu have much faster loading times, giving them a significant advantage for speedruns. To combat this unfair advantage, we previously decided as a community to add 45 and 38 seconds to all Any% runs on VC and emu respectively.

Now on to the current issue:

We decided to only factor in the loading times to Any% because it was by far the most competitive category, and no issues had arisen with loading advantages in other categories. We decided that we would bring the discussion back up if/when there was an issue of loading advantages in another category.

That time has come. GregLvk just got a new personal best in Cannonless, with a time of 46:29 played on VC. This is faster than the current world record by 6 seconds. However, it would not be the world record if we factored in loading times for Cannonless, as discussed in previous threads.

I propose a rule change: factor in loading times to all categories, not just Any%, to keep things fair for everyone. I will calculate the loading time differences for both VC and emu for all categories. Below, I have linked a poll for deciding what to do with the rules. I would appreciate it greatly if everybody voted on the poll and responded to this thread explaining why they chose which answer. Thank you.

https://www.strawpoll.me/15706434

It's more than a second. My gold is a 2:15 (lol) and you can save 3 seconds on it max. You must have split really late for 8-T2 because 2:09 is definitely impossible.

Is that including your 2:09 Bowser's castle gold? Because I'm pretty sure that split is wrong

I'm excited to see what you can pull off. Good luck TASing with DeSmuME though ? I don't know how Hartmann does it. I have even more respect for him after messing around with it.

Do you mean backup shell on 8-3? Like from the hammer bro at the beginning of the level. I guess you could go for the yump in bowser's castle, but I don't know if anyone has ever done it successfully (and complete the level).

I don't know if you should even include fast 8-2, because I'm not entirely sure ILSMB is a human ? ?

What do you mean pick up another blue shell in 8-T1? It's impossible to even get the red block on 8-T1 unless you die and have to redo a level; the red block is not congruent with your possible map locations otherwise.

Definitely don't go for any mini mushroom strats because no one has come close to doing them in a run I'm pretty sure.

For the overworld RNG, I can link you to videos of all my gold splits to see what good RNG looks like generally.

Did you check if the red block/hammer bro crossed the bridge?

Yes to both questions.

You'll beat this PB ezpz.