There should probably be a rule against it, as it is one of those things that would bring the run down to "Who has the fastest computer" rather than "Who has the most skill and/or dedication".

I want Kaleidoscope added.

We never got a third person to record a run.

And yes I support adding pogostuck

Scratch might be doing something weird with prioritizing certain scripts over others... looking at the scripts, I might be able to restructure them so the timer is directly increased by the movement scripts itself, but that might be hard as the movement scripts use a lot of "Repeat until" or "Wait until" blocks that would prevent the timer from increasing correctly.

Uh... I might have done something wrong with the autosplitter. As you missed the upwarp, this should definitely be slower than my run... The in-game timer might be messing up again. I'll need to check the code.

Again, my BSF time is invalid because I accidentally used lag to mess up the in-game timer. It can't be used. Use your time for it.

Congratulations! It's insane how you can have such clean execution. My runs always have minor mistakes in them. While there are still small timesavers (1-walljump pre-upwarp thing, trickjump, etc.), you were able to gain that time elsewhere just through clean execution. Honestly, I don't think the time can go much lower. I think a TAS would at best be able to do a 8:36 or something like that.

I agree, math is always easy. I'm four grades ahead in it.

My timer is slower because my game is lagging. Everything should be lagging at the same rate though, so the lag isn't an advantage.

That glitch is interesting. It appears you're on the world map and in the game at the same time... I'll have to experiment with that and see if it will be helpful for the run. Can't do it now though, that will have to wait for around 12 hours when I have time.

I know I haven't beat CBE2009 in CP or BSF, but I probably do have a better run for SAM or L City. I'll look through my runs. As for a TAS, that could be difficult as the game is pretty complicated. I'll do some ILs for SAM and L City, as they are definitely beatable.

Yeah video audio not being present is allowed, all of my runs have it missing (My computer updated to allow audio, but currently I haven't gotten a run with that). Also the SRC rules say to keep everything PG-13 so some swearing is allowed.

I don't have time to verify it at the moment, will do that later if someone hasn't done it by then.

v1.3 is now here. What it did: Re-fixed split timers showing up in segmentless mode, added an option to not reset timer upon green flag, made it so turning a reset option on auto-resets the variables. Also fixed default settings.

Honestly I'm not sure what the setting would be used for... count-to-25% is really the only thing that comes to mind

Here's my (extremely inoptimal) run:

Removing space and win trigger too, but those are obvious.

You can use v1.0 if you want. I personally would add both, but if that would junk up the leaderboard with "too similar" runs you can do just one

Player hitbox, removing space, the win trigger, the respawn trigger, edges, the level itself, and the level spawning scripts are the main things. The level spawning scripts should be fine if the level is ported over as a sprite instead of costume-by-costume.

The arrows (backlevel) and background need the same treatment.

I also: Removed the water and the sound that goes with it, re-designed the pot, removed all narration (normal and post-fall), changed color of the numbers of timer, changed many of the costumes in the "splash" sprite, removed the stars from the win screen, changed the music, replaced the "narration" setting with a "music" setting, changed the color of the cursor, changed the thumbnail, and removed the background scrolling on the menu.

The hitbox is changed so the player can't slip through certain gaps. The triangle size present is the only one that works nicely with the chunk size, so the best way to make the player not slip through those gaps was to heighten the player hitbox by 2 pixels (approximately).

Also I was originally trying to port the map into the standalone recorder (not the modpack), which is why I thought that recorder to map was easier than the reverse. If you're planning to put the map in the full modpack, than yeah map into modpack is going to be a lot easier.

I can do one. Also your run is technically invalid, as the player hitbox is slightly modified in my map. I'll port the map into the standalone recorder because I remember everything I changed.

Edit: There's so much I changed that it might be easier to do the reverse and port the timer/recording feature to the map. I don't know where those scripts are, so someone else will have to do that.

Well, in those, nothing in that planet uses the "wait" or "glide" block, so everything is slowed down. Same goes for everything except Brobot and Bowser. So the only things that still need retimed is mine+Jalex's 24-bolt runs.

I have the autosplitter updated. I also did the same for the Brobot battle, although that was basically no timesave. I'll retime my runs now. You also might want to retime Jalex's runs, as he also had a decent amount of lag going by the fact that his RTA-LRT was also above 45. (I can't because Twitch is blocked on my computer)

Okay - I've got that part done. https://scratch.mit.edu/projects/927682316/

It currently desyncs at the moment because Scratch decided to run above 30 FPS...

Oh - I finally realized what the problem was.

Mario and the timer run based off the in-game frames. This means that they are slowed down by lag.

However, Bowser's actions are based off real-time. You can see this because he is using the "Wait ( ) seconds" block. This means that he is not slowed down by lag.

So the timer is running slow, but bowser - who controls the switches and other stuff - is not.

I think that the best way to solve this is to edit my run with RTA bowser, which will set it to around 8:49. Then, I could edit all of Bowser's actions to be based off the in-game frames. Is that okay?

Edit: Actually it might be hard to edit all of them - a decent chunk is based off the glide block, which uses real-time. That could be a problem. I'm going to try to see if I can use some trigonometry to make an alternative block.

Well, in Mario 64 there are so many stars that each new category brings a completely different route. 1 star is a glitch showcase, and 0 star is an even bigger glitch showcase. 120 star is our 24 bolt, 70 star is intended%, so 16 star is the only thing that seems a bit strange. However, unlike 21 and 22 bolts, 16 star has a large difference in route from the other categories.

As for the RTA/LRT thing, I'm honestly not sure why it wouldn't work. There could small optimizations that I have in my run (like better-timed dives to land quicker) that you don't. If you send me a run, I could see what the difference is.