this is way harder than a walljump in smb1 tbh. On a good day, I can get fast accel walljumps 3/4 times, sometimes even better, But I cant get anywhere near as consistient at this trick. However, I have investigated my input faults on emu by making a movie of just me failing fast 8-2, and it turns out Im actually releasing B early relative to when I release X and press A. ILoveSMB, by any chance do you know the number of frames it is from the point you release A in the initial jump, to when you repress A and release X to activate the doublejump? I wanna make This semi consistient, cause at this point, Id compare the difficulty of fast 8-2 with mavpole

eh, I think somebody will get a low 23:1x, maybe me, probably walrus

Im probably not gonna have w5 done for a while, I have school to work on and also RNG manip is ass. Dont worry tho, shouldnt be delayed too much

no, but Ill study your gold

update: W1 is completed. the split with loads factored out is probably roughly 3:38.3

btw this doesnt replace fast 8-2, fast 8-2 can save up to 10 seconds. this only saves 3. however fast 8-2 sucks and this is cool

wait I just realized If I ever get fast 8-2 3rd try or better, And finish the split, Ill have a 22 SOB. thats fucking nuts

ok through some testing Ive determined that rng changes by date/time roughly every 0.3-0.5 seconds, So I just need to find a window that works, this is nice progress. My current setup is manipulating warpless luck, so I need to just mess around with the date/time

also I might know why your rng manip isnt working zach but Id need to see footy of you playing to confirm

Im gonna begin RNG testing, I have a VC, EMU, and OG console copy of the game, so I will be going to be looking for everything. Im no goddamn genius, and im a shite speedrunner, but I can TAS for my life. If I find this shit, I will learn and go for fast 8-2 in runs. sub 8:05 pace out of w5 is free with rng manip.

in hundo helixfossil is actually quite competitive for WR Tho, and a lot of other categories are easy to factor out

ILSMB is hitting these 50000000000000000IQ points RN. Also I choked a 23:18 to 8-7, should prob post here

We should have a standard measure of adjustment for loads for every category/platform to make it identical to DS loading times. ofc it will be thrown out of wack when unnecessary deaths are taken, but enough unnecessary deaths to offset the load adjustment enough to be actually unfair would probably make your run not competitive anyway

yeah probably

I mighta split 1 second early

human SOB is a 22:54 im pretty sure

oh yeah also my gold on bowsers castle is actually a 2:09.9, dont quite know why I typed a 1 instead of a 0. 2:19.9 would be a really terrible gold. Its insane to think that the TAS does that split in under 2 minutes.

yeah for some reason the TASvideos rules are completely contradictory to those reccomended by the developers of the emulator. TASvideos needs to get its shit together fr, they rejected an optimized NSMB DS low% TAS because it was too boring. Like, excuse me sir? Did you even bother to watch the damn thing? Some of their rules and regulations have just become stupid and unnecessary at this point. I however do agree that 9.9 should be the used emu, cause despite being one of the laggier builds, it is the most console accurate

Turn off advanced bus level timing and run on 9.9, if you run on any version other than 9.9 and/or have advanced bus level timing on, it will desync

If you playback with the same emulation settings as the recording, there is no reason why it should desync

Desmume crashed while I was recording inputs for the TAS so rip the half hour I spent manipulating an extremely rare redblock pattern. How the fuck hasnt hartmann been drivin clinically insane by this shitty emulator. This is nowhere near as precise or optimized and I already hate it