Backing up. Rerecord count is low but will likely spike hard once I start actually getting to the battles.

Reaches the room where David is fought for the Stradivarius without any RNG manipulation... i.e, David ain't there.

55 on Practice using #6934: Evan0512's Genesis Marble Madness in 02:34.31 with minor input adjustments. Definitely not optimized. Frames can also be saved on menuing.

stress TASed up to getting jailed. not really optimized. figured out how to manip away overworld encounters though. attempting stradivarius route.

Using superjump, in-level warps, and stage clear glitches.

Uploading for comparison purposes. Also, uh, spoilers, new Splatterhouse run coming soon. Five frames needed to be added.

Stage 2 boss. Can't seem to use fire wheels the frame after jumping and I have no idea why.

EDIT: This actually isn't the final branch (it enters the boss 1 frame later than optimal) but it doesn't matter since a 1 frame delay is needed to get the boss kill to work (and another 1 frame delay is needed on top of that). I'll just be moving on with a two frame loss to what I can only assume is emulation differences.

Team 3 (Bear and Bird): 1, 5, 6, 8, 12, 15, 17, 20

Team 5: 2, 3, 4, 9, 10, 11, 13, 14, 16, 18, 19

Team 6 (Team 5): 7

Team 1 (Sisterhood of the Travelling Frames): 21

i'm beginning to think there might be a problem with how to classify ess enn ee ess test program dot ess eff cee

Hot damn!

46 frames faster than goofydylan8's abandoned(?) WIP up to the end of Stage 1. Boss playaround probably not final.

That's a long, useless title!

Same exact strategy (with faster outs), no player editing, can still be improved.

Still absolutely destroying bosses. 3621 frames ahead.

Thank Christ. 2541 frames ahead of the published run (42.35 seconds).

Gamegear Heroes got nothin' on this.

Backup. Partway through form 2 of 7.

Backing up Green again.

Another 9 frames saved. Why haven't I just taken the Block 1 from the Grant submission and tried to improve that?

Not much better though. 5 frames saved.

42 frames faster than Random's cancelled submission at the end of Block 1.

Backing up first minute-ish of Green's stage because I'd really rather not lose everything and have to redo it.

718 frames (12 seconds) ahead of the published run. Most of the improvement comes from better boss/miniboss strategies.

boy howdy gosh and golly i love me some weapon management

dragster is the ultimate stress test because it tests how stressed i can get over a video game older than i am

This movie tries to recreate the currently published run in BizHawk, but loses frames everywhere due to needing to do different things to get the same results. Not sure if it's more accurate emulation or just my general incompetence. Note that the file is automatically 5 seconds longer due to the GBA BIOS, actual time lost compared to the published run is ~10 frames.

Stops before Drac fight.

Old Axe Armor "hard, level 1", resynced.

16 frames faster. Everything up to Behemoth chase is arandomgameTASer's input.

.15 seconds faster than TASeditor in level 2

Global timers are the bane of my existence.

6 frames faster than published run by direct comparison: Actually about 19 frames faster but 7 are lost due to the emulator and 6 are lost due to enemy spawning being on a global timer. Because ugh.

Red's shouting is basically how I react whenever anyone asks me "Hey Samsara, do you like Gunstar Heroes?"

No comparison to the published run, this is only a test anyway.

speaking of garbage you can't understand, SNOOPY MAGIC SHOW

Had to copy-paste my original input to a new blank file to properly switch the core from Performance to Compatibility, which wiped the re-record count. Actual count is ~5000.

Updated watch file using the addresses listed on the Game Resources page.

shouldn't crash bizhawk anymore

Currently 155 in-game frames faster than the published run: 129 in Field 0, 24 in Field 11, and 2 in Field 23. Still working on Field 23.

I think I'm done now. Please no more improvements.

shoutouts to Tompa for 1 frame improvement

Not synced past frame 11623, coming back to this after a long rest

it's like 31 frames faster or something

thank you atma

Backup. Through JodValley.

Removed an extra frame of lag that slipped in by mistake. One frame improvement.

This is a couple years old at least.

Another cheap 1 frame improvement, though this one required a tiny bit more modifying to make work.

Tried a few alternate strategies, then gave up and improved rchokler's improvement.

Posting this here since it's been a while and I haven't wanted to do any work on it. I've been working on this on Bizhawk 1.8.1 so if it doesn't sync on later version then use that. It should sync just fine on both the (U) and (J) versions, though it was specifically made on the (U) version ROM.

Feel free to take a crack at completing the run: The thread in the TG-16 forum has my reasons for wanting to take a break from the project.

I'm starting to figure it out.

I implemented a further route improvement found after collecting the bomb, saving about 10 frames. The rest of the stage is almost exactly the same as the published run. All the difference will come from Stage 2 and beyond.

This won't be completed for a while, but it was highly requested so I figured I'd at least start it. This is only a second slower than the published run up to this point, even though the boots are collected. There are two more rope boosts that I missed in the published run, so there may be an improvement to the health underflow run in the future, depending on how easily I can implement these two rope boosts and whether or not some of the other ideas I have will work.

The rerecord count is, again, carried over from previous files. I'll edit it away as soon as I remember to.

3237 frames faster due to a new skip in the Prize World. This is the last time I'm touching the run.

30 frames faster.

Emulator: FCEUX 2.2.2 (with TASeditor)

Had a lovely July 4th and didn't really have much else to do so I, uh, spent the last 8 hours working through the rest of the game.

It's decently optimized, though there's still a lot of work left to do: The prize world code could very well end up different in the final run, fadeout lag needs to be optimized further, I still have to do some testing with some of the tricks I found, and some of the playaround sections could be more entertaining. Thankfully, the game behaves nicely and it shouldn't take more than a day or two to finish up. I'm expecting a final time of around 20:15 after full optimization unless I find something big.

This run gets up through Stage 3 976 frames (00:16.26) faster than micro500's WIP, and finishes the game 29,712 frames (8:15.19) faster than CeSaR74's rejected submission. The prize world code entry is all micro500's input since I started working off of his WIP, so if it remains intact by the time I submit the finished run (which is highly likely), then he will be listed as co-author.

203 frames ahead. Most of the time saved is from skipping the boots, but there are small movement optimizations all throughout. The invincibility glitch is activated near the end of the stage.

This is probably the most boring item skip.

The boots apparently aren't necessary to beat the game, so it seems, and in a TAS using the invincibility glitch it ends up being around 80 frames faster to skip them. However, it could be possible to save more with a different damage route and more optimization (this was only a quick test), and the enemy/platform patterns later on in the stage could be better or worse.

TAS ADHD is a serious disease. Serious cases can lead to never finishing a run, letting people down, and claiming that everything in Super Mario World is memory corruption. Made in FCEUX 2.2.2 using TAS Editor.

I used the passcode entry from micro500's last posted WIP, which is why the rerecord count is that high. I only used about 723 rerecords from the looks of it. Stage 1 of that WIP syncs in FCEUX 2.2.2 but desyncs in Stage 2. Like everything I've been doing lately, if this run were finished it would not be submitted.

Compared to that WIP, I save 5 frames (I climb the rope 2 frames faster, kick one time less, and something around getting the controller gives me the other two frames), but I lose 3-4 of them (1 frame of lag and a couple frames delaying before that bastard caterpillar). I'm trying to find a way to manipulate that thing into shooting twice like it does in the other WIP. There's a way to manipulate it to shoot later and a way to manipulate it to not shoot at all, but I still have yet to figure out shooting twice.

Either way, I don't have enough health for later, so when I come back to this after the several other projects I have going, I'll start from the beginning.

Through Stage 4. 153 frames faster than my test run so far.

• Stage 1 faster through better control of Superman and lag reduction.
• Stage 2 faster through lag reduction.
• Stage 3 faster through optimizing clipping through the laser beams and a different key route.
• Stage 4 faster through lag reduction.

Optimized will definitely be sub-6.

http://tasvideos.org/forum/viewtopic.php?t=15339

Merely a test, not optimized.

I did say I had a few frames left to save in the stage itself... 89 frames is a few, right? New strategies, tighter movement optimization and pixel positioning, and reducing 41 lag frames (meaning less charged buster shots and less damage taken).

Stops right before the Star Man fight.

343 frames faster than HTWT's rejected submission, 148 frames saved in the stage itself and the rest on the Star Man fight. The Star Man fight is actually 16 frames faster than Mothrayas' when timed from the first hit to the last hit. However, the faster Star Man fight actually wastes far more frames, likely due to Star Man's lower positioning and the fact that it seems to take far longer for the boss explosion to despawn, so I'll need to redo the entire fight and manipulate him better. There are probably some frames to be saved in the stage itself, too, so I might just redo the entire thing.

Okay, bye.

Completes the first five levels and goes up to the first boss(es) of the 6th. 162 frames ahead of the currently published run, again due to reducing lag and optimizing movement as much as possible. I'm estimating a total of 200-220 frames saved by the end of the run. Made entirely with TASEditor in FCEUX 2.2.0

Completes the first 4 levels and goes through a few screens of the 5th. 125 frames ahead of the currently published run due to lag management and pressing Start to end levels faster. Made entirely with TASEditor in FCEUX 2.2.0

Improvement WIP, abandoned in favour of a much cleaner and improved version that I'll probably never start.