TASVideos

Tool-assisted game movies
When human skills are just not enough

Submission #3119: Samsara's SNES Run Saber in 13:38.48

Console: Super NES
Game name: Run Saber
Game version: USA
ROM filename: Run Saber (U) [!].smc
Branch:
Emulator: (unknown)
Movie length: 13:38.48
FrameCount: 49109
Re-record count: 5130
Author's real name: JP
Author's nickname: Samsara
Submitter: Samsara
Submitted at: 2011-04-20 08:32:20
Text last edited at: 2015-08-07 22:54:05
Text last edited by: Samsara
Download: Download (6149 bytes)
Status: published
Click to view the actual publication
Submission instructions
Discuss this submission (also rating / voting)
List all submissions by this submitter
List pages on this site that refer to this submission
View submission text history
Back to the submission list
Author's comments and explanations:

(Link to video)

Run Saber is a quite fun SNES game developed by Hori Electric and published by one of the few companies I'm an active fanboy of: Atlus. The gameplay is quite similar to Strider, in that it almost seems like a blatant ripoff.

Game objectives

  • Emulator used: Snes9x re-recording 1.43 v17
  • Aims for fastest time
  • Takes damage to save time
  • Uses death to save time
  • Plays on hardest difficulty
  • Minor speed/entertainment tradeoffs
  • Manipulates luck (kinda)

Comments

Note: These comments are not finished and are subject to change.

This is a 4562 frame, or 1 minute, 16.03 second improvement to the currently published run by boct1584.

I technically started this run a couple months ago, making a rather lazy run to the first miniboss out of boredom. On a whim, I decided to compare it to the currently published run (after learning how ZSNES TASing works...) and found that my testrun was about a second faster. I restarted on April 5th, 2011, and ended up finding a few new timesavers in Taj Base that allowed for a much faster run. I finished late April 19th, 2011.

This run uses Alow, the male Saber. Both Sabers are nearly identical. However, Alow's attack is more horizontal, while Sewa's is more vertical. Alow is chosen due to this, and due to the fact that he can hit with the vertical range Sewa has anyway.

Tricks Used

Dashing (Power-walking)

Pretty standard, double-tapping forward when on the ground or on a ceiling allows you to dash. This is done as early as possible as often as possible.

Animation Cancel

Mostly used when landing from a high distance. Hitting Down one frame before landing will cut the landing animation by 5 frames.

Damage Cancel

When in the air, taking damage normally wastes quite a lot of time. If you're near the ground, it will put you on the ground with no hurt animation, and you can act immediately. This is mostly used to save time on jumps up to platforms.

Saber Abuse

Jumping before attacking will shorten the attack animation. Alongside that, the saber can hit behind you on the first frame it's out, which is used in some cases.

Stage by stage comments

Title Screen (66 frames slower)

Switching from ZSNES to Snes9x costs 66 frames at the title screen for an unknown reason. This means that my exact run on ZSNES would be 66 frames faster than on Snes9x. However, it's a rather cheap improvement, and Snes9x is a much more suitable emulator anyway.

Everything else is the same.

Stage One: Taj Base (583 frames faster so far, 649 frames total)

I found out that slashing is 7 frames faster than using a jump attack, so I slash enemies as often as possible.

Sliding off ledges saves a frame or two each time.

Landing on a ledge like that saves a couple frames rather than overshooting and landing upright. I found a method to halve the "stun" time it takes to land, however it was halfway through Tong City, so it's not utilized here.

The jump attack to slide here saves quite a few frames. You can slide as soon as you hit the ground, and it immediately takes you to dashing speed, and as long as there's a ledge nearby, it saves time. The slide through the enemy also saves the time it would've taken to slash it and start walking again, probably about 10 frames.

The ceiling catch looks a little awkward. I might fix that in an improved version. Seems like falling straight and sliding off the ledge would end up being faster.

I like the way this room looks in the TAS. Yes, I'm sliding off each step.

There's an enemy in the way. I could slide under it, but it would end up wasting way more frames than slashing it took.

There are a couple ways to do this room. One way is to land at the bottom, then jump up the wall to the right and ceiling-dash along the platform. The way I do it is more sensible. It also contains the first use of the "damage-cancel", where if you're hit in the right spot, it cancels the stun animation and puts you right on the ground, allowing you to jump immediately. It's mostly used to control jumps onto platforms when enemies are around, and it saves the time of having to land while overshooting, or having to slowly inch your way up to the apex of the jump.

Jump-slashing here allows me to hit the miniboss two frames earlier than if I had stood and slashed.

I like this miniboss. Doesn't move around much, has an even invulnerability time (50 frames between hits), and most importantly, DOESN'T FLOAT OUT OF RANGE EVER. Attacks don't matter here, the only thing that does more damage is the X-button attack, although it's usually never a good idea to use. 300 frames for the equivalent of 3 hits, when most bosses have an inv.time of 50 frames. Either way. The last hit being angled to the left is oddly the only way I could hit the boss without taking a hit myself.

Jumping from the girder not only skips the boss defeat animation, but it also saves climbing time.

The jump before the wall miniboss was necessary to scroll the screen up, allowing it to move right.

The miniboss itself was hellish. Everything I did was basically luck manipulation, which is why there are whiffed hits and weird looking jump-attacks that do nothing. Even so, it doesn't do much. The jump after the wall is destroyed was barely enough to skip the boss defeat animation, allowing for more running.

I'm not *great* at being entertaining during scenes like this. It's much harder to do in frame advance, so this section was one at 25% speed, which oddly enough allows for more control. Game sound was also unmuted, which allowed for the musical slashing. The plane jump is one of my favorite parts. Any frame between 5900 and 5975 would be a good screenshot, I'd think. There's a weird warping thing that boct did in his WIP, but I wasn't near the right position to do it. Maybe version 2? :3

The plane boss always spawns in the same order: Middle, back, front, cockpit.

The cockpit part itself is the weird one. The time between hits varies so wildly that it's hard to tell exactly how optimized it is. The last few hits were more for "entertainment" than anything. The rapid slashing somewhat cancels out the sound of the last hit connecting. So, I guess it makes it look like I'm about to fall to my death, and then PLANE EXPLOSION.

Stage Two: Tong City (1504 frames faster so far, 1570 frames total)

It's always nice to watch that long, relaxing run right at the beginning, isn't it? Unless you're me, who still frets that a desync will occur as a random enemy pops out of nowhere on the right. It cost me a few hours of work, but it was sort of a blessing in disguise, as it allowed for a solid second of improvement over my first time through.

I'm pretty sure that sliding right when I landed on the lightning statue thing would've gotten me hit by the lightning coming out of the side.

In the sliding corridor, I slide as close to the fourth "claw" as possible, allowing me to stand up and walk through it. This, in turn, allows me to walk through the lightning rods, saving quite a lot of time. It was rather nice to be able to figure both of those out at once.

The Orb Head miniboss... Way more fun unassisted. Aiming for time is a bitch because of his movement patterns and tendency to float out of reach. Your location manipulates him more effectively than anything, and this is the fastest pattern I could come up with. I like the last three hits, all as early as possible.

Bye bye, extra life. I won't need you at all.

Red Saber (apparently known as Kurtz) will not spawn until the screen is clear of enemies, which is why I need to stop and kill them. Kurtz himself is a dick, having variable inv.times (mostly when he's jumping), and casually refusing to be manipulated into a fast pattern (again, position matters here, but you don't get too much time to position yourself accordingly). His inv.time can be anywhere from 50 frames to 74 frames. Yeah. It's not fun.

The enemy setup here is awful. You can avoid the boss defeat animation with a slide, but I couldn't slide at the appropriate time without getting hit.

"This is a weird looking area. Pretty sure it's slightly faster to take my route over jumping over the spire thing and landing on the upper platform. The ceiling catch lets me go to the ground slightly faster than landing just before. I could travel along the ceiling but it's ultimately slower, as it's apparently not possible to ceiling dash there." That's what I wrote in the comments before going back and redoing the run from this point. It's actually quite a bit faster to land on the upper platform. I have no idea what I was thinking, clearly.

Another damage-cancel here.

Jumping over the fire statue thing involves the screw attack, which has the honour of being one of the best moves in any game. I'm not entirely sure what it does aside from making you an invincible spinning beast. Well, not COMPLETELY invincible. Either way, my hitbox shrinks enough to make it through the top of the fire. Boct did it in his run, too, and no you can't just short jump through: You'll get hit. The screw attack isn't perfect, ya know.

There's a lot of awkward moments where it looks like I can do something, but don't. Alow is locked into those positions for a few frames.

Rock Tosser is boring, but that's the only position where I can hit twice per jump and not die to the rocks he tosses. I found out in my redux of this area that Rock Tosser only takes damage on certain frames of your attack. Utilizing this, I officially took him out as fast as possible and saved 10 frames.

I debated on sticking around and grabbing the blue health-up, but it doesn't seem like it'd save more time than it'd take to get it. I got a white one as a drop without losing time, that should suffice.

"Had to kill the Cloud Runner to get the screen to advance." Another part of the old comments. Getting to the bridge earlier allowed a better set of enemies here, and there was no Cloud Runner to kill this time, so the screen scrolled immediately.

Also, I utilized the screw attack move here, most notably on the bridge. It shortens the height of your jump, allowing you to hit the ground earlier. It's also used instead of slashing, as it's much faster to do so on the upward slant.

"Undead Princess is another boring battle that requires not a lot of movement. This is where speed/entertainment needs to be decided. The battle's more entertaining in boct's run, but it ends up slower due to needing to jump up to the head. This is the improved battle, which ended up 21 frames faster due to hitting earlier and not losing frames due to repeated swings. There's one point where I swing repeatedly, which costs me two frames, but it's necessary to avoid taking damage." The old battle was boring, I sacrifice 17 frames to make it much more interesting. The Undead Princess has a much bigger hitbox than her sprite details. It's kinda weird, but I don't mind one bit.

Stage Three: Jodvalley (2454 frames faster so far, 2520 frames total)

I both hate and like this miniboss. It doesn't take an offensively long time to kill, but the forced wait between hits hurts a little. I delay the final hit, as it sends the boss to the ground slightly faster. That's one of two unskippable boss-defeat cutscenes in this stage. Yes, you need to see it for the lizard thing to take the eagle head. That is a really weird sentence.

The single biggest timesaver here is taking damage on the spikes and just walking through them. This is probably the only time where taking damage to save time actually saves a sizable amount of time.

Lots of abusing the slide here. This is probably my least favorite level to play, but it's surprisingly fun to TAS, and interesting to watch.

Again, miniboss won't spawn unless the screen is clear of enemies. Not killing that fly thing (too lazy to look up the FAQ name) would've cost a few seconds, as it flies downward, into the miniboss screen, and takes a while to leave it. I'm sure sacrificing 7 frames or so to avoid at least 100 lost is fair, aye?

Somewhere during this level, I found that you can shorten the landing pose by another frame by hitting down the frame BEFORE you land.

I... really don't like the minibosses here. They're pretty standard and easy enough to kill, but I just don't like how they are identical. They also both have their own boss-defeat poses, meaning I need to skip two. These two minibosses have no clear indication of when you hit them, so the head-bouncing is not only an "entertainment" thing, it's a security thing, since I know it'll always hit first frame. I originally used this strategy for the second one only (as it skips the first victory pose), but I figured it'd be faster to use it on the first one, too. And it was, by a couple frames, so yaaaaay.

If there's one thing that frustrated me about this stage, it's the mecha-lizards. They're hard to avoid, and screw attacking through them isn't an option considering they jump up in place when hit, meaning you have to overshoot your jump or risk getting hit (and in my case, dying). In the case of the second one, he just happened to be coming toward me, which allowed a good jump through.

Nope, can't go below the screen. It might be a bit faster to scroll the screen over before dropping down, but I like how it makes Alow seem like he's floating for a bit.

It's not necessary to kill the first mecha-lizard but the screen won't scroll for that amount of time anyway, so why not.

Hate hate hate hate this miniboss. I attack him quite a bit earlier than you would normally expect, as he's vulnerable the frame the actual miniboss forms, which is a nice timesaver over the published run, but he has the Kurtz-like variable inv.times where it just randomly takes 6 more frames to damage him. It's annoying.

Welcome to the most frustrating part of the run. It looks relatively smooth, but I had a horrible bit of trouble with the mecha-lizards here, especially the one after the drop. Getting past him was a bitch and a half, but what I came up with was decently fast enough, I suppose.

The rest of the path up to the boss is pretty standard. I hate landing on slopes, since you have to sit through the full 8 frames of the landing animation.

The boss. It's boring, but it's the fastest way to kill it. A couple seconds were saved here due to shorter jumps, which saved 6 frames each time over a full jump. I call it the 7/3 jump, because why the hell not. B for 7 frames, idle for 3, then hold up. Hits the boss at the end of his invulnerability and lets you land sooner to boot.

OH GOD I DIED RUN REJECTED FOR SLOPPINESS. The death was intentional. It wasted absolutely no time (since I died during the long boss death animation), and it gives me back my health. Plus, the death animation is kinda neat to see.

Overall, this stage is 950 frames faster than the published run, and 2454 frames faster in general up to this point. Adding the 66 from the title screen means I've saved 2520 frames: Exactly 42 seconds. I'll take that as a good sign.

Stage Four: Grey Fac (3433 frames faster so far, 3499 frames total)

I like this stage. I just wish it could be longer, with less terrible minibosses.

The fire geysers are on a set pattern that's actually rather forgiving. I don't have to wait long and I can pass right through the rest.

I discovered the fastest way of climbing here, meaning the first stage of the run is definitely improvable. Version 2 might be a possibility, now.

This is where the game starts throwing enemies at you left and right. It might be faster to take out that group of five with a longer screw attack, but that would mean having to fight the Engine Knight again...

I'm glad the enemies before Engine Knight cooperated in the way I wanted. I needed the sword powerup back due to the earlier death, and they gave it to me in a way that only cost me a couple frames. And yes, they need to be killed for Engine Knight to spawn.

Hate this miniboss. The hit I took was to get me into a better position to keep him grounded. I like how the fight turned out, I just hated optimizing it.

Actual platforming in this level! Not really much to say here, except that one screw attack is faster than two swings of the sword, and hitting all three of those aliens would've taken two swings.

Kurtz again. He takes 16 hits this time. I take another hit in this battle for positioning and manipulation purposes. There's a noticeable delay around frame 30460 where it looks like I can hit him but I don't. This is due to his inv.time being terrible and not letting me hit him even though he's done flashing.

See? Stage boss already (after a bit of straightforward platforming). The chase sequence is boring, not much can be done to spice it up. The first form of the boss itself takes around 20 hits or so before it starts moving upward, although any damage it takes during this time carries over between phases. I land about 27 hits on it before it starts moving, saving quite a bit of time. The boss has three points where you can hit it when it starts moving upward: Head, upper torso, lower torso. I hit these all as they were available, although I'm not sure which one actually affects the health of the boss, or if they all do. Regardless, I take the boss out much faster than in the published run. Half the improvement in this stage was on the boss alone.

Overall, I saved 979 frames in this stage, bringing the total up to 3433 frames saved.

Stage Five: Bruford (4562 frames faster, 4628 frames total)

Best music. Pity that it has to be covered up by the miniboss music so often. This is the most sizably improved area of the run, with 1129 frames saved here alone.

The first enemies here are rather annoying. I wait around to collect the health they drop, intending to do some damage cancelling later on, though I didn't end up needing it.

Snakelike bosses aren't that fun, as it's rather difficult to tell if you're hitting them or not considering they have so many hitboxes that can get hit on any frame, regardless of the invulnerability time. There are two in this level. Yeah.

Holding on to the ceiling skips the jumping down animation, saving a good number of frames.

This is a long fall and there's not much to do to make it interesting, so I bring back the musical slashing. It's a little off due to the game's music being somewhat glitchy. Also, oddly enough, the heads that spawn don't bounce off the walls on Snes9x like they do in ZSNES. Huh. Doesn't make a lick of difference, it's just a little strange.

Orb Head is back and more annoying than ever. There's much less room to position yourself properly, so I end up spending a lot of time not hitting him so I can manipulate him into a better pattern. I manage to manipulate him to go over the platform on his last hit, which saves the time it takes for him to drop down. It would be faster if he was in the wall, but I couldn't manipulate that out of him.

The elevator sequences are sped up by the rapid jumping, and it's one of the only places in the game where you can start a jump two frames before you land, making it seem almost seamless. The elevator gets no chance to slow down, so it stays at that sped-up pace the entire way up. This is used three times and makes up most of the improvement in this level.

Final Kurtz fight, and he was being even more of a dick than usual, refusing to be hit once every 50 frames and often taking 64. I managed to save 30 frames here, not counting the frames I saved from taking him out on the ground, so it's likely that more can be saved.

Directly after is the Abomination, my least favorite boss. It's stationary, which is boring (doesn't allow for much playing around), and it has a snake head. Its attack pattern wasn't all that great, so I lose some frames here, but it's a small loss.

And directly after the Abomination is the final boss, and boy do I hate how this battle looks. This is the reason I stockpiled health, because it's faster to take the damage here than to spend time avoiding it. It's a pretty standard, seemingly sub-optimal fight, but I can promise I hit the heart as early as I could in every situation. I end the input as early as possible here, on the same frame as the second to last hit on the boss, with the last hit connecting 290 frames later.

Yup. Ended with a bang. A flashy one at that.

Potential Improvements

Two Players?

Some people have brought up the idea of two-player, although two-player is rather laggy from my experiences. It would allow for faster boss fights, but the added lag may negate that time.

Known Improvements

Overall Optimization

I found a few frame-saving tricks later on in the run that aren't incorporated into the first levels. My determination to finish this run wouldn't allow me to go back and incorporate them into the earlier levels, so I estimate a few seconds can be saved through doing that.

Boss Manipulation

It's a pain in the ass. It's entirely determined by the player's position, which I didn't find out until Tong City, and I've been hesitant to really grind at it considering the bosses are easy enough anyway. I mostly used it to manipulate the more mobile ones into better patterns, although I assume it can be used to cut down on randomly long invulnerability times as well.

Final Comments

Finishing this TAS fills me with a sense of pride, and it's calmed me down immensely even through my various crises. It was a great experience, something I'd like to do again for sure (Terranigma is calling me). Even if this is rejected for whatever reason, I'm still proud of myself for working through this, and as glad as I am that it's over with, I can't wait to start on another project. This paragraph probably makes no sense. Oh well.

Suggested Screenshots

Action

5960, 15796, 22148

Glitchy

32987, 33486, 47366


klmz: Claimed.

feos: HD encode.

klmz: Accepted a an improvement to the published movie.

Nahoc: Processing...


Similar submissions (by title and categories where applicable):