Fighting Game Tool-Assisted Speedruns

[Maj]

Lately there have been a few tool-assisted speedruns released for console versions of fighting games, usually on SNES and sometimes on Genesis. They're just full playthroughs of classic fighting games in arcade mode on hardest difficulty, usually starring the most versatile characters (meaning the ones who can copy opponents' moves). The people making them obviously have speedrun skills but they aren't necessarily experts at combo engines. Anyway i thought i'd compile a list of what's out there right now. You need a BitTorrent client to download the video files.

Genesis Pit Fighter (July 1991) in 06:31.53 by Mário de Almeida Azevedo Junior (aka FODA) (starring Ty)

SNES Super Street Fighter 2 (USA v1.0) in 11:21.47 by W.S. (aka Saturn) (starring Zangief)

SNES Super Street Fighter 2 (USA v1.0) in 11:32.43 by W.S. (aka Saturn) (starring Ken)

SNES Mortal Kombat II (USA V1.0) in 12:21.57 by M.S.M. (aka Samhain-Grim aka VANDAL) (starring Shang Tsung)

N64 Mortal Kombat 4 (USA) in 09:14.72 by Alex R. (aka Xyphys) (starring Shinnok)

SNES Killer Instinct (USA v1.0) in 09:59.18 by Flávio Garcia Soares Rodrigues (aka Dark Fulgore) (starring Spinal)


Genesis Streets of Rage 2 (USA) in 30:43.82 by Gavin Ward (aka SprintGod) (starring Max)
- side-scroller by Sega, featuring fighting-game-esque combat and 4 diverse characters

SNES X-Men: Mutant Apocalypse (USA) in 15:45.62 by Alvaro Augusto Peixoto Júnior (aka nibelung)
- side-scroller by Capcom, featuring fighting-game-esque combat and 5 different X-Men (all of whom appear in the video)

GBA Dragon Ball Advance Adventure (JPN) in 44:34.07 by L.V. (aka AnotherGamer)
- side-scroller by Banpresto, but if you were only shown a clip of any boss fight you would think it was a (helluva broken) fighting game

[Maj]

SNES "quirkfest" Mortal Kombat II (USA v1.0) in 17:54.45 by M.S.M. (aka Samhain-Grim, aka VANDAL) (starring Shang Tsung)

Anyone here into MK? Are these actually good? The speedrun format doesn't really suit fighting games, but they manage to do some cool things here and there.

[fullmetalross]

I haven't actually watched the ki one yet. But it seems by taking advantage of cinder's easy infinites and manipulating luck so they happen at the soonest possible time of the game, you could kill very quickly... but i don't know I don't play KI.

[Maj]

Well, i don't care so much about completion time as i do about technical skill. The two SSF2 speedruns aren't all that crazy, except for a select few portions.

Ken mostly mashes jabs and does absolutely no light attack jump cancels. Most of the time he's just doing reactionary DP's which isn't impressive at all if you're going frame by frame. Zangief works out way better because at least the random counter moves are actually random. The last match against Bison is really cool cuz he stuffs everything that Bison tries using a different obscure counter. Also, Zangief does one cool thing on the brick wall bonus stage: breaking the segment he's standing on and then doing j.HK as he falls to break the last remaining piece.

Just wondering if the MK/KI stuff is more advanced. So i guess my real question is, did the authors do their homework before making these? Cuz the SSF2 author clearly didn't bother looking through any SSF2 combo videos before making this. Kinda goes back to what i said (in another thread) about the emu scene having a tendency to start from scratch and ghetto it up for a while.

[Maj]

Ok, these aren't fighting games but i'm gonna mention them regardless, cuz really enjoyed watching them.

Genesis Knuckles in Sonic the Hedgehog (Revision 01) in 14:45.95 by JXQ
- hacked version of Sonic 1, allowing play as Knuckles in a game where Sonic didn't even have spindash

Genesis "100%" Sonic 3 & Knuckles (World) in 45:29.6 by Jody Northup (aka Upthorn) (starring Knuckles)
- collects all of the Chaos/Super Emeralds (boring) which allows him to transform into Hyper Knuckles (exciting)

Genesis Aladdin (USA) in 14:20.73 by M.T. (aka slash_star_dash)
- one third of my childhood (alongside SoR2 and TMNT Hyperstone Heist) wrapped up in one neat 14-minute package

I'd also mention the Genesis X-Men 2 speedrun, but there's something fundamentally wrong about beating that game in under 25 minutes without ever getting killed. Everything about that game was creppy as fuck when i was a kid, and making it look easy completely ruins the atmosphere of constant unrelenting impending doom.

[Magnetro]

the sonic 3 & knuckles speed run... ....I... I had no idea over half of that stuff was possible...


Wow. to say the least











sheesh

[Xenozip.]

It's tool assisted. Meaning they were playing the game frame-by-frame. Literally pausing at each frame and inputting commands at each frame.

Boo.

[Mike Z]

Having frame control is the same as using macros or program pads - all it really gives you is superhuman speed and the ability to try something as many times as it takes with the exact same conditions and inputs. It eliminates the randomness and lag of human control.

Magnetro will kill me, but TAS's are just as valid as that stuff. The only things humans supply is the thinking. It's making the video for a different reason - not "look what you and I can do" but "look what can be done if you stress the right parts".

Mike Z

[Xenozip.]

It's a shady subject IMO, but here's something you may (probably) or may not be aware of: thanks to -undisclosed source- some tools allow you to save states and "undo" to return to a previous frame position while still recording inputs. This allows you to re-try over and over until you get the desired result per frame. Essentially, you can be psychic on a frame-by-frame level and always get the desired results from random variables.

On one side I see it being of great value and benefit because some one could try and create a combo 'over and over and over', always failing due to extreme execution requirements but succeed using a tool. TAS' allows you to find out what is technically possible regardless of weather it's humanly possible or not, which gives you and the viewers something to strive for at least.

On the other side I completely and entirely disagree with TAS' being the same as what is possible through trial and time via normal means. There is no way that you could ever humanly produce the same results as a TAS of games like Tetris, Megaman, Mario, Final Fantasy, Metroid, Sonic, etc (to name a few series -- give me some time and I can find examples that are humanly impossible to be on par with a TAS). And this is why I have a problem with it.

Anyone can pick up a tool and have the damn thing do it for them via trial and error on a frame-by-frame level. Although it's time consuming and entertaining to watch there's no skill there, no brains no brawn so to speak. How is that possibly a good thing?

Now, I could certainly see it being put to good use. I understand what great value it could have to benefit the community if properly used by responsible people, to both authors and viewers. But I don't see it being a good thing in general as it's far too easy to misuse and creates illusionary ideas about the games they portray.

And I say this based off a lot of experience with the subject. The "problem" originated from the Quake Done Quick series when players started using tools to record demos at very very slow framerates while recording demos which allowed them perfect execution and timing, and then later removed frames to make it appear as though the game was running at normal speed. And I believe this is also why Radix over at SDA is so adamant about not tolerating TAS runs.

I tried one of these tools for a Shoot-Em-Up (shmup) not too long ago, just to test the extent of the stupidness. I was able to beat it without losing a life having never played the game before. I then tried playing it real-time and the results were laughable. It made me realize I was able to achieve something with a tool that would normally take hours and hours and hours of practice to achieve. There's also that other little factor called skill -- my TA run looked like I had perfect execution and precognition which was impressive at real-time, but I didn't actually do anything smart or innovative, nor did I display any "real" skill outside of execution (skill != execution).

If a Tool assist is used to win every single hand/round at: poker, blackjack, chess, go fish, hearts, heads up, etc. --- is it a good thing?

Personally, I don't think it is.

[Magnetro]

Having frame control is the same as using macros or program pads - all it really gives you is superhuman speed and the ability to try something as many times as it takes with the exact same conditions and inputs. It eliminates the randomness and lag of human control.

Magnetro will kill me, but TAS's are just as valid as that stuff. The only things humans supply is the thinking. It's making the video for a different reason - not "look what you and I can do" but "look what can be done if you stress the right parts".

Mike Z

DO YOU KNOW WHO YOU'RE TALKING TO >=^O

Of course I understand it, but it was like seeing something you'd never thought possible in MvC2 or any other game. It was really cool looking, I understand the purpose of macros for stuff like that though. It's just like marvel.

c'mon!

[jchensor]

Just curious, then, Xenozip... what's your opinion of Sai-Rec, Meikyousisui, kysg, and the like? All of them are programmable controller players, and I am, in particular, a huge fan of kysg's. The stuff he comes up with is awesome, and you wouldn't ever get to see those things if we didn't have programmable controllers and only relied on human input.

Is it preferrable, then, that combomakers like kysg announce and fully admit to using programmable controllers? Are you worried about people who use them but do not admit it? Or do you just feel they take the essence of combo videos away?

- James

[Maj]

Anyone can pick up a tool and have the damn thing do it for them via trial and error on a frame-by-frame level. Although it's time consuming and entertaining to watch there's no skill there, no brains no brawn so to speak. How is that possibly a good thing?

That's certainly a valid concern, but i think that the format of the main TAS website alleviates a lot of those dilemmas. That whole community is inherently competitive, because everyone is constantly trying to improve the speedruns for the more popular games. For example, you really can't get away with doing a braindead tool-assisted speedrun of any popular Sonic game. You have to understand the game really well, be totally up to date on the latest approach to playing the game, and you have to innovate.

For games that nobody really cares about, i'm sure there are some instances of people cranking out a tool-assisted speedrun without much effort. But there are also lots of instances where some random kid absolutely loved some obscure game when he was growing up, so he puts a lot of effort into the speedrun even though there's no competition to get in his way.

And even in these cases, there's a user feedback system in place to decide which submissions are good enough to get published. If the community decides that it's polished enough, then the player has to go back and bring it up to par.

[Xenozip.]

-snip-
- James

I'm very glad that you brought up specifics, like KYSG. Yes, I absolutely feel there should be a disclaimer that the video is in fact tool assisted. If the content is displayed vaguely or without any information at all then I personally feel that it's very deceptive.

I'm also a fan of KYSG for those exact reasons: not only are they very clearly tool assisted but it's even mentioned in the videos that a programmable HORI was used.

I would consider it incredibly taboo to try and pass off a TA video as non-TA.

That being said, I take KYSG at face value. I enjoy watching them as a source of entertainment and nothing more. Kind of like watching invalid combos that the opponent could recover from or combos that require an impossible amount of super or some such. Clearly it's useless, but entertaining no less.

Like I was saying in the other thread, I feel people can "get the wrong idea" if a disclaimer is not properly advertised. If all we ever saw out of SFA3 was invalid juggle combos that the opponent could tech flip from, then what would you expect people who've never played the game to think? They might not even try the game.

-snip-

Yes I agree that the competition has taken it to new heights. Like with Metroid, which my brother and I are fans of. Certainly they have improved the Metroid runs by leaps and bounds since they originally started doing them.

And also, if not for -undisclosed person- hacking the data that was processed from Super Metroid the exact mechanics for speed boosting might never have been found (the game only checks every 8th frame).

I bring up Metroid for a very specific reason: recently there was a non-TAS done of Metroid that was done in a fashion to mimic the fastest TAS as closely as possible.

Know what? The non-TAS is infinitely more impressive in my book, dispite it being full of flaws and slower than the TAS.

Point being, like I said before I can totally see how the two can compliment one another, but I feel they should be kept very very separate and treated as a novelty at best. In the genre of fighters it could lead to serious discoveries, but then you know about entertainment and realism...

I mean, think of movies. There's certainly value in unrealistic kung fu movies, but then they should also not be treated as anything more than fantasy.

[Edit]: Actually that may be the best way to simplify my opinion on the subject. Kung fu movie = TAS --- Martial Arts/Boxing Tournament = non-TAS. Once again I feel the VG culture mirror/shadow sports.

[Mike Z]

Essentially, you can be psychic on a frame-by-frame level and always get the desired results from random variables.
--snip--
On the other side I completely and entirely disagree with TAS' being the same as what is possible through trial and time via normal means. There is no way that you could ever humanly produce the same results as a TAS of games like Tetris, Megaman, Mario, Final Fantasy, Metroid, Sonic, etc
--snip--
Anyone can pick up a tool and have the damn thing do it for them via trial and error on a frame-by-frame level.

- Fighting games for the most part do not have "random" random variables, or you couldn't record matches. It's always deterministic. If you wait the same number of frames for something, you get the same result - and usually it's not even dependent on that.
- Savestates allow you to look ahead and retry, but everyone here has spent hours trying different timings for a combo at some point. Savestates let you skip the initial setup to get to the part you're interested in trying.
- Tools do not do it for you - they allow you faster reaction than humans, and that's it. Humans still apply the thought.
- You aren't hacking the game - there is nothing that is not POSSIBLE by normal inputs, except that humans aren't that fast or precise. Playing back an emulator movie of a TAS (or a macro from a PPad) only gives a stream of inputs. If a human could time everything that quickly and precisely, they're equivalent. I do take issue with mostly-emulator-specific abuses, like pressing up+down at the same time.

I suppose for me it comes down to: Non-ppad/macro, or using ppad/macros to get difficult-but-humanly-possible parts right over and over, is interesting as a "look what's actually doable". Superhuman stuff is interesting as "look what you can do to this game."
The Kung-fu movie analogy is pretty good as long as you're talking Jackie Chan, not Crouching Tiger. Not a display of skill-to-be-useful, a display of skill-to-be-skillful.

Mike Z

[Maj]

Know what? The non-TAS is infinitely more impressive in my book, dispite it being full of flaws and slower than the TAS.

Point being, like I said before I can totally see how the two can compliment one another, but I feel they should be kept very very separate and treated as a novelty at best. In the genre of fighters it could lead to serious discoveries, but then you know about entertainment and realism...

Well, as long as everything is properly labeled, i think everyone will get their proper credit. I'm definately 100% in favor of mentioning any usage of tool-assistance in the video itself. Anyone who uses programmable controllers or emulator tools should mention it in their video in giant glowing letters. We're definately in agreement here and i think anyone who's ever made a video by hand would agree that it's important.

However, as long as people aren't lying about using tools, then i don't feel that those two scenes need to be kept separate. I mean, i don't think that people from one community should avoid contact with people from the other. Back when Sai-Rec was around, i never let myself get intimidated by the fact that they used programmable controllers. Sure, there are certain situations where they can get more hits out of a certain concept than i ever could, but i found ways around that. Either i optimized the combo in a less execution-driven direction or i simply replaced the unreasonably difficult parts with stylish alternatives. However, if i came up with a good idea, i never allowed execution limitations to hinder me from showcasing that idea.

That said, there are definately elements to doing combos by hand that get lost in the tool-assisted approach. I'm sure we've all come up with hella clever, helluva elaborate ways of setting up combos to reduce human error. Everything from using normal moves to set up spacing before a crossup attempt, to devising ways of making the combo more reliable, to developing command shortcuts and redundancies to loosen strict timing. A lot of that stuff gets lost when you take execution out of the picture. These are exclusive/unique problems that video makers had to solve when they performed combos by hand, and it's definately unfair to rob them of the credit they should receive for their solutions.

In other words, there are a lot of clever things that get overlooked when you solve a game frame by frame. Therefore, if someone releases a speedrun done by hand which is slower than a tool-assisted speedrun, it makes sense to evaluate them by separate criteria because they are solutions to different problems.

I think both are valid. I don't think that the best tool-assisted speedrun is worth more than the best manual speedrun, and vice versa.

Personally, i've gotten to a point where i'm sick of spending countless hours testing out one thing only to conclude that it "most likely" doesn't work. From here on out, i'd like to try the tool-assisted approach. However, i'm still going to keep the grading scales for manual gameplay separate from those for tool-assisted gameplay.

[Xenozip.]

- Fighting games for the most part do not have "random" random variables, or you couldn't record matches. It's always deterministic. If you wait the same number of frames for something, you get the same result - and usually it's not even dependent on that.

Specifically combos don't normally have random variables. There are games with random variables like critical hits in Melty Blood. I've never played SSBM but I hear Peach has a "random variable" on her turnips. But that's for combos. You're right though, I suppose random variables aren't that big of a deal.

- Savestates allow you to look ahead and retry, but everyone here has spent hours trying different timings for a combo at some point. Savestates let you skip the initial setup to get to the part you're interested in trying.

Again we thinking of combos. But what I was thinking of specifically was things like the Mortal Kombat TAS that very clearly displays the player being "psychic". The player simply played, saw what the CPU AI was going to do next, reloaded a save state and countered the AI's action.

If you were to take that to the extreme, imagine red parrying everything in 3S. With Tools it's possible. Could a human do it? Sure, probably after countless hours of trail and error until they got somewhere close to the desired results. But any monkey could pick up a tool and red parry every second hit and always perform Makoto's 100% stun every time they tried. Or how about CvS2 Geese always getting the correct counter whenever the opponent attacks and roll canceling his rekka a few frames before any attack connects.

Now for the sake of combos the undo function let's you experiment, that's for sure. That's not necessarily a bad thing. That's until you consider extremely long combos that have a lot of different attacks. XvSF and Melty Blood immediately come to mind. With save states and frame manipulation you could created 10 minute long combos with lots of different actions, not just looping the same thing over and over. And you could probably get it right "on the first try".

If you made a 10 minute long XvSF combo by hand that didn't repeat any loop more than once then you'd have my upmost respect for being that talented and for working so hard to make it. But using a tool, it's like you didn't even have to try.

- Tools do not do it for you - they allow you faster reaction than humans, and that's it. Humans still apply the thought.

Humans apply thought, yes. But that's really all that's being applied.

- You aren't hacking the game - there is nothing that is not POSSIBLE by normal inputs, except that humans aren't that fast or precise.
-snip-

Actually I brought up "hacking" and it was an improper use of the word. What I actually meant to say is that some one looked at the internal code for Metroid and discovered the game only checks for the Speed Boost flag once every 8th frame. So TAS' were then able to get Speed Boost by walking, and only running on every 8th frame. Which means you did not need a "runway" large enough to maintain full run in order to Speed Boost.

I don't disagree with you.

I suppose for me it comes down to: Non-ppad/macro, or using ppad/macros to get difficult-but-humanly-possible parts right over and over, is interesting as a "look what's actually doable". Superhuman stuff is interesting as "look what you can do to this game."
-snip-
Mike Z

Yes, I kind of agree with that. For me it's more like "look what's actually doable" and then "look what can be done with a tool".

With the kung fu analogy. If Jackie Chan and several other martial artists were to put on a live exhibition of their skills it would be completely different than what's seen in Jackie Chan's films. Just look at all the blooper reels in Chan's films: countless amounts of trail and error and mistakes that are made to look like amazingly perfect skill through movie magic. For me this is the difference between a match video and combo video.

But if Jackie had a Tool then things would be a lot different for him. He would not have to do the entire scene over and over just to get every movement perfectly correct, he could just undo at every point and get it right all in one take. He also would be frame manipulating so his reaction and speed would look amazing, but with the Tool he would just be seeing everything in slow motion. If Tools existed that let you manipulate reality then you probably wouldn't even need to practice martial arts ever, you'd just need the basic knowledge of martial arts so that you could emulate it. What would be the point of extensive training if you could parry every attack and hit every target dead center?

Same with fighting games. Anyone could pick up a tool and make it look like they knew what they were doing and get every combo perfect on the very first day they started using the Tool, but without the Tool then the player would actually have to work hard for the result.

Good thing/bad thing, I don't know. For all I know it could be a wonderfully awesome thing. But I personally perceive it as being something that could potentially be misused by irresponsible people. I also have a lot more respect for exhibitions when they aren't tool assisted because they display hard work and skill, not just knowledge.

Same result, significantly different process.

-snip-

Yes, I agree with your points of view on this.

But just to clarify, I agree that neither a TA or manual run is worth more than the other. But that's because I look at the two differently. They are both sources of different kinds of entertainment.

To throw another analogy out there, I don't think I'd like to watch basketball or football if it was played entirely by robots. Humans would be programming the robots, but you're still removing the skill, reaction, strength, speed, and need for training. In other words the games would be played by the coaches and no one else. If something like that existed I might watch it as a different source of entertainment, but human teams would be more entertaining and get more respect from me personally, due to all the hard work and effort they put in to be the best.

[Maj]

When a tool-assisted combo video is done right, it should take as much time as a combo video done entirely by hand. At the end of the day, what makes a project good is the amount of effort that went into it. Even creativity falls under the umbrella of effort.

When you use emulator tools or programmable controllers, you're not only eliminating execution obstacles - you're also raising the bar on expectations. If someone uses tools out of laziness, then that's just weak and it shows in their results. If someone takes the same "as good as it can possibly be" approach to tool-assisted projects, then that's commendable and it comes through in the final product.

The bottom line is that when i'm working on a project, my ideal litmus test isn't the question of "is this good enough to get away with putting into a video"? The real question is almost always "is this as good as i can possibly make it"? And when you bring emulator tools or programmable controllers into the mix, all they do is raise the bar on the "as i can possibly make it" condition.

However, there is one advantage that emulator tools and programmable controllers will always have over manual execution, regardless of the quality of the result. I'm talking about absolute transcript precision. Since you always need exact inputs for tool-assisted videos, you always end up with a frame-by-frame command sequence. If something inexplicably groundbreaking happens, you can show everyone the exact input string. This is always a more satisfying answer than "i had a hunch that it would work and after a hundred tries, it just did." And this degree of precision is very useful in collaborative projects, where one person submits an unpolished combo idea and another person decides to improve the setup, add a different finisher, make one subsitution in the middle, make it more stylish, etc.

[Xenozip.]

Something interesting my brother had to say: Manual runs have proven beyond a doubt that there is a vast range of player skill.

The best TAS runs have been created by collaboration, competition does not produce nearly the same results as a collaborations.

No one should watch any combo video with the expectation that it was done by a human. TAS when done properly stick out like a sore thumb, so they don't really need to be labeled. And honorable people will label TAS and TA-combos, but it should not be expected of people to do so. There's really no point in trying to trick people into believing that it was done without a tool.

-

Rather interesting coming from him. In regards to his "expectation" comment, I believe that's the same thought process as my feelings on the kung fu movie analogy.

I don't disagree with you guys. But I stand by my feelings that kung fu movies are fantasy and should be treated as a certain type of source for entertainment. But still a different kind of entertainment than tournaments and live exhibitions. And I feel the same about TA and manual. Another thing I can't help but point out is programmable sticks are banned from tournaments.

I suppose continuing with the anology it's like watching: Jackie Chan / Robo-Jackie Chan / Tournament or Live Exhibition

I agree with what you're saying Maj and I think they're a good thing. I respect them all and enjoy watching them all, but I personally don't feel they're all the same. Like I was saying with the football analogy, it's like watching two top-tier football coaches coach a team of robots. Entertaining sure, but a different kind of entertaining IMO.

[Maj]

I think we're pretty much on the same page. My only objection is that managing a team of robots is not trivial either, for two reasons. First, you have to learn a whole new set of skills to get the most of your robots and to keep them in shape. And second, suddenly it's not enough to win the game because even the slightest unforced error is enough to get you booed.

You mentioned using emulator tools to beat some shump game without losing a life, but that's not good enough. Once you get into the realm of tool-assisted speedruns, you automatically raise the bar very high and you can no longer get away with a 20% time improvement over a manual speedrun. Suddenly you have to cut your completion time in half, have to make sure no bullet misses an enemy, have to redo the same sequence a hundred times to manipulate the CPU character's AI into punching instead of kicking to save that extra 5 frames. Because if someone does release a stylish manual speedrun that's only 20% longer than your tool-assisted speedrun, then you get embarrassed.

Between the raised expectations and the hassle of dealing with imperfect homebrew/foreign technology, i don't think it's fair to undermine the effort that goes into making tool-assisted videos.

[Xenozip.]

Yeah, I don't disagree that effort was put into TAS'. I've said that my brother and I follow the Metroid runs regularly (actually he follows everything) and have been impressed with the constant advancements. And without Tools, things like the latest Manual run might never have come about (since the Manual run was done to mimic the TA run).

But again, if I were to watch VG videos as though they were movie productions, I personally feel TA and Manual are separate genres.

So, I'd like to get your opinion on this, please. I'll explain later.

[Maj]

Regarding that pair of crazy Storm combos from COTA, i don't know what to say. Did you make those? The last time i played COTA, it was over a decade ago - way before the internet taught me that there was more to fighting games than beating the boss on one quarter. Honestly, even when i watch that video frame by frame, i can't tell where a lot of the hits are coming from. If it's legit, then it's more advanced than any COTA combo i've seen before.

If you're asking whether i think it was done manually or with emulator tools, i couldn't offer anything more than a blind guess. Seriously, COTA is so foreign to me that i may as well be watching KI or DOA videos. I'd definately like to know what the hell is going on, though. Is there a transcript or explanation somewhere?

[Xenozip.]

Sorry for the delayed response. Here's details on the vid:

1) I made it, it's tool assisted, and I made it in less than a day. I made it on Saturday right after my post on Saturday.

2) I was able to create in incrementally, which is why it didn't take me very long. Every time I screwed up I only needed to return to a point right before the screw up, rather than having to do the whole combo again. If I wanted to make another ending for the combos (eg. end the combos with a super instead of air throw) then it would be very simple, I wouldn't have to redo the whole combo.

3) I know very little about COTA and never play it. I don't play any of the VS games either and know very little about them as well. What I've seen in videos and what I've read is the extent of my knowledge about the Marvel games.

4) "false" kara-canceling. In COTA, characters are able to cancel any of their normals into any of their valid cancel options (eg. magic chains, super jumps, specials/supers) prior to the active hit frames, which works a lot like kara-canceling. Storm is "special" because if you cancel her normals 1 frame before hit impact and the game will not force Storm into a hit-freeze, but it will register it for the opponent.

Basically, in most games when an attack connects both opponents freeze for a short time to allow for a dramatic impact effect. What Storm is doing is bypassing that freeze, but still forcing the opponent into that same freeze, which grants her several more frames than would normally be possible.

5) Other generic VS series stuff like fly/unfly, tri-jumps, and whiff canceling. The only other COTA-specific things are the "false" karas and Storms wind move. The wind move is able to push the opponent away or suck them toward her which is what allowed me to pull Psylocke out of the corner.

---

Yeah, picking COTA was a bit of a double edged sword. I chose it because I don't play it at all either which was supposed to help me make a point, but oh well. I guess I should have chosen another game.

The video kind of speaks for itself once you kind of see what's going on. I literally never play COTA, yet somehow I was able to magically create that. That's why I feel Tools do things for you.

There were a few other points I was trying to make, but I guess it's moot anyway. It's not like we disagree with each other, so really we're just debating on how much we agree with each other.

[Maj]

If someone shows me an impressive-looking COTA combo and asks me to judge the level of skill required to make it, my lack of experience with COTA obviously puts me at a disadvantage. However, i'm fortunate to be a part of a large community of very knowledgeable people that i can turn to for advice. My own personal area of expertise is only a small handful of games - the CvS series and maybe the OG SF2 series. Other than that i'm just a spectator.

But the community aspect is absolutely crucial. It's our best line of defense against someone trying to slip one past us. Cuz if you do something stylish in an MSH combo, then i'll give you props for that. As long as there are no glaring warning signs of dishonesty, i'm always glad to give positive feedback. But i'll definately keep an eye out for jchensor's, ToXY's and Mike Z's opinions. If any of those guys come along and say that the combo is unoriginal or suspect or humanly impossible, then their opinion is the one i'm going to adopt.

Thing is, not every tool-assisted combo is impressive. There are a lot of Japanese CvS2 videos out there that look crazy if you haven't kept up with the combo video scene for that game, but to me they're just copycat bullshit. My general policy is always to be supportive and always to offer encouragement, because that copycat's next video could be absolutely amazing. However, if anyone like that comes around SRK bragging about what is essentially a rehashed compilation, i'll be there to steal his thunder.

That COTA video clip was definately shocking and amazing to me. But if two other people from this forum decide to compete against it, suddenly you won't have such an easy time coming out on top, right?

Plus, do we really care about every game out there? I mean, if someone makes a video for a game that absolutely nobody plays and that nobody is going to challenge, does that require me to do extensive research into the game's combo history to determine the real value of the video? In most cases, i simply don't care enough about those obscure games. I'll watch it and i'll enjoy it, and i'll comment on the stuff that seems interesting to me. But how much is my compliment really worth? Any comment i make towards a Super Muscle Bomber video isn't worth nearly as much as any compliment i make towards a CvS2 video. Any good review is certainly worth something, but nobody will ever use my opinion of a Super Muscle Bomber video as the true test of its success. Without a solid understanding of the game, there's no way for me to rate the substance of a video. All i can do is reward style points, and even those are weighed on a sliding scale.

The general rule is: If i don't know the game's combo system inside and out, then i'm not the person you should be trying to impress.

So, i dunno, i guess i'm just not that threatened by tool-assisted videos. I don't think they'll stop people from making combo videos by hand. The problem is that the older a game gets, the more difficult it becomes to find new stuff within the realm of human execution. Damn near all of the games we play now are well over five years old. When CvS2 first came out, there were dozens and dozens of combo videos made by all sorts of people without programmable controllers. When Tekken 6 comes out, the same thing will happen again.

Plus, it's 100% impossible for tool-assistance to replace manual skill in the fighting game community because the heart of the scene is still competitive play. As long as Evo/ECC/MWC/SBO/etc. are around, people will always value execution. Street Fighter is a 2-player versus game at the most fundamental level. The entire scene will die out before that changes.

Regarding the tool-assisted speedruns linked in this thread, i'm only familiar with one of those fighting games. For the record, i didn't find those two SSF2 speedruns particularly impressive. If you want, i can list every single interesting thing in each of those videos and i'm sure i'll end up with less than five items per video.

Also, this is a topic that we're all going to be discussing a hundred different times in a hundred different ways over the next few years. Tool-assisted videos are steadily going to become more popular in the fighting game scene and i hope we make the right choices to avoid becoming segregated from the core competitive community. I've heard that most players in Japan don't pay attention to the combo video scene anymore. Even if i totally give up on making combo videos by hand, i'll still play at the arcade, still engage in strategy discussions on SRK, and still remain involved in the competitive scene. After all, us "combo video people" are good at analyzing and explaining technical matters to the rest of the community. None of these cool social networks/relationships need to come to an end anytime soon.

Edit: Just adding a couple of bold tags.

[Xenozip.]

If someone shows me an impressive-looking COTA combo and asks me to judge the level of skill required to make it, my lack of experience with COTA obviously puts me at a disadvantage. However, i'm fortunate to be a part of a large community of very knowledgeable people that i can turn to for advice. My own personal area of expertise is only a small handful of games - the CvS series and maybe the OG SF2 series. Other than that i'm just a spectator.

But the community aspect is absolutely crucial. It's our best line of defense against someone trying to slip one past us. Cuz if you do something stylish in an MSH combo, then i'll give you props for that. As long as there are no glaring warning signs of dishonesty, i'm always glad to give positive feedback. But i'll definately keep an eye out for jchensor's, ToXY's and Mike Z's opinions. If any of those guys come along and say that the combo is unoriginal or suspect or humanly impossible, then their opinion is the one i'm going to adopt.

It wasn't necessarily about judging the combo, it was about judging what can be done with a Tool.

Of course that combo is humanly impossible, IMO. No human is going to be able to input every single attack in that combo on the exact frame that is required to false-kara the last attack.

But that's what I was talking about before, Tools allow you to do stuff you'd never ever see a manual run doing. This is what Mike Z said:

I suppose for me it comes down to: Non-ppad/macro, or using ppad/macros to get difficult-but-humanly-possible parts right over and over, is interesting as a "look what's actually doable". Superhuman stuff is interesting as "look what you can do to this game."
-snip-
- You aren't hacking the game - there is nothing that is not POSSIBLE by normal inputs, except that humans aren't that fast or precise. Playing back an emulator movie of a TAS (or a macro from a PPad) only gives a stream of inputs. If a human could time everything that quickly and precisely, they're equivalent. I do take issue with mostly-emulator-specific abuses, like pressing up+down at the same time.

My nitpick is that no human could. There's just no way that combo is humanly without frame advancement.

That COTA video clip was definately shocking and amazing to me. But if two other people from this forum decide to compete against it, suddenly you won't have such an easy time coming out on top, right?

With the ability to combo anything into anything else, all it becomes is a freestyle, IMO. How does one guage which combo is "better"?

By how fast it kills the opponent means whatever Storms 2 strongest attacks are looped, that could be finished in less than an hour. By what you're able to combo is just a matter of testing each variable. Anyone could create a clip displayed Storm comboing all of her attacks into each of her other attacks given enough time to go through each variation. So all that's left is freestyle, AFAIK.

I'm not saying there isn't a higher level to be achieved there, history tells us that there's always a higher level in everything. But the process is entirely different from manual, and so is the result. If competitors in the Olympic games had frame advance and undo points then things would get pretty linear and silly pretty fast, everyone would be perfect.

My other nitpick is that it didn't take me any time or effort. I could spend hours and hours learning CvS2, designing combos, putting forth a ton of effort to get execution down, and coming up with a script and running through it. Only for all that to be thrown in my face by some one who picked up a Tool and did the same thing in less than an hour having never played the game before, only theirs had tighter/perfect execution. I don't know why that bugs me. I suppose it shouldn't, but for some reason it just erks me.

Also, this is a topic that we're all going to be discussing a hundred different times in a hundred different ways over the next few years. Tool-assisted videos are steadily going to become more popular in the fighting game scene and i hope we make the right choices to avoid becoming segregated from the core competitive community. I've heard that most players in Japan don't pay attention to the combo video scene anymore. Even if i totally give up on making combo videos by hand, i'll still play at the arcade, still engage in strategy discussions on SRK, and still remain involved in the competitive scene. After all, us "combo video people" are good at analyzing and explaining technical matters to the rest of the community. None of these cool social networks/relationships need to come to an end anytime soon.

That's true. I apologize for marring this thread with redundant banter. If it's possible to split the thread then that might be a good idea?

[desk]

2) I was able to create in incrementally, which is why it didn't take me very long. Every time I screwed up I only needed to return to a point right before the screw up, rather than having to do the whole combo again. If I wanted to make another ending for the combos (eg. end the combos with a super instead of air throw) then it would be very simple, I wouldn't have to redo the whole combo.

Do you mind saying how you did that? what programs you used and stuff? I would save so much time if I was able to do that.

I had a huge problem when I first found out about tool assisted stuff. It was after I had watched the sairec CvS2 preview with the silly guile combo's. I though it was insanely good and whoever 'performed' it must be awesome. Then I was taken by surprise when people started talking about it being programed. I felt like I had been tricked even though it clearly says, 'combo programmer' in the video. I'd never come across it before so I didn't know anyone did it. Now however, I have absolutly no problem with it and I'm hoping to explore the area myself with my next vid

[Xenozip.]

http://tasvideos.org/EmulatorHomepages.html

That site has what you need and info on how to do it. Look for my brother Blip's tools.

[Maj]

But that's what I was talking about before, Tools allow you to do stuff you'd never ever see a manual run doing.

My nitpick is that no human could. There's just no way that combo is humanly without frame advancement.

Yeah, but there's nothing inherently wrong about that, right? It's just the next level of combos. The complaint isn't that tool-assisted combos are humanly impossible to execute. The complaint is that tool-assisted combos are easy. But when i hear about ozy working on the same XSF combo for 3 weeks even with a programmable controller, that not what i would call "easy." Granted, emulator tools are even more flexible and even more forgiving than programmable controllers, but there's no way they would have cut down a 3-week programmable controller combo into a matter of minutes. Maybe it would have taken one week. That's still a lot. And even then, better tools would only have given him more difficult ideas to explore. Good combo video makers never shy away from that, so that'd put him right back at 3 weeks.

With the ability to combo anything into anything else, all it becomes is a freestyle, IMO. How does one guage which combo is "better"?

Yeah, but you gotta admit that COTA is kind of the extreme abnormal case here. It's almost in a category all by itself. The game is absolutely broken beyond belief. Improving a Sai-Rec CvS2 combo or a Meikyo MvC2 combo is nowhere near that easy. Hell, even Killer Instinct combos have maximum limits that are very rarely broken these days. COTA is like a constant CvS2 Custom Combo with no meter draining, not even when you use supers, and not even any juggle restrictions on throws. Combos in that game were all about freestyle way before emulator tools showed up.

That's true. I apologize for marring this thread with redundant banter. If it's possible to split the thread then that might be a good idea?

Oh, dude, this discussion has been awesome. Nothing to apologize about. I'm just saying that we're going to be seeing a lot of these debates. It comes up every time Magnetro announces a new video, it's going to come up when desk releases his next video, and it's definately going to come up when i finish the video i'm working on. The trick is to handle every instance calmly, so we don't accidentally end up alienating a large portion of the scene by acting elitist or isolationist or defensive.

[Xenozip.]

Yeah, but there's nothing inherently wrong about that, right? It's just the next level of combos. The complaint isn't that tool-assisted combos are humanly impossible to execute. The complaint is that tool-assisted combos are easy. But when i hear about ozy working on the same XSF combo for 3 weeks even with a programmable controller, that not what i would call "easy." Granted, emulator tools are even more flexible and even more forgiving than programmable controllers, but there's no way they would have cut down a 3-week programmable controller combo into a matter of minutes. Maybe it would have taken one week. That's still a lot. And even then, better tools would only have given him more difficult ideas to explore. Good combo video makers never shy away from that, so that'd put him right back at 3 weeks.

Nothing wrong with it when it's done for that specific purpose and it's appreciated for what it is. If TA-combos and manual-CV were treated exactly the way TAS' and manual runs were, then I don't see any harm in it and there's really nothing wrong with it. But this is why I have such a great appreciation for people like Radix.

But not everyone feels that way. A lot of people want to "see what players can do" and are only vaguely interested in seeing "what people can do when cheating". Though keep in mind that's not my personal views on it -- just what I've heard from people who are apposed to the concept.

My personal reasoning for appreciating manual runs more is: watching manual runs you're seeing the application of skill, thought, and effort. With TA' you're only seeing the application of thought. I personally like the idea of using TA' as a template for manual runs -- to show what is possible and what to strive for.

PPad combos are entirely different IMO. They are done in real-time and if you screw up you have to try again. With the latest batch of Tools my brother created everything is done frame-by-frame so you have complete control over what inputs are done at each frame, and there is no "screw up and retry" with the ability to undo. If you screw up due to an unforeseen error on combo hit #33 then you can just load to combo hit #32 without doing the whole combo over, and either try again slightly differently or try an alternative. That's why those Storm combos took me less than a day to create having little to no actual experience with the game, I just tried different variables until I got the desired results. Although, it was incredibly boring doing everything frame-by-frame.

If ozy had used my brothers FBA hack it would not have taken nearly as long and he may have even stumbled upon potentialities he's never seen before.

But the thing about potentialities when analyzing the game at a frame-by-frame level is I've come to understand it's a double edged sword. You really need to look at the game from all angles if you really want to get into the full depth of the game.

Yeah, but you gotta admit that COTA is kind of the extreme abnormal case here. It's almost in a category all by itself. The game is absolutely broken beyond belief. Improving a Sai-Rec CvS2 combo or a Meikyo MvC2 combo is nowhere near that easy. Hell, even Killer Instinct combos have maximum limits that are very rarely broken these days. COTA is like a constant CvS2 Custom Combo with no meter draining, not even when you use supers, and not even any juggle restrictions on throws. Combos in that game were all about freestyle way before emulator tools showed up.

Right, it's not even a question of how to do it or what's possible. It's only about picking what you want and putting them in the order you want. The difference is you don't even have to try finding that stuff anymore, nor do you need to actually put any work in to do it. Being that COTA is totally and completely broken there really isn't limits, only freestyle variants and/or conditions.

With games like CvS2 there are hard-coded limits though. Juggle limits prevent most characters from doing anything more than what their juggle count allows, so therefor it's a matter of analyzing the character's damage output and finding the optimal route. If the stipulation is actually combo hit-count rather than damage output then things get a little more interesting, but it's still a matter of process of elimination: you go through every possible permutation until you find the limit.

But that's really the interesting thing right there: "possible". Which is why TA' stuff still holds some value. People get to analyze and think outside the box and discover every single possible situation ever. Likewise, if the stipulation is links and combo variants then it becomes even more interesting and potentially wacky (much love for Ode to the 2hit).

But again, this is all the application of thought. Manual exhibitions are thought, skill/execution, and effort via trial and error.

You also get to see what's humanly possible -- as in "hey, look what I can do!". As apposed to what's technically possible -- as in "hey, look what can be done!".

Problem with the "look what can be done" thing is it's not necessarily humanly possible, but really anyone can do the same thing using a Tool.

I like being a fan of sports players and video game players who have specific styles and are very talented and experienced. It's fun rooting for your favorite team/player and being entertained and awed by their display of skill. It's nice to have a role model/hero. But, I don't think I'd be much of a fan of Robot#2501, but I can appreciate the thought that went into it's programming.

Oh, dude, this discussion has been awesome. Nothing to apologize about. I'm just saying that we're going to be seeing a lot of these debates. It comes up every time Magnetro announces a new video, it's going to come up when desk releases his next video, and it's definately going to come up when i finish the video i'm working on. The trick is to handle every instance calmly, so we don't accidentally end up alienating a large portion of the scene by acting elitist or isolationist or defensive.

Problem that I see is that people have generally already made up their minds. The way we handle things counts for something, but it's probably not going to make the majority of people suddenly change their views on the subject.

The fighting game community is sort of an odd beast. It's not at all like the other communities I've been a part of.

[Magnetro]

/ . Yeah, people usually have their minds made up, maybe they'll just understand how it should be appreciated but they'll still think that a video done by hand is cooler =/. Or in my case, a guy's video made by hand is faster than a PROGRAM PAD.....*sigh..marvel...*

All that's left is to get people to understand what to expect from a ppad video, that's pretty much it. [/b]

[Xenozip.]

Regarding ozy's combo stuff. If I were to take that Storm combo, replay the input recording, save a state right before that last throw and "undo", I could instead finish the combo with a super instead of a throw without having to redo the entire combo. As seen here. Courtesy of Blip (who also rerecorded the last portion of the combo for me).

That's why it takes infinitely less time with a Tool than it does with a PPad.