I was originally going to try pikmin, but could not for the life of me get it to sync consistently, so instead i thought this would be a good game to tas. And it turns out, it syncs fairly well. My first run through act one had one desync at frame 15000, and after replaying it, the next 6000 frames synced just fine. Not perfect, but not bad either.
Currently working on a quick play through on shura to generate a save file, and have learned a few things in the process.
Most notable, what i thought were forced battles (i fought 4 encounters in the move in the first post), can in fact be avoided. These fights are changed, and can be avoided entirely by things like changing screens (some times this requires moving back and forth between two screens upwards of 5 times, which is still faster than fighting), or picking up items.
Once i finish on shura, i'll gather frame data, and such, and start working on a serious run.
Frame data on jumping/falling:
falling and landing twice: 3730
jumping once: 3724
Falling once then jumping from the branch: 3730
falling then jumping midair: 3724
double jumping: 3724
falling twice, hold down on second fall: 3732
jump, then extra jump for no reason: 3729
jump, then double jump from the ground: 3729
jump, double jump, hold down after second jump peaks: 3739
testing was done on a screen that started up high, has a small drop onto a branch right away, then quickly drops again, much further, to the ground. The number is the first frame that the screen is completely black to load the next area.
Based on this data, it seems that jumping does not cost any frames, but landing costs 5. Holding down while falling is slower, with the exact amount of frames lost obviously depending on the height falling. I'm not sure where the extra frame is coming from when falling from the ledge to the branch. My first thought was that falling cost 1 frame, but it can't since falling then jumping from midair still ended at frame 3724. I doubt it will ever come up in an actual run though, so whatever.
others:
changing weapons/items outside battle costs no frames
picking up items also costs no frames.
pressing down for 1 frame (used for luck manipulation to avoid random encounters) costs 1 frame.
pressing back for 1 frame costs no frames (? triple checked this, it's right, but i have no idea why).
releasing the joystick for 1 or 2 frame also costs no frames, but after 3 frames, it costs 1 frame.
Interesting results at the end. I was not expecting to be able avoid encounters without wasting any frames. It's especially strange since she does indeed turn around, or start to stop, but apparently keeps moving forward at the same speed regardless. Anyway, I'll work on attack patterns and damage another time.
current route:
act 1: hold right, kill boss.
act 2: small one screen detour to buy 2 bronze mirrors, and smoke bombs, then hold right, kill boss.
act 3: use mirror to warp to starting screen. In suruga, take a 10 screen (5 screens, then turn around)) detour to activate shrine. This saves 3 screens over using the previous shrine i pass in Izu.
act 4: use mirror, buy more mirrors and smoke bombs in suruga from merchant.
There is a bronze mirror in shinano that can be had for 94-99 frames. This would save 30 mon buying one. Not sure if that's worth it, will need to see exactly how much money i have when i reach the merchant in suruga (as well as exactly how many mirrors i need at that point). If it allows me to buy an extra smoke bomb, it will almost certainly be worth taking 99 frames to get to it.
Will plan out route further as i continue playing through.
Just finished act 5 boss on my shura play through. Everything is still syncing pretty well, and with read only fixed now, it's fairly painless. I should hopefully be done with this in a few days, and then i can get started on my actual tas through shigirui. Which, i honestly don't expect to take more than a week or two.
Honestly though, as i replay through it, i'm starting to think this is a fairly terrible game to tas. There aren't any glitches that i'm aware of, and with little to no random encounters, the game is like 80% travel (50 mins in so far, boss times were 1:30, 2:00 2:00, 3:00, 3:00 or something like that, and they could be shortened by probably 20% at least with tasing). Still, i've already done so much work prepping the run, i'd hate to not do it, especially since it really would not be too difficult to do. And of course since i've never tasd before, it never hurts to start with something easy.
Joined: 4/21/2004
Posts: 3517
Location: Stockholm, Sweden
Trust me, Baxter and I aren't certainly the only ones who look forward to this. Tasing isn't all about zipping and whatnot. It's also about "superhuman" reflexes aswell :D And besides, with all due respect, it would definitely be nice to see something else besides Mega Man 10 and Melee submissions :)
Nitrogenesis wrote:
Guys I come from the DidyKnogRacist communite, and you are all wrong, tihs is the run of the mileniun and everyone who says otherwise dosnt know any bater! I found this run vary ease to masturbate too!!!! Don't fuck with me, I know this game so that mean I'm always right!StupedfackincommunityTASVideoz!!!!!!
Arc wrote:
I enjoyed this movie in which hands firmly gripping a shaft lead to balls deep in multiple holes.
natt wrote:
I don't want to get involved in this discussion, but as a point of fact C# is literally the first goddamn thing on that fucking page you linked did you even fucking read it
Cooljay wrote:
Mayor Haggar and Cody are such nice people for the community. Metro City's hospitals reached an all time new record of incoming patients due to their great efforts :P
Agreed. I will love to see this TAS, because Muramasa is already a very artistic game, and combining that with superhuman combat that would make wuxia films jealous seems like a winning combination.
(Besides, a game being 'lots of running in between particular fights' pretty much defines the RPG games we have here, and many of them have no shortage of fans.)
First a movie gets submitted, and ends up accepted despite breaking rules other runs have been rejected for. And when I vote less than spectacularly on this movie, I become the victim of harassment and threats.
Yay, favoritism.
Oh god, some of the enemies in hell don't scale with level. I'm currently level 9, in an area where you would normally be level 25+ (with weapons that do 5x more damage). One enemy just took over a minute to kill. Guess i better make sure i have smoke bombs to avoid those.
And i'm glad to hear people are actually interested in a run of this game.
I've never played this game, but I would definitely like to see another Wii(?) game in the submission list, especially since I know it takes hard work to make a TAS in Dolphin and the result should be an enjoyable run. The game sounds like it would make for an entertaining run, too.
Just finished through shura with a time of 1:43. Proper routing should knock off 10-15 minutes, and i'd expect better optimization to take off a additional 15 minutes or so. 1:15 seems like a safe bet. sub-1 hour may be possible, but i doubt it.
Act 1 is really short and easy, so it should be done tomorrow or the day after, depending on how much time i have between family activities. I'll upload it to youtube as i finish each act.
Took longer than expected because when i played back the end of my movie from the start, it ended up desyncing 50k frames before the end, so had to redo most of the final act. dtms for my shura run can be found here. I'll probably have act 1 finished and encoded by morning.
Damage:
All done against first boss, Quick draw was with a 13 damage weapon, normal attacks with a 12 damage weapon. Moonlight damage may be wrong, since it hits like 7 times, and numbers overlap, and is sometimes impossible to see, etc which is why i only tried it 3 times, but it shouldn't really change the overall conclusion. Arts can be used during the animation for a QD, which is why they are separated.
Switching over 2 weapons requires 10 more frames than switching to the next on, which will always happen every other time i switch weapons during a boss, since i'll never have more than 2 useful weapons at once.
Conclusions are pretty obvious for the 3 starter weapons: only use arts right after QD, which should always be used as soon as it is available, otherwise just spam normal attacks.
I have not tested the odachi yet, but i'm fairly sure it's completely useless against single enemies, i'll update if i find otherwise.
Ariel attacks don't require testing, as it's incredibly obvious that it is always best to combo attacks, but never finish a combo in the air except right before hitting the ground (since the final attack brings you to the ground, and all but the last hit of it do extremely low damage). The final attack can be interrupted with arts, or that moving attack thingy.
Manipulating damage is possible, but since the only method i have of doing so is via brute force, and it wouldn't save more than a few seconds per boss, i'm not sure how much i really care to do so. Also, in 30 combos, i never once saw a 7-7-7-11 damage round, so fuck trying to manipulate that repeatedly.
edit: downward swipes take 60 frames, and average 16.7 damage each (15-18 range), so with about the same damage, but 5 frames faster, they are better than 3 hit combos if i'm unable to finish it for whatever reason, or especially to use them to finish a boss off, since they do the entire damage after ~25 frames, which i believe makes it the fastest finisher i have.
edit2: 14 damage odachi gets ~47 damage per 200 frames, so it is indeed worthless.
edit3: I'm really enjoying TASing, but it kills me how much effort goes in for such tiny improvements. Just finished the first boss, and got a time of 1:30, fighting it nearly perfectly. I doubt it could possibly be improved by more than a second. Meanwhile, my test run which looking back on it, was fairly terrible killed it in 1:34, and using almost no save states i beat it on shura in 1:35. I just spent nearly two hours to save 4 seconds on a fight that requires little more than mashing A. And now i get to redo the last 30 seconds that desynced too.
Oh well, it's fun anyway. Encode of act 1 should be uploaded within an hour or so.
more edits: and now i have to redo the entire first boss. Following up a 4 hit combo with an uppercut, flying dash attack, and then A spam until i hit the ground is about 33% faster than just mashing A, because you can start the uppercut as soon as the fourth swing hits, without waiting at all. This takes ~220-230 frames per cycle, and does ~95 damage.
even more edits: uploading in glorious 720p to http://youtu.be/YufGFanxBfs now. I'll upload to nico and megaupload after, and edit yet again when it's done.
Link to videohttp://www.megaupload.com/?d=H6HN1GN7
Joined: 4/21/2004
Posts: 3517
Location: Stockholm, Sweden
rog, you did great! This is your first tas and it's obviously not easy to strive for perfection. Keep hunting all frames you can get, even if it means redoing for instance 1k and whatnot (you will feel good about it afterwards, even if the award is little). That boss is one tough son of a gun. Boy can it take a hit. Is this due to you are playing on the highest difficulty? Regardless, I am very keen to see your next wip and congratulations in the making of a Wii game. History is once again being written in the hands of rog :P
I've also never seen a game where you can do more than 300 combo damage :D
Edit: I just noticed that the act 1 encode is over 300mb. I wonder how big the final encode will be. I also wonder if nitsuja or grunt could somehow optimize the encoding method of dolphin even further so that the savestates, movifiles and encodes aren't this large. On the other hand, this is afterall a Wii game we are talking about so we must cease thinking this is fceu or some sort :)
Nitrogenesis wrote:
Guys I come from the DidyKnogRacist communite, and you are all wrong, tihs is the run of the mileniun and everyone who says otherwise dosnt know any bater! I found this run vary ease to masturbate too!!!! Don't fuck with me, I know this game so that mean I'm always right!StupedfackincommunityTASVideoz!!!!!!
Arc wrote:
I enjoyed this movie in which hands firmly gripping a shaft lead to balls deep in multiple holes.
natt wrote:
I don't want to get involved in this discussion, but as a point of fact C# is literally the first goddamn thing on that fucking page you linked did you even fucking read it
Cooljay wrote:
Mayor Haggar and Cody are such nice people for the community. Metro City's hospitals reached an all time new record of incoming patients due to their great efforts :P
Bosses all have a lot of hp, i don't think the difficulty changes that. It's definitely the same on shura anyway, i've never actually played on muso. I should easily get a 700+ combo on one of the later bosses.
Dolphin actually spit out a 15 gb file. I probably should have compressed it more/better, but it's not something i've really done before, so i wasn't too sure how to do so without destroying the quality.
Also, the .dtm: http://www.mediafire.com/?hg8vpi7uj4w2mvg
Great job so far! The WIP is fast-paced and looks optimized. I agree with AngerFist that it's worth it to redo unoptimized parts and then continue.
Thanks for taking the time to encode it. Good luck with the rest of the run!
Ugh, wtf is this. Refought boss one, and after 4 hp bars, i was ahead of my previous time by 9.5 seconds.
Then he got an extra hp bar, which cost me 17 seconds.
It wasn't based on anything before the fight, as i started rerecording 5 or 10 seconds into the fight. Hopefully it's entirely random, so i can manipulate it, and not based on how fast i damage him.
Could be the game has some form of dynamic difficulty? If it sees that you're really good then it makes the game harder.
Or it could be random. But the easiest way to test would probably be to make the fight slower by increments and see if there's some cutoff where it consistently doesn't add the extra health bar.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Could be the game has some form of dynamic difficulty? If it sees that you're really good then it makes the game harder.
Or it could be random. But the easiest way to test would probably be to make the fight slower by increments and see if there's some cutoff where it consistently doesn't add the extra health bar.
Yeah, i have no idea. I didn't even notice it until i finished the boss, and saw my time. About to start testing it now.
edit: It's not random :(
If i finish off the first hp bar before frame ~12650 it takes half as much off the second hp bar vs if i finish it after that frame. I'm going to need to stand around for 2 seconds there -.-""
Best possible time for the first boss should be something like 1:14 +/- 2 seconds i think.
edit2: just managed 1:19 for the boss. With massive amounts of brute forced luck manipulation, i could likely get a few more seconds off, but i don't really see more than a second of improvement happening any other way. Frankly though, i'm not really willing to do it for a 2-4 second improvement on a 1:15+ run, so i think i'm just going to stick with what i've got now. I'll post a new encode later.
edit3: Apparently the battle timer in game is bullshit. My latest attempt says 1:19 (first was 1:21), and is 6 seconds faster. Guess that isn't at all useful for comparison.
When in doubt, use the frame counter. It won't lie to you. I'm not at all familiar with the game* but it's entirely possible that the in-game timer is inaccurate for all sorts of reasons. Lag is the obvious possible explanation, but it may also purposefully run the timer inaccurately during special techniques or the like.
Kinda lame that the game makes bosses take longer if you're really good at killing them. I wonder if this might be a situation in which the most entertaining option isn't the fastest? That is to say, going all-out and inflicting as much damage as possible even if that causes the boss to grow extra limbs and organs for you to lop off, instead of sitting around wasting time so he won't. Again, not familiar with the game, and this is a judgement call at the best of times. Go with whatever you think is best.
* Though, having seen it for $10 on Amazon, I went ahead and ordered it after seeing your encode; it has a good pedigree.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
When in doubt, use the frame counter. It won't lie to you. I'm not at all familiar with the game* but it's entirely possible that the in-game timer is inaccurate for all sorts of reasons. Lag is the obvious possible explanation, but it may also purposefully run the timer inaccurately during special techniques or the like.
Kinda lame that the game makes bosses take longer if you're really good at killing them. I wonder if this might be a situation in which the most entertaining option isn't the fastest? That is to say, going all-out and inflicting as much damage as possible even if that causes the boss to grow extra limbs and organs for you to lop off, instead of sitting around wasting time so he won't. Again, not familiar with the game, and this is a judgement call at the best of times. Go with whatever you think is best.
* Though, having seen it for $10 on Amazon, I went ahead and ordered it after seeing your encode; it has a good pedigree.
Yeah, i will be now. I just assumed it was accurate and was using it for a basic comparison. I can't really imagine why it was off by so much. I didn't use any special attacks, except at the beginning, which was exactly the same in both attempts, and lag isn't really much of an issue in this game except in certain places. But whatever, it doesn't really matter.
New encode
Link to video
Changes from the last video start at 3:20. At 3:25 you can see where i had to wait. I lose ~3 seconds there and another 3 seconds at 4:18, and it looks silly, but in the grand scheme of things, it's still only 6 seconds, which despite looking like poor gameplay, is still better than spending ~17 seconds dealing damage that doesn't need to be dealt.
One thing I noticed in the downloadable encode is at about 0:35. Is it possible to jump onto the left branch (the one with the sparkle), and use that to jump to the upper right branch and then up, rather then hitting the wall? This is where I mean.
[URL=http://imageshack.us/photo/my-images/850/muramusasnap.png/][/URL]
One thing I noticed in the downloadable encode is at about 0:35. Is it possible to jump onto the left branch (the one with the sparkle), and use that to jump to the upper right branch and then up, rather then hitting the wall? This is where I mean.
Yes, but i need to waste a number of frames anyway to manipulate the rng for the next screen (if i enter the next screen as fast as possible, i just end up with an unavoidable battle). Also, i'd now have to redo the entire act to change it.
All done against first boss, Quick draw was with a 13 damage weapon, normal attacks with a 12 damage weapon. Moonlight damage may be wrong, since it hits like 7 times, and numbers overlap, and is sometimes impossible to see, etc which is why i only tried it 3 times, but it shouldn't really change the overall conclusion. Arts can be used during the animation for a QD, which is why they are separated.
Switching over 2 weapons requires 10 more frames than switching to the next on, which will always happen every other time i switch weapons during a boss, since i'll never have more than 2 useful weapons at once.
Conclusions are pretty obvious for the 3 starter weapons: only use arts right after QD, which should always be used as soon as it is available, otherwise just spam normal attacks.
I have not tested the odachi yet, but i'm fairly sure it's completely useless against single enemies, i'll update if i find otherwise.
Ariel attacks don't require testing, as it's incredibly obvious that it is always best to combo attacks, but never finish a combo in the air except right before hitting the ground (since the final attack brings you to the ground, and all but the last hit of it do extremely low damage). The final attack can be interrupted with arts, or that moving attack thingy.
Manipulating damage is possible, but since the only method i have of doing so is via brute force, and it wouldn't save more than a few seconds per boss, i'm not sure how much i really care to do so. Also, in 30 combos, i never once saw a 7-7-7-11 damage round, so fuck trying to manipulate that repeatedly.
edit: downward swipes take 60 frames, and average 16.7 damage each (15-18 range), so with about the same damage, but 5 frames faster, they are better than 3 hit combos if i'm unable to finish it for whatever reason, or especially to use them to finish a boss off, since they do the entire damage after ~25 frames, which i believe makes it the fastest finisher i have.
edit2: 14 damage odachi gets ~47 damage per 200 frames, so it is indeed worthless.
edit3: I'm really enjoying TASing, but it kills me how much effort goes in for such tiny improvements. Just finished the first boss, and got a time of 1:30, fighting it nearly perfectly. I doubt it could possibly be improved by more than a second. Meanwhile, my test run which looking back on it, was fairly terrible killed it in 1:34, and using almost no save states i beat it on shura in 1:35. I just spent nearly two hours to save 4 seconds on a fight that requires little more than mashing A. And now i get to redo the last 30 seconds that desynced too.
Oh well, it's fun anyway. Encode of act 1 should be uploaded within an hour or so.
more edits: and now i have to redo the entire first boss. Following up a 4 hit combo with an uppercut, flying dash attack, and then A spam until i hit the ground is about 33% faster than just mashing A, because you can start the uppercut as soon as the fourth swing hits, without waiting at all. This takes ~220-230 frames per cycle, and does ~95 damage.
even more edits: uploading in glorious 720p to http://youtu.be/YufGFanxBfs now. I'll upload to nico and megaupload after, and edit yet again when it's done.
Link to video
http://www.megaupload.com/?d=H6HN1GN7
[/URL]
