Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Great improvement. I knew the old movie was improvable, but thought it could be a few seconds at best. Shows how much I know. Hopefully sub-7:00 is eventually possible..
The last level is really impressive! I could barely handle those clunky columns. Ryu actually looks smooth in this movie.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
I thought that was you behind that comment! Unfortunately for this game, it is fairly dull to TAS, so I never tried making a second run... Cool to see it console-verified anyway.
MESHUGGAH wrote:
edit2: a question for the author: are you aware that your youtube channel name doesn't reflecst your username change here at tasvideos?
edit3: and the submission file as well
Yes, I have no problems with that name. It's still my real name (outside of the last initial), so people can use it if they want.The submission file, nothing I can do about that unless I learn to travel through time. :-p
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Just for the record,there's no rule against necroposting on this site. In fact, having only one thread is preferred.
MODERATOR EDIT: That is indeed correct, so I've merged the two threads and deleted the resulting duplicate post. — moozooh
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
What a pleasant surprise to see another TAS of this by you. I didn't expect there to still be any significant bugs hiding in this game, and the new route in 5-3 was very clever. Also, I'm glad you fixed my dreadfully out of sync dancing:p
Easy yes vote!
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Many years ago, I tried to improve the published movie. I made it about a quarter in, but I found the game extremely dull to TAS, so I dropped it. I thought of picking it back up recently, but I had other projects that were more interesting, so I'm glad to see someone else finally knock it out.
Anyway, I just checked right now, and my WIP has two rounds faster than this submission:
- In level 10, it is a few frames faster to go down the ladder before walking off the edge after pushing the rock down.
- In level 17, after you kill the first enemy, it's a bit faster to go down the ladder before walking off the ledge... so basically the same idea as the above.
There may be some other improvements like that, but I haven't checked the entire run. Regardless, congrats on saving close to a minute on one of the oldest submissions on the site. Going to do Boomer's world?:)
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
You can't do that trick if you don't have a slope or start from roomtransition. Otherwise you could be flying at top speed whenever you jump.Also, grapping an item doesn't reset your speed to 0, so it wouldn't be the proper conditions for the trick anyway.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Two straight submissions of mine with an improvement found after? I must be slipping ^.^ Here's that implemented: http://tasvideos.org/userfiles/info/65633912286313509
I also took the liberty of making a high-def encode, since I noticed my original was pretty blurry at times.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Dying at the goal is an improvement of 72 frames, which means we have 3.5 seconds saved from the default three extra lives given to you by the game, while the rest are (72 - however many frames you needed to waste to collect a 1up, usually 20-30). The closest call comes from the detour to pick up the life in the very first stage, which only winds up saving a handful of frames, even though it's visually very different. The suicide after grabbing the magic bag only saves 2 frames, but luckily it doesn't require having an extra life. It's also worth noting that the position of the character at the goal isn't relevant to time, despite what came up in the last submission thread.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
I did see that movie, but it doesn't seem like the same situation since it was obsoleted for containing the exact same gameplay as an all levels run. M.C. Kids all levels requires collecting all 6 cards in each level, in addition to every secret card, in order to access the secret 7th world, so it's closer to a 100% run than a run that just plays the any% run input with the skipped levels.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
I think pressing B immediately is better. You can access the credits from the menu without playing the game at all, so it's not like it's inherently an "ending" that must be seen by the viewer.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Don't feel offended. I agree that this game is crappy... It would be accurate to say I made this just to get rid of an unoptimized published movie. So it's not the most passionate I've ever been about a TAS. I think that this run would have a chance at Moons if the music wasn't so awful, since the gameplay isn't the worst thing ever. Though it gets boring how every level is about hitting switches and running to the exit.
3 frames improved on a jump? I haven't checked what you did differently right now, but I'm not surprised another pair of eyes could find an improvement, this game is pretty "fluid" to TAS in that almost every non-trivial moment can vary wildly in frames. Luckily, even though it's dull to TAS, it's almost completely edit friendly (large chunks of this run were taken from my first version). So, if you did something that I didn't think of, I could implement it throughout the run without much trouble. I'd prefer to avoid another Wolverine scenario...
EDIT: Okay, I did another pass and improved the run. It's 149 frames faster now. I fixed some sloppy jumps left over from my testrun, which was the bulk of frames lost, whoops<_<. There were improvements in most levels, but I couldn't improve some because of the enemies, which aren't manipulatable, so they're kind of a frame rule. Anyway, here's the file: http://tasvideos.org/userfiles/info/65254474712194872
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
arkiandruski wrote:
Quick question: at 41 seconds into the provided encode, I noticed that both you and the published run turn around to spit at the whatever-it-is. What's preventing you from being able to hit it from the left?
The jump to that switch needs a dash to clear it, and you can't dash and spit at the same time. So, trying to spit midair there would result in falling to your death.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Boring TAS update: Unfortunately, I was nearly finished with this run when I discovered by a pure chance that Milon runs 1 frame faster if you tap right the frame he first appears. This is too significant of a timesave to ignore, given that every frame in this game is important to randomness, so I had to start over for the third time. (I recorded the first 2000 frames again, which you can see here, but it's nothing too different from the published run so far, so it's only really notable as a historical curiosity.)
Even though I'd still like to finish this movie, I'm going to place this game on low priority for now, since I'm bored of TASing it. It requires quite a lot of brainpower with all the randomness, lag, and subpixel optimization, so it's not an easy game to TAS for months on end. If someone else feels interested in giving this game a shot,feel free:) Here are the RAM addresses I collected:
003F ypos
0041 xpixel1
0045 ypixel1
0046 xpixel2
0047 speed (fastest when 1)
0048 ypixel2
004B xpos
0077 bubble timer
007A RNG
008E RNG
0090 RNG
0099 block timer
00B2 milon hp
0615 boss hp
0638 enemy hp (balloon)
0639 naruhito hp
07BE kill count
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
scrimpeh wrote:
I didn't know Milon was on Gameboy also. Are all the world layouts the same as the NES verison? If so, I'd be curious to see what route would be possible with the glitches, and how fast the TAS would end up.
There is nothing radically different in the layouts from what I've played,though I haven't gotten too far, so don't take my word for it just yet. Each level is understandably quite smaller than on NES, so definitely not complete parallels, but they should still be familiar if you know that version.
The biggest strategy difference will probably be from the fact that items in the GB version are cheaper to buy than in the NES version, so maybe it's not necessary to enter the first room like in the NES TAS. Also, unlike in the NES game where the umbrellas are fixed from killing enough enemies, here they're fully random, like the hearts. So it's possible to upgrade your weapon a lot easier.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
This game is glitchier than its NES counterpart. Milon has the ability to walk through walls with ease. He can also cause enemies to horizontally glide around the screen by pressing left and right at the same time. I made a demonstration here.
Link to video
I've been working on a TAS of the NES game for a long time now, but this might be a good co-exist project if I ever finish that one.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Link to video
I made a TAS of this game in the past few days that beats the published run by 2471 frames. It is still pretty improvable (I've started a v2 that finishes the first four levels 5 seconds faster) so I won't submit it, but it is much more optimized than RachelB's movie, so I figure it is worth posting for any fans of this game or the published TAS lurking about.
Most of the time saved is just from tighter movement in every level. Though there are a few "route changes" to trigger some switches earlier, I forget exactly which levels are different in that regard. I also beat the jaguar boss a little faster by using the second rock to deal two hits of damage.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
FYI:the list of active pre-500M submissions on the first page is missing Kendo Rage and Deja Vu 2.
Also, I looked at Super Smash TV recently, but it's a more complex case than I anticipated, since the current movie only sort-of aims for fastest time. So, IMHO, it would only be appropriate to obsolete it with something of equal entertainment, which isn't easy to decide on your own.
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
Yes, I intend to stick with the Japanese version. Given the sensitivity of luck in this game, there is no chance a run made on the U.S. ROM would sync on the Japanese ROM and vice versa, so, in my opinion, it is probably best to be directly comparable with the published movie.
I didn't mention this above, but it should be noted that L+R in this game allows Milon to walk right while facing left, which saves time killing those two enemies at the start of the first room. I also use it in the WIP to kill the fourth boss, but it's slower than an optimal fight with a better luck pattern would be (like in Hisatoki's TAS).
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
I have been slowly grinding out a new TAS of this game for quite a while now. Here is what I have so far (made with BizHawk 2.4.0):
Link to video
Hisatoki's movie is very tightly optimized, and has hardly shown its age in the way other decade+ old TASes tend to. There is not much obvious room for improvement. I would normally leave it alone, but, unfortunately, the TAS was recorded with a bad dump, and cannot be played back on any other ROM because the randomness is totally different. Because of this, I've set out to make a more accurate run of this game, using a new emulator and a verified good dump.
To the naked eye, my WIP most likely does not look all that different from the published TAS, because the lag reduction/movement optimization is far from immediately apparent. In my opinion, there are two timesavers that stand out:
- In the well, it is possible to jump after grabbing the balloon, which shortens the slow ascent to the ceiling.
- In the last room (in the WIP), it costs no time to turn into little Milon, which increases precision when turning.
The biggest barrier to making a good TAS of this game is the appalling way this game handles randomness. The RNG in this game is completely frame-based, and does not change based on player control. In addition, while certain luck-based things (like item drops from enemies) change from frame to frame, others (which are more important, like the position of enemies when you enter a room) have something like a ~10 frame cycle to change. It is hard to describe how annoying this is, because killing enemies to upgrade your weapon is a major part of the run. As a consequence, it is not rare for any savings you have to get quickly neutralized just because the game feels like it.
My WIP is generally optimized, but I am not entirely satisfied with it, so it will likely be redone. I was handed rotten luck on the fourth boss, which eliminated a lot of what I had saved. But it's a good proof of concept anyway.
It doesn't seem like there is a massive speedrunning community for this game, but since the current run is, as stated, very old, I doubt I am the only one who has expressed interest in TASing it since its publication. So, if you have any other info, or just want to express interest of your own, feel free to share. (I could've made this post more detailed, but it seemed like overkill for an introduction.)
Experienced Forum User, Published Author, Skilled player
(1173)
Joined: 7/13/2015
Posts: 102
Location: California
That's pretty cool, thanks for posting. There were a handful of other minor optimizations pointed out in the original submission threads, so it might be worth taking a look at those runs again in the near future...
