1 2
8 9 10
Joined: 10/20/2006
Posts: 1248
I'm currently trying to get my script to simulate the random events of 0-1 properly (provided you take forever to beat it). Proper simulation isn't required to work for that particular stage, but will be useful for some later ones. It's a nice stage to test my code with though. Edit: Wow, this turns out to be a lot more annoying to do than I thought it'd be. And that only for a very minimal reward. Looks like it's going to take me quite some time still. In 0-1 Donkey Kong puts down a new barrel every 136 frames. Sometimes that barrel calls the RNG the instant it has been put down, but sometimes there's also a 1-3 frame delay. This is totally unexpected and I presently have no idea how to simulate that weird, seemingly random behavior.
Active player (268)
Joined: 7/14/2010
Posts: 12
Well I think I'm still going to give TASing this game a shot, with help with the RNG it shouldn't be TOO horrible. Hopefully I'll get around to TASing this some over spring break. We'll see how things go.
Active player (268)
Joined: 7/14/2010
Posts: 12
Well I went ahead and redid World 0 in this speed run. And it came out suprisingly well, I saved a total of 14 frames over a previously thought fully optimized run... I saved 8 frames on 0-1, 3 frames on 0-2 and 3 more frames on 0-4. 0-3 was not changed. http://www.youtube.com/watch?v=Y-uz2mub2_Y
Player (36)
Joined: 9/11/2004
Posts: 2624
Yeah, that's slick. You landed even farther up in the air on 0-4, and I guess you landed on a pixel edge and popped down a few pixels to make the jump shorter on 0-1? Nice. I don't understand how you saved the time on 0-2 though. Does the hammer affect your jump any? I didn't think it did.
Build a man a fire, warm him for a day, Set a man on fire, warm him for the rest of his life.
Active player (268)
Joined: 7/14/2010
Posts: 12
No if you watch closely, I fall off the edge on 0-2, since I'm lower when I jump I land on the next up platform quicker.
Joined: 6/24/2007
Posts: 119
It appears there is a 53:46 TAS http://www.nicovideo.jp/watch/sm12582961
Former player
Joined: 1/17/2006
Posts: 775
Location: Deign
That run is absolutely sick! here is part 2 http://www.nicovideo.jp/watch/sm12583066 :O~~~ wtfwtfwtf i wonder how does he fall forever like that>?
Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign aqfaq Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign
Joined: 1/26/2009
Posts: 558
Location: Canada - Québec
Love it, maybe he could just mess a bit more around the donkey kong hand, but that's negligeable. Hopefully, we can get them submit to this site or get someone re-upload it on youtube.
Post subject: Donkey Kong '94
Player (12)
Joined: 6/17/2006
Posts: 501
I was considering the possibility of doing a Donkey Kong '94 TAS in the future. However I was wondering about the following: - Currently, VBA-rerecording is the only acceptable emulator to use for submission. However, I'm not familiar with the accuracy of the SGB emulation of VBA, and since this game is so dependent on the Super Game Boy capabilities, I was wondering if lsnes wouldn't be a better emulator choice for this particular game. - There's a spring in stage 2-2 that crashes the game on the Game Boy if you jump on it in a specific way, but not on the Super Game Boy. I haven't found any documentation about this bug, so I was wondering if anybody more familiar with the Game Boy hardware architecture could help me debug why this occurs and maybe find together some crazy exploit that could be used to break the game. - I'm not sure whether to do an overall best time (like most Mario TAS) or per-level best times (like most Sonic TAS). The reason is because the total time left for every set of 4 stages is rounded up to the hundred to determine the number of 1-ups you get, and every 1-up gained that way loses time, so it may be possible to lose some frames during levels to avoid a 1-up time loss. I have a very small incline for best overall time since I don't think people usually compare individual level times for this game, but I wanted everybody's opinion on the matter.
Jigwally
He/Him
Active player (418)
Joined: 3/11/2012
Posts: 119
You may have seen it already but there's a TAS already on Nico http://www.nicovideo.jp/watch/sm12582961 http://www.nicovideo.jp/watch/sm12583066 I have no clue how optimized it is
Former player
Joined: 1/17/2006
Posts: 775
Location: Deign
If that's the run I think it is, it is quite optimized. Also another topic here though that run probably makes most/all of the discussion in it obsolete. I posted a demo somewhere in that topic of bugging/crashing the game in 2-2, but didn't even know how I caused it to happen and couldn't replicate it on my console.
Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign aqfaq Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign Deign
Player (12)
Joined: 6/17/2006
Posts: 501
I can't believe I missed the original thread! But yeah, most of its content seems pretty obsolete. As for the Nico run, I haven't done any research for current records yet, so thanks for the link! There's definitely something wrong with 2-2 worth investigating. I can crash this level very easily on my original Game Boy while jumping off the spring, but not on my Super Game Boy. I don't think I can watch jimsfriend's demo considering how old the VBM file is though, but it shouldn't be a problem considering how I can easily reproduce the bug myself. I can probably handle the debugging part too, but without knowledge of the GB hardware architecture it's going to be hard by myself to understand what's going on.
Cooljay
He/Him
Active player (392)
Joined: 5/1/2012
Posts: 468
Location: Canada
http://dehacked.2y.net/microstorage.php/info/1257050345/Donkey%20KongTAS.vbm I was planning on doing a TAS of DK94 with a friend as a team effort, but he couldn't really get the times I could so he disbanded the project. I might take a jab at it again sometime in the future.
N._Harmonik
She/Her
Joined: 8/27/2006
Posts: 502
Location: Canada
Why, oh, why do I even <i>try</i> to understand my own species?
Sir_VG
He/Him
Player (39)
Joined: 10/9/2004
Posts: 1911
Location: Floating Tower
N. Harmonik wrote:
Ahem. http://tasvideos.org/forum/viewtopic.php?t=2017
AHEM. 3 posts above you.
Taking over the world, one game at a time. Currently TASing: Nothing
N._Harmonik
She/Her
Joined: 8/27/2006
Posts: 502
Location: Canada
Sir VG wrote:
N. Harmonik wrote:
Ahem. http://tasvideos.org/forum/viewtopic.php?t=2017
AHEM. 3 posts above you.
My bad.
Why, oh, why do I even <i>try</i> to understand my own species?
Player (36)
Joined: 9/11/2004
Posts: 2624
The Japanese TAS is missing at least 1 trick. It's possible to have the stone thwamp things push you through floors (provided the floor can be jumped up through.) This will save time in 3-4 and a few other places (and was in some of my old WIPs) I'm also not sure if the author is aware that jumping is slightly faster than running. He grabs ladders as he walks past, but overall I think that's slower because of character acceleration (I'd have to double check though.) I feel like 8-8 could have been handled better, but I haven't tried. That being said, the nico TAS is overall much better optimized than my old WIPs. Beyond that I really can't say. I can't remember well enough
Build a man a fire, warm him for a day, Set a man on fire, warm him for the rest of his life.
Active player (268)
Joined: 7/14/2010
Posts: 12
Hey guys, originally I gave up TASing this game because the Japanese TAS was top notch and didn't have the drive to optimize it. But I bring a new glitch to the spectrum. This glitch was found by Tropicallo, but I went back to test it more to find some promising results. http://www.twitch.tv/moltov/c/1702903 I was able to manipulate what I think is corrupted memory to complete the level. If you can complete a level with corrupted memory can't you do just about anything? I'm no expert on corrupted memory so if anyone who has knowledge on the subject wants to shed some light on this, that would be sweet. Other things that happened include modifying my lives, my score, tilesets, causing the game to have a mind of it's own(moving mario around by itself) and yes even making mario dance. Cool beans eh?
Joined: 10/12/2012
Posts: 16
Hi people. I'm a new guy in the TASing world. I'd like to make a TAS of Donkey Kong (I didn't red each pages of this forum). I tried to TAS the 4 first levels of this game here : http://www.youtube.com/watch?v=mGEMBVVtNJo So does smo still TASing this game ? May I work with you ? What do you think about it ? Etc
Joined: 2/15/2009
Posts: 329
Level 1 Faster than OmnipotentEntity by 6 frames. I copied a lot of his input. Grincevent seems to be doing some RNG research as well.
Working on: Legend of Legaia, Vagrant Story
Joined: 6/24/2007
Posts: 119
nice, keep it up
Active player (306)
Joined: 8/21/2012
Posts: 429
Location: France
So, the RNG (if there is only one) seems to be stored at "DEFE" (2 bytes), and changes when it's used. Examples: throwing an enemy in 1-1, when it gets back on its feet, it has to decide to move right or left. In fact it looks like the RNG is called at the start of a level for each object that needs to be told which way to go at first. In 1-4, it is called when DK jumps, then once for each falling object that appears after that. I didn't find a way to make it change manually (of course you can pick up and throw an enemy for that, but it takes too much time). The player can change the starting value of the RNG by waiting before pressing start in the "pre-title" screen, too. 2 more RAM addresses: -the timer: C711 -the "sub-timer": C710 (it only changes every 4 frames) For miscellaneous info: Mario has to accelerate for 4 frames before reaching his max walking speed (181 subpixels per frame), but jumping is a better idea to start moving. A test with Mario standing still at the beginning: 1-walk until max speed, jump, then continue moving 2-jump immediately, then walk Option 2 is faster, so jumping helps moving faster, but also helps accelerating. Grabbing a wire/rope (whatever it is) puts the subpixel value to 0, and Mario moves at 64 subpixels per frame on an horizontal rope.
Active player (306)
Joined: 8/21/2012
Posts: 429
Location: France
Just posting to show a little glitch that I've found. I don't know if it's already been discovered, and I really don't know if it can be useful somewhere :P It's done by pressing down just after placing one of these blocks at Mario's position. Easy to do in real time. Link to video
Post subject: Dk 94 Credits Warp
Shiekrunner
He/Him
Joined: 11/19/2015
Posts: 1
DK 94 was always a game that had huge potential for a credits warp and the dk 94 community which includes me(shiekrunner) ,Sparkcity3 and Tropicallo, only listing people who helped finding those things. we recently found many different ways/items on how to do a Credits Warp but we all lack the ability of TASing and the knowledge that is needed to find a Credits Warp so my question is anybody interested in finding a Credits Warp it would be amazing if someone could help us with that. Here are some videos of the Glitches we found that should allow a Credits Warp. - Infinte Fall Glitch Link to video - Double Tumble Glitch Link to video Link to video Link to video
Active player (306)
Joined: 8/21/2012
Posts: 429
Location: France
Hard to say if any of the glitches found can lead to a game end glitch, but they're all promising. I'll add one, for completion's sake. MoltovM has shown the one in 5-1, where if you destroy only one block with the hammer (by letting it go at the right time) and jump in the little gap created, funny things happen: I've found something similar in 3-3, the earliest level where I could replicate that setting. You have to place the timed block (didn't find a better name ^^) where shown in the picture to create the same kind of gap, then use an enemy to go and jump in it from the right: Don't do this on VBA, when I tried it stopped and gave me error messages about some opcodes. It works with BizHawk, though.
1 2
8 9 10