Posts for Randil

Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
You want to use the program Joy2key for that. It binds your gamepad buttons to keyboard buttons, so assign the gamepad buttons the hotkeys on the keyboard.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Tompa wrote:
Thanks for your comments. Arne has found out a new technique which will force me to redo some of the run... This will never be finished :(.
How edit-friendly is this game? Any chance this trick might be edited in, or is it one of those tricks that will force you to redo large sections?
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
How many possible shots can you make in this game? If the number isn't too large, perhaps it's possible to write a lua bot that tests all possible combinations, to see if hole-in-ones are possible on the levels with more than one shot.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
In this game, luck can be manipulated by all input, so there are a lot of possibilities to manipulate the numbers to appear in your favour. I made a brute force lua bot that randomly checks input until it finds one where the number appears next to you. Using this bot, all levels could be beaten this way. Here's the .fcm. The high rerecord count is due to the bot. Watch with input display to see the strange input from the bot!
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
You should take a look at the movie Moon. Very good, very deep, and very fascinating.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
What video player are you using to play the movies? VLC should be able to play them without you having to download any additional codecs.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
After looking around some (I just looked at around 20 other random submissions), that one seems to be the only published movie that has this bug.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Dwedit: You are right, my apologies. I certainly don't want to get Gannon Banned. Kyman: Have you done some more work on this? It's been a while since your last update.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
It seems that the frame count field isn't working properly in some submissions. My Kid Icarus TAS has the frame count 66878 in the frame count field, while the movie file in reality has the frame count 80375.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
In this case, "legit" means that the input in the movie file does what it's supposed to, when the game is started from power-on. If only an .avi is provided, there is no way to know that no RAM addresses have been messed with. In theory, I could load up the Super Mario Bros ROM, start recording an .avi, then poke a bunch of RAM addresses while I'm playing to new values (such as what stage the game thinks I'm at, so that once I beat the first stage, I get to the last stage), and be able to beat the game in under 2 minutes. If I only post an .avi with this, no one would be able to prove that I did this feat by changing RAM addresses, or if I found a new bug that warped me automatically to the last level. Posting a movie file proves that there is indeed an input sequence that can do this, without manipulating the game's memory. Movie files cannot change the game's memory, and if they start from power-on, you know that this input does indeed beat the game. This is more of a concern with submitted movies. The easiest and best solution is to both provide movie file and .avi, but this is of course up to the author. Regarding usage of a a bot - a bot is still just a program that finds what input to use - after this step, this input might as well be recorded by the TASer. What it all comes down to is that TASes are nothing more than controller input that beats the game from power-on (without any manipulation of RAM addresses) in the shortest time. How this input was found is of no concern for its validity.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I wanted to take a break from any serious TASing and do something just for fun, so I picked up this game and TASed the first 6 levels. I aimed for maximum score, but unfortunately, it's not possible to kill every enemy in this game (only a few slip by). Again, I did this just for fun, and I'm not at all sure I will continue it, and even if I do finish it, it's probably not submitting material. If you have 10 minutes to spare, put on some good music, make yourself comfortable, and take a look: .fcm .fm2 youtube
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Ah! This was what my "No move" RAM address was for! I knew it was something! Your X speed (RAM address 00BD) won't increase whenever RAM address 055C is either 144 or 160. 005D increases by 112 each frame, so on average this occurs every 8:th frame. Depending on the value of 055C when you start accelerating, you will reach top speed on different frames.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
adelikat wrote:
Haven't we already had this submissions + debate?
Yeah, this and this were submitted not too long ago, both of them "game B" runs. The points I brought up in this thread are pretty much exactly the same as in sergio89's run. Anyway, I think that any further "game A" vs "game B" debate should be carried on in the NES thread. I don't want to see a fourth game B submission without any discussion before, and in particular if that submission is slower than the run I posted in the NES thread.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I have posted my record on game B in the Circus Charlie thread in the NES section. Here's that post. The .fcm I posted there (here) is 31 frames faster than this run. Because of this, I vote no. I also asked in the CC thread if people preferred the game B run over the game A run, but not many seemed to have any opinion. With all that said, I would be extremely interested if you can show any improvements to the published game A run! Regarding which mode is hardest to TAS, level 1 is harder on game A, but level 5 is harder on game B due to the rope trick that has to be pulled off here. Overall, I would say that with all tricks included, the two game modes are about equally hard to TAS. And I've said this before, and I'll say it again, using the in-game timer to compare runs is not a good statistic, because the timer starts counting down depending on a global timer. That means that even if run A has a lower score than run B at the end of a level, run A might still be faster. Use frame counts rather than the in-game timer to compare runs. Pausing the game at certain frames also stops the in-game timer, as shown here.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I have a few questions about the Hex Editor in 2.1.4-interim: *Clicking on "Unfreeze all" doesn't crash the emulator, but it does make it freeze for several seconds. Is that supposed to happen? *I like the feature of adding Read/Write/Execute breakpoints, but is there a(n easy) way to remove a breakpoint from a RAM address?
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Watched this with a friend today, and both of us enjoyed it very much, so consider this yes vote as a double yes vote! :)
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Nice and short! Here's another yes vote.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
What's the status on this, NitroGenesis?
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Picking up magic pellets won't be a problem, since these give 16 magic, so the remainder will still be 3 mod 4.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Well, me and Aglar have kept ourselves busy with this game these last few days. Aglar's WIP was, as mentioned above, faster than mine, but thanks to our colaborate effort, we found even more tricks and timesavers, as well as understanding more about how this game works. Jumping is quite complicated in this game, so I made a program in Matlab that mimics the game's physics engine and can test all possible jumps and pick the best one. Most of the input in the WIP is Aglar's, while I have focused on dissecting the game's physics and RNG, as well as writing the matlab program. Anyway, enough small talk, here's the WIP (first level done, just like my old WIP): .fcm-file Youtube'd Some info about the WIP: *It's 1282 frames (21.37 sec) faster than the published run. *It's 324 frames (5.4 sec) faster than my WIP. *We have 255 magic (i.e. maximum magic) after the boss, and have therefore pulled off the "infinite magic" trick - by getting 255 magic, our magic will never run out - each attack drains 4 MP, and once we hit 3 MP and attack, it will go back to 255. So we won't have to collect anymore MP in the run. There's a lot more that can be said in terms of tricks, but I think we'll leave it to the submission text when this is done. Enjoy!
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I liked it, it was short, and it kept me entertained. The quirky mechanics seemed to invite potential improvements, but it looked well optimized to my eyes. Here's another yes vote for you.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I'm sorry, I can't remember... I don't think it's an important address, though. Perhaps I should have named it something better.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Looking good! One thing you should look into is that it might be faster not to turn into the Sylvester on 3-1. It wastes quite a few frames to turn into him, and you don't actually need him on this world. The only places where it saves time to have him are around frames 21480 and 21950 in the published run, but with the new "turning around faster trick" the part around frame 21480 can probably can be done almost as faster with Buster. The part around frame 21950 probably saves some time with the cat, but I think it can be done almost as fast with Buster. Anyway, you might want to test this out, it's quite a big potential time saver.
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I have the most RAM basic addresses, it should work if you just copy the text to a .txt-file and load in memory watch.
0090 X|pixel|pos.
074D X|subp.|pos.
00BD X|speed
03DD P|meter
0515 P+|when|0-7
x055C No|move|9/A
| |
00CF Y|speed
00A2 Y|pixel|pos.
075F Y|subp.|pos
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
Experienced Forum User, Published Author, Skilled player (1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
If anyone's interested, ktwo just published his PAL console speedrun on youtube, here.