Posts for Yumiko

Experienced Forum User
Joined: 10/29/2005
Posts: 31
Wow, I actually got farther than I have playing normally now, using my "don't fire any arrows" tactic.
I've gotten up to level 9, with 74% of the castle complete, and there was no way I could've held any longer with my present tactics.
I have now gotten to level 9 with 84% complete! Winning Level 8: http://img98.imageshack.us/img98/9623/8qd8.jpg Level 9: Screen after losing, shows the low score I got! http://img139.imageshack.us/img139/8787/9sg7.jpg I didn't use the up-down trick this time. I am trying to time my use of Z X C more strategically, though. BTW does anyone know how to make "save states" in a Shockwave file? This game is annoying in that if you lose a level, you have to start all over from the very beginning, and I'd like to be able to keep trying different things on the same level without having to start over every time.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Interesting, I just got 96% on level 8, with the no-Spacebar rule. I think I could've beaten that level if I was just a *little* more efficient. That's about as far as I usually get playing normally! (Though I have gotten to level 9, but lost.)
maybe you can use uparrow+downarrow to "exploit" and make them build faster
This trick can backfire. Sometimes, when you press uparrow, the worker who gets converted to an archer is NOT the worker who just dropped off his load, so that it gives less (or possibly even negative?) benefit. However, if I have a line of 6 guys dropping off their stones in a row, the trick will work for all of them. Also, it's risky to do this when there's enemy crossbowmen on screen. I've had this happen: press up crossbowman fires at the first worker press down press up crossbowman's arrow hits and kills the second worker who just popped up In the later levels, I only used this trick when the screen was mostly clear (especially after having used up the 'C' move for the logs).
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Here's a little challenge: How far can you get in this game WITHOUT pressing Spacebar? The point of the game is to use your archers to shoot down the enemies who approach your castle... but if you can't press Spacebar, then you can't fire any arrows at all! So you're limited to putting all your men to build the keep as fast as possible (maybe you can use uparrow+downarrow to "exploit" and make them build faster), as well as Z X C special weapons. I managed to get to level 7 (96%). I used Z X C, but maybe I could've gotten to level 8 if I was more efficient about it, and I haven't used the up+down trick yet. A "speed run" of this game would actually use as few archers as possible since it gets your keep built faster.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Lua is an example of a language that's designed for integration like this, but I'm sure there are others like Io or Pawn that'd work.
There's Ruby, too. http://www.ruby-lang.org/
Post subject: Castle Attack 2
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Interesting little game, but very hard: http://www.2kgames.com/stronghold2/minigame/ Can anyone beat level 10? I've gotten up to level 9, with 74% of the castle complete, and there was no way I could've held any longer with my present tactics.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Pasky13 wrote:
Im not sure, I havn't played in a long time. I liked the replay though. Although to me, I find it would be more stellar and much harder if there weren't just one entrance through a bridge to your base.
Well the truth is, if there were two entrances, I'd lose, lol.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
I'm not sure why they didn't send tanks... There is no way to set the difficulty level in a melee game anyway, is there?
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Is "Heart of the Alien" (the sequel of this game) available on any emulators?
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Phil wrote:
I don't like the idea. Why do you want that? It's easier with emu because it can check and tell you if your ROM match. Within emu = excellent idea. Within site = bad idea.
Because when someone's on the site, they're already in their web browser so then it's easier for them to go search for the ROM since they would be able to see the md5sum then.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Phil wrote:
Why asking that when it's already implemented in all emu?
So you mean I can open the downloaded video file in the emu and it'll tell me what MD5 the ROM is supposed to have? That helps, but I think it'd still be easier if the MD5 was listed on the webpage.
Post subject: MD5 checksums of ROMS
Experienced Forum User
Joined: 10/29/2005
Posts: 31
I was thinking, how about publishing the MD5 checksum of the ROM used to create each video? This would make it easier to determine whether one has the correct ROM to be used to playback a video. (It might also be useful when searching for the ROM, assuming any file sharing services let you search by md5sum.)
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Also my second last hit might look a bit sloppy but I had to swim in that evil electric sea-grass (?) to be able to use the same turtle (he's pretty useless anywyas ;P) to die with.
That hit was deliberately to lose health so that Raphael could die faster on the next level, right? I think you could have also switched from Donatello to Raphael one health point earlier, and avoided taking that hit (just to look less sloppy).
Experienced Forum User
Joined: 10/29/2005
Posts: 31
CtrlAltDestroy wrote:
I haven't watched, this, but is it a Melee (Everyone attacks you) or a Free For All (Everyone attacks each other) game? I've beaten 1v7c Free For All games many times, but I'd find a Melee victory to be extremely impressive.
It's a Melee (everyone attacks me).
Experienced Forum User
Joined: 10/29/2005
Posts: 31
There's still macro programs that could be used, as well as playing it on the slowest speed. But obviously without better tools it'd be impossible to achieve the same level of precision (like, not even wasting 0.2 seconds) that people do in tool-assisted console games.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Maza wrote:
mm.. what's .rep file? Not that it matters much anymore since Onega has promised to make and .avi of this. But I'd really like to know, to raise my knowledge ;)
If you have Starcraft installed, you can watch it using "Load Replay". If you don't have Starcraft, you can't watch .rep files.
Post subject: Starcraft 1v7c
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Here is a replay of me beating 7 computers on the map "Scorpion Ravine", using Terran. I chose this map because it has narrow chokepoints, making my tank defense much more effective. http://mint.aaanime.net/sc/1v7c%20Scorpion%20return.rep I'm not sure what the elapsed time is. When I played it on "Fast" speed, the time came out to be ~1:15, but when I watched the replay on "Fastest" speed, it was under an hour. I wonder if I can win faster by using more marines/tanks instead of battlecruisers, and attacking multiple bases simultaneously. My army was maxed out at 200 supply, and for the same amount of supply, marines/tanks do more damage. I'm not very good at multitasking in Starcraft though. Maybe I could pull it off if I played at Slowest speed. Thoughts?
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Paulygon wrote:
Part of the rules for having movies published here is that patched ROMs cannot be used. Hacks, translations.. doesn't matter. Since the Japanese version of this game is the only official one, it's the only one that will be allowed here for movie-making.
That's not what the rules way necessarily... http://tasvideos.org/Rules.html#2_hacked_games_in_general_are_not_allowed_because_ However, we may sometimes make exceptions of this rule, but the idea has then to be discussed beforehand (to avoid unnecessary disappointments).
Experienced Forum User
Joined: 10/29/2005
Posts: 31
You cannot save games which are played offline, if that was the case. It sounds like it, since you saved and loaded.
Ah, I just realized I could test it out in a few minutes by doing this: 1. Started game. 2. Saved. 3. Killed my own Command Center. Game over. 4. "Save Replay" option was available. 5. Loaded. 6. Killed my own Command Center. Game over. 7. "Save Replay" option was not available. So I think: * offline games CAN "Save Replay" * loaded games CANNOT "Save Replay" * a game that has been saved, but not loaded CAN "Save Replay" Damn, the fact that a replay can't be saved unless you play the whole thing straight through makes it harder to make cool replays! Does anyone know of a third party hack to make loaded games replayable? I'm guessing it could be that the savegame files don't contain the data needed to produce the replay in the first place though, so it would be difficult...
Post subject: Starcraft: Why is "Save Replay" greyed out?!
Experienced Forum User
Joined: 10/29/2005
Posts: 31
I just finished playing a Starcraft game, but... The "Save Replay" button was greyed out!! Any ideas why? Does saving and loading disqualify me from being able to Save Replay? (This was a Melee game.) Also, does anyone know how the "Elapsed Time" display at the end of the game is calculated, and how slow the "Slowest" speed setting is compared to "Normal" or "Fastest?
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Isn't "B" the jump button? If you never pressed it, what glitch makes him jump?
Experienced Forum User
Joined: 10/29/2005
Posts: 31
With two players, you'd definitely be much more powerful. You'd have twice the firepower, and you get to use each special weapon on not one, but two stages! The Megabeam, which kills *any* boss in a single hit, would be able to be used on two stages (once by player 1, once by player 2). That would eliminate the two toughest bosses almost immediately. I used to think that this game was impossible because of how one hit from anything kills you immediately. But then I played it a lot with my friend, and we ended up beating the game on Otaku (hardest) difficulty. Learning how to make our shots hit (so that enemies died soon before they fired too many shots) and dodge enemy shots helped, although our special weapon selection was pretty important too. For example, with the dual guided missile launchers, I could fly around in random evasive patterns while still killing enemies from any angle. We also made use of the invincibility shield. This weapon makes you invincible for a period of time, and you can almost use it throughout the whole level. On the 2nd last and 3rd last levels, we alternated between one of us having the shield and the other having the megabeam, and we'd combine ships so we'd drift through the level being invincible and then take out the boss using the megabeam.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
You probably mean genetic, not generic.
Omg, what an embarassing typo! Yes, I meant "genetic algorithm". T_T
I've been thinking about it, but there isn't any "learning", or room for learning, in the current system. I don't see how it could work...
My idea is, if you're trying to use a GA to optimize how the game is played for a 10-second stretch in a 30fps game (for example), then there's 300 frames to be played. With each frame, the direction pad can be in 1 of 8 directions (3 bits), the A button is either on or off (1 bit) and the B button is either on or off (1 bit). So a complete genotype for playing 10 seconds would be 1500 bits (187.5 bytes). In practice, the space of possible solutions could be shortened by eliminating redundant ones. For example, in some games you can't change your trajectory while jumping, so if you just jumped, then for X frames after the jump, the direction pad input is meaningless. Applying heuristics like this to weed out redundant and obviously-bad solutions significantly reduces the amount of work that the GA has to do. The genetic algorithm would initialize a random population of 1500-bit strings. Then it would play through each one and evaluate how well it did. Then it "kills off" the bad ones, reproduces new ones using crossover/mutation, etc. and starts the cycle again until some time limit is reached, or a solution that performed above a certain threshold is found. The theory is that if you take a good solution and change it around, you have a better chance of getting a better solution than if you picked something at random. (This probably wouldn't work if you're trying to manipulate luck since the outcome is so random, but for doing something more normal it could.)
Experienced Forum User
Joined: 10/29/2005
Posts: 31
If I'm watching a video of a game that I'm not very familiar with, then I'd prefer to be able to have time to read the dialog. But in speedruns they dismiss the dialog as soon as it comes up.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
I wonder if Generic Algorithms would be of any use here? Although calculating the fitness value for just *one* genotype could be quite expensive.
Experienced Forum User
Joined: 10/29/2005
Posts: 31
Truncated wrote:
Unless the cheat code is in-game (press up-start-A or whatever), you need to give it to us, because those things are not stored in the movie file and it is likely to desynch without it.
Ahh, I didn't realize that. The code is: RFST-G6ZC Thanks for telling me.
> jump-kick, slash, wall-jump, dashing slash, etc... all performed with a single action button. No, they're not. Actually not a single one of the attacks you listed are executed with the same button command as the others...
They're all executed by pressing the "attack" button (but what move comes out depends on whether you're standing, running, and the directional pad), I meant... (oops, wall-jump uses the "jump" button instead, I made a mistake on that one)