TAP TAS?

I'm not sure a mortal can successfully synchronize rotation and movement at every possible opportunity or even set up for it. 1 frame precision is just too precise for 100% accuracy.

 

you could do interesting things with stack shapes.


i think it would be way way faster than a human record. and i think you guys aren't really acknowledging the possibilities for saved time with strange piece placements. you'd want to get as many pieces locked on frame 1 as possible. would be awesome. i want to do it--someone tell me how i can do it.

 

i don't know anything about mame rerecording, but i found a couple discussions about it here and here.

 

Tetris for NES follows the model of UNIX /dev/urandom, continuously mixing user input into the PRNG, which is why you see its TAS videos turning preview on and off. Several other games, especially those that deal identical pieces to both players, use a PRNG that does not draw from user input. My impression is that TAP uses the latter, and in this case, a TASer can beat a real-time superplay by taking the .inp, deleting all frames of hesitation, and setting the initials as "TAS".


And as I think I've said before, the problem becomes more tractable if the goal is to approach the lower bound for Master and Death completion time rather than to try to hit it in one sitting.


Another question: What's the Japanese word for tool-assisted speedrun?

 

There is no word in Japanese for Tool-Assisted Speedrun as the culture does not exist there. Some people recognize and understand the "TAS" acronym, but many people only have half-assed understandings, not being able to discern them from non-assisted runs.

 

I've definitely heard of Japanese TAS'ers though. Videos like this lead mean to believe they use the term TAS just like we do, though perhaps fewer people understand what that means.

 

Then is there a word for "played in extreme slow motion and sped up"? Was that Prince of Persia game for Xbox with rewinds as a special power released there?

 

It seems there's interest. I think I should be able to whack together something that will make xmame (Linux .106) rerecording capable, and I'm a TAP S6-S7 rank player. Do you think I stand a chance?


Yes, this means I'm going to make the effort, but first MAME needs rerecording capabilities. I might be able to make something working as soon as tomorrow night (North American timezone)

 

I don't think many people use Linux here. If you could instead modify SDLMAME, which has a Linux, Mac, and Windows flavour, you'd get all major platforms for free and we'd be more likely to see some action.

 

I use the SDL build of MAME for Linux. Better support anyways.


Why, does the official MAME come with SDL support? Admittedly I haven't bothered checking since the Linux version came with my distro and the TGM2 patch applied nicely to it.

 

I'm not entirely clear what you're trying to ask, but the official SDLMAME homepage is here. The only truly "official" MAME is a windows commandline program which does not use SDL, for reasons I could only guess. But the SDL version is maintained by one of the core developers and is always up to date, so it's not so bad.


Apologies if you knew all this and I haven't answered your question.

 

There was some discussion about this earlier in the thread:

I think this is true. To make a good TAS, you need to take a very analytical look at a game. You need to know some common things about the game, but also need to know some things that aren't commonly known by doing some tests (by timing some things using frame advance). You however do not need any skill of the actual game. The knowledge is learned pretty fast. Really good players are mostly good at a game because of their skill, not because of their knowledge.

 

The link interests me, and I'll look into it later. I doubt my changes will have any serious effect on the cross-platform nature of the game. For the time being, I'm just going to smack it into my existing executable and run it. If i succeed, I'll probably place a video of it on youtube and offer the .inp for anybody who wants it (are .inp files version-independent?). After that, I might cough up a patch against the windows version of mame or the SDL version you linked to. Whatever works in the end. The windows version will need to compile under mingw though. We'll deal with that when the time comes.


My modified xmame already does TGM2(p) and AVI dumping with my own code. I'll just add rerecording, learn some of the internals of TGM2p and then see if I can make GM in 6:30.00 while making the invisible challenge look like a piece of cake.

 

I have seen your videos. I recommend a huge watermark saying you are cheating during the whole game.

 

Yeah, tool-assisted runs should be well-labelled.


However, if they're fast enough, it could help with the people who say that videos like the Ti GM one are tool assisted. It'd be an indication of what a proper tool assist would look like.

 

To Lost: Not a problem. I do plenty of video work for the tasvideos.org site. I'm also going to make it look AWESOME and sign my name as TAS. I'm also planning on having a video overlay with useful information, like the rank promotion numbers so you can see how far I am from a grade increase, but that'll come later.


To Rosti LFC: That's the idea. No frames wasted.


I've been doing some checking and for those of you interested, it appears MAME already has frame advance. Holding a SHIFT key while pressing your PAUSE button will act as frame advance. I've modified the savestate code, and I'm getting ready to try it out while I let it compile..


Now then, for maximum show-off value, I'm going to start with T.A.Death, then try Master in TAP. Raw speed is one possibility, but would there be interest in runs that don't aim for fastest time but aim for something else, like most gold medals (ST is a given, but gold RO, AC, CO, RE and SK would be totally showing off), or S9 before level 900?


I haven't even tried it yet, so there's every chance that a savestate bug might wreak the whole thing and make the game not sync stable. Opinions? If you could have your perfect TAP run, what would it look like?

 

Actually I was going to go by the info which is in the wiki, but same difference. I want to show the little leaking accumulator so I can say how close I am to my next promotion, so I can optimize for slamming down a quick tetris before rebuilding my stack while the accumulator is at zero. It would also be a good step to show that the movie is not played on a real machine.


I've also determined that the piece generator is deterministic once it starts, so I can see as far into the future as I want and plan. Unfortunately it does mean I can't decide my own pieces. Oh well, you win some, you lose some.


I'll be in the TASVideos IRC channel, #nesvideos at irc.freenode.net if anyone's interested. I'll be in and out though, but I plan to have something to show in an hour or two if everything works as intended.

 

I can tell you from experience that it really doesn't matter what the grade points are. The way the system is designed, all you pretty much have to do is play 750-999 well and you'll get S9.


There's some pretty non-obvious time attack techniques related to the DAS. Give me a day to put this info on the wiki, otherwise your run will be significantly sub optimal.