Sorry if this has already been asked, but is a limit of enemies/sprites known? I think there are only 4 enemies that can be on screen at a time, and I have know idea how many sprites can be loaded at one time, so want to check.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
Well, imo, I think a glitchless TAS of Kanto should use Firered/Leafgreen, since the original versions have too many glitches to really warrant it begin "gltichless". For example, all moves there, even the ones that supposedly never miss, can miss in Blue/Red.
The difference is that I'm not suggesting glitchless, I think off skipping to the credits as a game breaking glitch, so the lack of skipping to the credits but skipping brock is a different category completely, even though it isn't glitchless and isn't possible in firered/leafgreen.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
I'd have to agree with this, but linked movie is very sub-optimal. Despite this, I think that it should be done, and added as a new category. This run plays the "full game" e.g. skips warping to the credits, whereas the next one does not. They are completely different runs with completely different categories, so IMO they should NOT be in the same category.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
No you can't actually get into the first pipe 1 frame faster. It doesn't matter which frame Mario disappears. Any actual 1 frame improvement would save 21 frames in 1-1, but that's very unlikely to be happening.
very unlikely to happen* (wouldn't normally, but I think you said something on your page about helping you with your english?)
Of course, the published movie is of a very, very high quality, but I have to search absolutely everywhere for improvements until I'm absolutely certain that there aren't any. Of course, there could be some unfound glitch, but I'm not a great glitch hunter. No disrespect, just my instinct to look at everything.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
Considering that it takes particular places for Link to get out of a wall (maybe he won't at all, I believe Link actually walks below the floor, preventing his escape), this technique may not be as useful. If we find a way to exit a wall whenever, it could be a more useful strategy.
The way it works (roughly) is as follows:
To get for the different floors in the game, the game has to use layers. When you use EG, you trick the game into thinking that you are still on the same layer you were, but actually being in a place where you should be on a different layer. To get out of it, there are certain rooms you can enter that will get you out of it automatically.(?) (tompa has a map of things to do with EG and what the certain rooms do if you enter them) But I think you can also switch between light world and dark world (using the mirror? I don't know LTTP very well), I can't really remember. I had a conversation with tompa about this about a month ago, so I can't remember the specifics very well, please anyone, feel free to correct me if I'm wrong.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
This is just my opinion, so it might not apply to anyone else, but I personally feel that this is the right way to do things. If you let 1 frame improvements not make a difference, people would then start arguing over 2, then 3, then 4, and so on. After all, it is only 1 frame more than the previous. It becomes a lot more simple just to say that if it's quicker then it's better (with a few differences here and there where slower movies obsolete due to entertainment). Despite only being 1 frame quicker, it is still 1 quicker, so yes, I think that it should obsolete for just 1 frame improvement, no matter how long the game is.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
uhh... i thought that saving 1 frame in 1-1 was actually enough for that frame rule, so theres always that...
It is, but there are different frame rules throughout the stage. When you enter the above ground pipe 1 frame earlier, it doesn't save any time, but if you got the flagpole 1 frame earlier then I think that is when time would be saved.
EDIT: Apparently I mis-interpereted when "getting in the pipe" is counted by the game. it is based on the first frame where pressing down will get you into the pipe, the published TAS does on the first frame possible for their positioning. That frame was frame 563. The same section in my TAS: first frame possible is 563. So it doesn't actually get down the pipe any quicker according to the game. I assumed that it would be when you cannot see anything of Mario, my frame is 579, the published TAS is 580. If this were the case, it would save time, although it actually may still've been on the same 21 frame cycle, I'm not entirely sure. That part of the published 1-1 is perfect, although I prefer the way I enter the pipe because despite being inside a wall, the game doesn't slow you down. However, it is a very slight difference that would be likely to go unnoticed by many people, and almost all of the casual viewers. I think being 1 frame quicker at the flagpole would be a 21 frame improvement as well, so although I doubt there will be one, I will search for an improvement. I will also search in the underground section for any improvements that would be almost impossible to notice visually, unless you were comparing frame counts on screen black-outs or watching with slowdown in an emulator. Well, I was wrong about getting in the pipe a frame earlier in overground, so yes, that bit is perfect (unless hopping glitch can give pixel/subpixel perfect values, I will have to check for that as well.)
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
BTW it is possible to get inside the first pipe 1 frame earlier than in the published TAS. doesn't save any time but IMO is better, although it probably shouldn't obsolete it despite it raising the entertainment value ever so slightly (this is my opinion, as it looks like mario is entering the pipe ‛through‛ the actual material of the pipe rather than the hole at the top.) but the real reasom I am posting is for a different part of the game. does anyone know the maximum falling speed, not underwater and not after bumping his head?
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
Consider a tool-assisted speedrun a story.
At the beginning, they may not look like much.
But when you look deeper into them.
And find the real reason within.
You have found something that is quite amazing.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
Hi,
did anybody ever try to solve a game by using an A*-search?
(http://en.wikipedia.org/wiki/A*_search_algorithm)
This algorithm guarantees (under some assumptions) a perfect and unbeatable result.
IMO this algorithm should work very well for many old jump'n'run-games and deliver perfect solutions in a very reasonable time.
Let me describe the most important part in a few sentences, but be warned: This might be a bit theoretical!
You can read for an hour in the wiki or just believe the following statement:
The core of the algorithm is the evaluation-function of a gamestate: "How long will it at least take from here to the end?".
As long as this function does NEVER report a value that is higher than the value that is actually achievable, the algorithm will definitely find the perfect solution. That's it! Let's leave the algorithm's internals aside and concentrate on this evaluation-function.
You might want to take a shortcut and take the easiest evaluation-function possible: Always estimate the remaining time with "0".
You can do that, since this function fulfills the requirement: NEVER report a value higher than the value that is actually achievable. Thus the algorithm will eventually find a perfect solution.
BUT: With this function, the algorithm will basically brute-force the game, which will take far too long.
You see where this goes: The better the function is at estimating the remaining time (without ever returning a result too high), the faster the algorithm will terminate.
So we should provide a function that comes closer to the actual results.
If you take for instance Super Mario Bros:
Just return the time it would take to run from the current position to the flagpole at top-speed. Since mario does rarely slow down during the run, this simple estimate should be very good. In practice you would need some more fine-tuning to account for shortcuts inside the levels.
So back to the initial question:
Did anybody ever try to do that in an actual game?
The problem with that is that it doesn't take in account glitches. I'm ignoring the fact that mario runs at different sub-speed values because of the hopping glitch. It won't take into account frame rules, or the flagpole glitch etc. lag frames... so this is not a good way to see perfection, as it will give a value lower or higher than perfection could be sometimes. especially if you trigger an end level early. It could be done, but it's not worth the effort really. When you take everything into account, we're better off just looking for improvements in the current records.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
Do you think there's a way to find a way to corrupt the value of 0057 (speed in subpixels) to the value of 7F?(or anything above 28) It's amazing the glitches that are caused with a speed value of 7F... you can practically move through walls with it... or has this been searched for
? I don't know how it would be done at all, but if it's possible...
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
is there a TAS-editor that doesn't strech lag-input out
Everything is there, in the Manual.
www.fceux.com/web/help/taseditor/index.html?ProgramCustomization.html#AutoAdjustInputAccordingtoLag
thelegendarymudkip wrote:
or will I just have to not use TAS-editor? (The mentioned stretching is the ones that comes with FCEUX 2.2.1)
As it is now, TAS Editor is not meant to be used for multigame TASing, so you may want to switch to traditional method (the one Angerfist uses). I imagine having 6 TAS Editors on desktop would be kinda scary.
thanks, didn't see that though I searched... I'm not sure my pc likes 6FCEUX, but as long as I only have one running at a time it should be fine.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
First of all, sorry for double-posting, but the previous post was over a day ago, so it's not really spam, and this won't get noticed if I edit the previous post. I would like this to get noticed as I have something quite important to ask.
Even during the opening cutscenes, this is important. Unless I want to wait for a LONG while, if I try to make the fm3 file and use it for all of the games, I can't make it, as it stretches out input during lag frames until the first non-lag frame. This means that nothing would sync on all the games... is there a TAS-editor that doesn't strech lag-input out, or will I just have to not use TAS-editor? (The mentioned stretching is the ones that comes with FCEUX 2.2.1)
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
I think I'm missing something here, do you have 6 different .fm2 files or 1? either way, how would you get it to work?
It is 9, not 6. But all those .fm2 files should have identical input tracks, so any should work with any of those 9 games (albeit 8 get wrong ROM warning).
For my TAS, I'm just using the 6 NES MegaMan Games. But I think I understand now, thanks :)
A quick question as I don't want to double post doing it in edit, is there a more efficient way to TAS 6 FCEUX than switching between windows? e.g. getting a TAS editor to show all 6 screens and use the input for all of them?
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
I think I'm missing something here, do you have 6 different .fm2 files or 1? either way, how would you get it to work? as in for the 6 NES megaman games. or is it not possible with FCEUX? my computer won't run 6 bizhawk... it'll barely run 6 FCEUX lol
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
Just a quick question, would it be possible to do ALL the main megaman games eg 1,2,3,4,5,6,7,8,9,10? Not sure if different emulators would mean it was impossible. If so, I'd love to see (or make) that, but I'm definitely going to make 1,2,3,4,5,6 as the technical difficulty required is huge AFAIK.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
Pokemon Yellow (and presumably the rest of the mainstream first gen games): Use a glitch move thats name is TMXX. Some glitch moves crash the game (screen turns a lot of colours, and sound freezes too I think.) Not sure exactly which glitch pokemon has one of these but I think it's Q or 44HY. Been done in real time, but probably a much faster way than this, but I don't know of one.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
^how are those times compared to the real-life TAS bot thing? are they more accurate or less?
If the real life TAS bot was done on N64, then those times would be the most accurate, however if it were on VC then these times would probably be more accurate.
Current TAS:
[SNES] Jelly Boy
[NES] Street Fighter 2010
