Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Thanks Dromiceius, it seems to work nicely!
alden: I'm not sure I understand, do I just copy that text to a .txt-file, convert it to .csv, and then open it with Excel? In that case, how do I convert the file to .csv? I tried just putting .csv at the end of the file name, but Excel didn't want to open it.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Yes, it's me again, and yes, I have a question again: is it in any way possible to make a lua script write data to a .txt file or an Excel file? If not, I think this could make a nice feature:
Let's say that you are interested in knowing the value of RAM address $1234 20 frames after pressing the A button between frames 1000 and 2000 while recording a movie. You're not looking for the highest or lowest value of $1234, you're looking for the all the values of $1234 depending on what frame you pressed A. In this case it would be great if the lua-script could write all 1001 data points to an excel file or a .txt file into two columns, where the first column is the frame you pressed A and the second column is the value of $1234 20 frames after you have pressed A on that frame.
I have looked through the commands on qFox lua page but I couldn't find anything like that there.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
sgrunt wrote:
No vote.
This cancelled submission is 27 frames faster and achieves a time of 3870 in stage 4 (even if it does worse by an equivalent amount in stage 5).
While I try to encourage new runners as much as the next guy, please do the research first before coming to the submission queue. It'll save everyone much hassle.
I agree with all this. I think it's great to see new TASers enter the stage, and even greater that they TAS a game I TASed. Still, I think it's a little sad when they don't make sure that it's faster than existing runs, even when those have been brought to light in axelito's CC thread.
I apoligize in advance if I sound a bit harsh here, it's not my intention, but I want to make a few things clear:
Basically, I want to bring up these points, and this goes both to the author and anyone else who is, or is going to, make a Circus Charlie submission:
1. You should not compare this game B run to the published run (not that you do, this goes to anyone who wants to compare this to the published run in terms of completion time), just like I didn't compare my game B run to Phil's run, simply because they play two different modes in the game. It doesn't matter if it's faster or slower than the published run, what this run should be compared to are other game B runs, and in that light, it's actually slower than some existing runs.
2. If you are going to do a game B run, I strongly recommend that you start a topic about it in the NES forum so we can discuss it in more detail. Having TASed both game A and game B, my opinion is that even if game B might be harder to play, it's easier to TAS than game A and I don't think it makes for a much more entertaining run. On the other hand, it's not less entertaining. If a game B run is to obsolete the published run, I think it should be of the same technical quality.
3. If you are going to make a game B run, make sure that it is at least equally fast, in terms of frames, as this run. Do not compare it using the in-game timer, since that is very inaccurate. I belive that that run can be improved further, in particular stage 1, but I wouldn't be too surprised if a few more frames can be saved.
Here are the comparisons between this run and this run:
*This run starts stage 1 3 frames slower.
*Your stage 1 is 7 frames slower.
*Your stage 2 is equally fast as mine.
*Your stage 3 is equally fast as mine.
*Your stage 4 is 37 frames slower.
*Your stage 5 is 19 frames faster.
Nice job on stage 5! However, in order for me to vote yes on a Circus Charlie game B run these conditions must be met:
*The community has come to an agreement that a game B run will obsolete a game A run.
*It must be at least as fast as this run on all levels, except perhaps stage 3, where bad RNG can cause you to lose frames. In that case, you should carefully explain why you lost frames, and why it's not faster to wait to manipulate the RNG timer. If you incorporate your stage 5 strategy and get at least as fast times as my old run, you might have something. :)
I vote no on this one, and please, I urge you to discuss this game B vs game A issue in the NES forum before submitting.
I don't want to see a third Circus Charlie submission in the workbench if it doesn't actually improve this runin terms of frames, and not the in-game timer. If we suddenly go for fastest in-game time in this game, you can make a run that finishes with 5000 bonus left after each level by pausing every 16 frames.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
The "game A vs game B" choise has been discussed for this game before, and the conclusion we made back then was to go with game A.
Either way, on a game that's been optimized as much as Circus Charlie, I personally believe that if you lost a frame somewhere to the published run, that you should carefully explain why you lost a frame there and why it can't be avoided. Otherwise I kinda feel that this run is only faster on stage 5 because it does game B and not game A, and that the published run is actually of overall higher quality. (EDIT: I have to compare this carefully to my own submitted run of game B to see if you might have found a new timesaver on stage 5).
I'll need to watch this when I have more time so I can give detailed feedback, but I would also like you to adress exactly why you lost frames to the published run.
EDIT: Oh, I had forgotten, I made a run of this game on game B back in 2005, and it's faster than this submission. See here. Note that that sumission can probably be improved by now, especially with the stage 1 trick used in the published run.
EDIT2: I have now compared this run more carefully to my submitted "game B" run. Here are the results:
*You start stage 1 12 frames faster.
*Your stage 1 is 17 frames slower.
*Your stage 2 is 2 frames slower.
*Your stage 3 is equally fast as mine.
*Your stage 4 is 47 frames slower.
*Your stage 5 is 8 frames slower.
It's nice to see someone else is interested in Circus Charlie, but I'm afraid this isn't publishable material, so I'm sorry, but I vote no. I hope you make a new attempt and beat my times! :)
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
mklip2001 wrote:
By the way, whatever happened to Kid Kool? I liked your submission for that game earlier, but I haven't seen anything further about the game.
Slightly off topic again: I have very slowly been working on an improvement to my first submission, and I've been thinking about finishing it up, so it's not completely abandoned. Good thing you reminded me of that one, I'll see if I can finish it. As you might have noticed, it's a bit of a habit of mine to start a project, almost finish it, abandon it for a few months and then finish it. :)
Thanks for all the feedback.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Zurreco wrote:
Why is it that you can't reach the spring at the end of the second level? It looks like the bounce height is the same on both jumps.
Even if it might not look like it, the first bounce height is slightly lower the first time around, that's why I have to land and make another jump, to get enough height to reach it.
Zurreco wrote:
Are all of the conveyor belts you jump over in the third stage working against you, or can you ride more of them?
I took careful consideration to ride the ones working with me as much as possible on this level. Sometimes riding one too far will result in screwing up a jump later on, so that's why I don't always ride the whole way.
Zurreco wrote:
Does jumping off of the key platform cost a frame?
That is does, I did not think of this. 1 frame lost here. Darn.
Baxter wrote:
I wish you spend the time making this run on making a Rockin Kats TAS! ;)
Rockin' Kats would be really great and fun to TAS if it wasn't for the fact that it lags like crazy, making it really counter-intuitive to plan a level, because stuff that should be faster might not turn out to be faster because they cause so much lag. Hmm, I guess this is a little off topic, but still. :)
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I had a friend once who was into lucid dreaming (and incidentally was the one who introduced me to TASes). I've had a handful of dreams where I had almost full control, and that was pretty cool. But yeah, lucid dreaming would be cool combined with emulator features! If there ever was a world where anything and everything is possible, it would be in that dream. :)
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Oops, seems like I posted the script with HTML enabled, which ruined the code. I have edited that post now with HTML disabled, so try it again and let me know if you run into something else. I think the only part of the code that got broken was that line you pointed out - and your guess was pretty close, though it should be xspeed>=238. To be on the safe side, you might want to copy the entire code again, you never know if something else got broken too.
If you have questions about the script, just throw me a PM and I'll try to help out. :) I realize that it's often confusing to try and interpret other people's code.
EDIT: If you're having too many troubles with FCEUX crashing on you, you might want to convert to FCEU with this game, I've had no troubles with this game on FCEU 0.98.28.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
It looks promising, I think an optimized run could be fun to watch!
EDIT: I played around a little with this game myself today, and did a test run of the first level (just for fun). Here's the .fcm. I beat the first level with 32 seconds on the timer, but I think more frames can be saved.
Here's a useful lua script I made:
while true do
local function displayer()
xpix = memory.readbyte(0x00B)
xbig = memory.readbyte(0x00A)
xpos = 256*xbig + xpix
xspeed=memory.readbyte(0x014)
if xspeed==0 then xpixspeed=0
elseif xspeed<=12 then xpixspeed=1
elseif xspeed<=76 then xpixspeed=2
elseif xspeed<=204 then xpixspeed=3
elseif xspeed<=236 then xpixspeed=4
elseif xspeed>=238 then xpixspeed=5 end
time1=memory.readbyte(0x33A)
time2=memory.readbyte(0x33B)
time3=memory.readbyte(0x334)
igtime=600*time1+60*time2+time3
ytext=40
gui.text(10,ytext,"X pos: " .. xpos)
gui.text(10,ytext+8,"X speed: " .. xspeed)
gui.text(10,ytext+16,"X pixel speed: " .. xpixspeed)
gui.text(150,ytext,"Ingame Time: " .. igtime)
end
gui.register(displayer)
FCEU.frameadvance()
end
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
MarcBelmont wrote:
-I noticed the input has some delay. Say, I press the A button (X on keyboard) in Frame 1, but the input is recognized in Frame 2. It works that way or am I wrong?
Hi, and welcome. This is exactly how frame advance works. If you press the frame advance button on frame X while holding down A, the A input will be recognized by the game on frame X+1. So basically the game detects the input you hold down while pressing frame advance on the next frame. I hope that answered your question.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
WOW. That's pretty much sums it all up. This TAS was totally mindblowing. So many new tricks and bugs abused - you ripped this game apart completely! Easy yes vote, this one will get very high ratings from me.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I watched this today and enjoyed doing so. I'll give this a yes-vote. I think the game is great, and the TAS looked well done.
However, I agree with what Comicalflop said about the slightly low action to movie ratio. I would love to see an encode where all these Alia messages and cutscenes are cut out, and only the action remains!
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I've watched run this before when you posted it in the forum, and I think that the fact that it improves the published run by just 1 frame shows the level of optimization that DK JR has reached. I watched this again now just to make sure that it synched and such (which it did), and I'll throw in a yes vote. "Bra jobbat", as we say, and hopefully we'll see another improvement soon enough!
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
adelikat wrote:
I started Solomon's key, however, it is a lot less fun than I thought it would be. I'm hoping someone (Randil) works on it.
Hmm, are you insinuating something? ;) If it wasn't for the great amount of randomness in Solomon's Key, I'd be more motivated to work on it. Well, I might pick it up again sometime, but I have other stuff going on right now. Although I do agree that the day when all Famtasia runs have been obsoleted will indeed be a day of great rejoice.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Jeopardy! for NES is a game that hardly needs an explanation. It certainly doesn't make a good TAS, but I took it more as a programming challenge to make a lua script that wins the game with as much money as possible. I finally managed to get a working script, and it beats the game in about 11 minutes, winning 235 100 dollars (the game can't show the 100 000 digit, but it's there in the RAM. Quite a lot for a game of Jeopardy.
Here's the movie file and here's the lua script (it's quite big, so I'm not posting it in text form here.) The bot recorded all input except for the first 10 seconds. Please note that the movie is not a TAS of the game because it's not as fast possible (though it is quite fast). This was more of a lua programming challenge for me, and fans of lua bots might enjoy the script and movie, even though the movie itself is pretty dull.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
Vykan12 wrote:
Also, is there a way to run algorithms when making a TAS? For example, "if memory address X has value Y then hold A for 13 frames". I really would like a way to eliminate excessive trial and error in TASing.
Yes, this is what lua-scripting is all about. You can make a lua script that does exactly that, plus a whole lot more. The latest SNES9X supports lua. There's some more information here.
Experienced Forum User, Published Author, Skilled player
(1890)
Joined: 4/20/2005
Posts: 2160
Location: Norrköping, Sweden
I realize this may sound silly, but would it be possible to compare times by calculating frames in the .avi-file for the old run? I remember doing that when I did my first Solstice run (the published run was an .fmv at the time), and it actually worked fairly well. I know BSPlayer comes with a frame counter and frame advance, but I guess a lot of other media players do as well.
[/URL]