Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Off topic: Here are the mosaics of the same pictures I posted earlier, using (almost) all pictures from the TASSnapshot package.
MarioFlying MarioBowserYoshiLuigi
EDIT: Moved images.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Raiscan wrote:
I like the Luigi one. Also, what's wrong with including obsoleted movies?
DaTeL237 wrote:
not sure how you extracted all the files Gunty, but several 'categories' are possible :P
There were really no thoughts behind only using the currently published movies :P. I happened to use these because the easiest way to get a bunch of screenshots was saving the published movie pages. If there is an easy way to get the obsoleted movie or perhaps even NesVideoAgent's preview screenshots, I'd love to know so I can include them.
I also implemented the improved tiling accuracy DaTeL237 suggested, and the results do look better. I forgot to upload the mosaics and I cannot access them at this time, but I'll try to upload them in a few days.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Basically, I first down sampled the target picture and matched the tiles based on their mean color value. In order to try and prevent from having several identical tiles next to each other, not the best tile match is assigned, but a random pick from the 5 best matches. I'm sure this method could be improved, but this seemed the most time efficient solution at the time.
I programmed this in Mathematica which is a fairly high level programming language, so I doubt I have the resources to do anything fancy like video mosaic (current memory usage already peaked at 2 GB).
Also, without increasing the number of tile screenshots, I don't know it would be desirable to increase accuracy. Typically, you would then see several identical tiles next to each other in homogeneous regions, which is quite ugly in my opinion.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Aqfaq wrote:
Cool tool-assisted paintings, alden!
Slightly off-topic: I wonder if there's enough screenshots available at TASVideos for generating a good looking mosaic.
I tried making some mosaics with all non-obsoleted publication screenshots. While the tile-set is lacking some bright colors, the result is not too bad I guess. Here are some (about 4 MB each):
MarioFlying MarioBowserYoshiLuigi
EDIT: Moved images.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
SmashManiac wrote:
I vote for locking them all, because I don't see any advantage for changing any of them.
I vote against this. Turning on the 'Sync Sound' option usually doesn't desync the movie, while generating more accurate sound effects. I'd like to keep that option. ;)
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Don't be sad! After using this for a few days, I might have two more new feature requests:
Disable/enable all manual input (for automating purposes)
Direct pausing on frame boundaries (not sure if this is possible)
That's all I can think of right now. Other than that this application is just perfect and I'm pretty sure I'll be using it for every future TAS I make. :)
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
I'm not sure if this is possible yet, but it might be nice if there's a way to make the emulator pause without having to stop the script or use the "memory.register" function.
Something simple like snes9x.pause() or snes9x.speedmode("pause").
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
I don't think it's necessary to add arguments to the registered functions, I only wrote this (useless) function as an example.
However, when using the (edited) code DaTeL237 posted, I'm getting an error about attempting to index a nil value. Even when removing all arguments like this:
myFunction = function()
snes9x.message("Written: 7E0521")
end
memory.register(0x7E0521,myFunction)
while true do
snes9x.frameadvance()
end
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
I must say the new features look really nice, but I'm a bit lost on how to use the memory.register function. When I use the following code:
function func(address)
snes9x.message("Written: "..address)
end
memory.register(0x7E0521,func(0x7E0521))
while true do
snes9x.frameadvance()
end
I get an error about the second argument not being a function or nil. The same thing happens when I put the memory.register function inside the while loop.
EDIT: I know this function is kind of silly since it shows the address in decimals, but that shouldn't matter. ;)
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Okay, thanks, that worked.
But now when I try to run the emulator, it asks for libpng13.dll and msvcr80.dll. When placing those in de emulator's directory (which shouldn't be nessecary in the first place), I'm getting an error about a failed initialization.
I'm sorry to bother you this much, but I'm a bit new to this all.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Interesting glitch. But shouldn't you be able to activate this glitch by the C-missle's explosion as well? If possible, it might be faster in some instances.
Also, I recommend enabling the 'Sync Sound with CPU' option for nicer sound effects.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
5 months later...
I finally found time and motivation to tackle this beast :). I've spent a couple of days rewriting some programs and I made a fast 'preview' of what the eventual tas would look like:
Click!
Recorded using snes9x 1.51 with the following pro-action replay codes:
82E99EA9
82E99F02
This movie gets all 9 normal Iris Treasures in 36 floors and grabs a Providence and escapes at floor 37. Naturally, the actual tas will go all the way to floor 99 and defeat the jelly, but I wasn't feeling like optimising another 62 floors when the first 36 weren't optimal to begin with.
The strategy I will use in the actual tas will aim for an Iris treasure every 10 floors, so I will have the last 8 floor to worry about an item/spell/weapon to kill myself with.
As for the version of the ROM and emulator I'm going to use, I'm not sure yet. For clarity: since this run will probably last less than 10 minutes in NTSC timing, the timing difference in PAL will be 2 minutes at most. For a normal-mode speedrun, which lasts about 3 hours in NTSC, the difference in PAL is ~36 minutes.
Fun fact: the probability of finding 9 Iris Treasures in 36 floors is 2.5459 e-11
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Despite the somewhat boring nature of the game to watch, I must say I'm very impressed with this movie. I think one of the most entertaining parts of this run was the seemingly random route you chose at first, but after a while it all started to make sense :).
Next to the route, the movie also contained a lot of small entertaining parts, for example punching some random Goron in the face after winning the race, killing four (?) aliens with one arrow and much more.
I congratulate you on a TAS well made! Yes++
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
P.JBoy wrote:
Chamale wrote:
What would the other 5 be?
But seriously. In my "directories" thing, all the paths are blank. Do I have to have the game running, or nothing running, or what?
It'll be here then
<picture>
Actually, that line is empty as well by deafult (at least for me). But this is really a non-issue since SMB2j doesn't create an SRAM file anyways.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
By default, FCEU saves SRAM to the folder "sav" , but I don't think SMB2j saves anything so that could explain why that folder is empty. Try running a game that does, for example The Legend of Zelda, and you'll see a file being created called something like "Legend of Zelda, The (U) (PRG0).sav".
So I guess this problem is probably not caused by SRAM related stuff.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Watching the run in frame advance it seems the game was played on difficulty level 'Novice', the same as Sleepz' run.
I liked it, especially the ending. You have my yes vote.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
I'm also leaning towards the PAL version, since the messed up graphics cover a significant part of the run compared to a normal playthrough. Also, the time difference will only be a minute or two tops.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Thanks for the motivating words :)
As for the run, I've got another minor setback. Starting at floor 16, the simulation data again became inaccurate. I'm not going to take any more chances now though; I'm going to run the simulations for every possible floor so I'll have all the possibe data I need to finish it.
Estimated simulation time: 3 weeks (I'll try and find a faster machine though)
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
To me, luck manipulating 9 Iris treasures and beating the Ancient Cave in 10 minutes is entertaining, but you're probably right, most people aren't going to find this entertaining. I don't really care though, if it doesn't get published here, I still had fun making it.
I had considered using a weapon to kill myself, but I was rather thinking of a spell or item since I wouldn't have to equip it first.
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Small update:
I wrote an algorithm to find the fastest way to go through every floor while collecting one Iris treasure every 10 floors based on the data I got from the modified emulator, but naturally, doing this for 100 straight floors takes way too long (we're talking about ~8^170 possibilities here). So I decided to limit it to 10 straight floors, leaving only ~10^15 possibilities which is acceptable.
I started making a movie with the data, but soon I found out I overlooked something. If I recall correctly, the first 8 floors are the only ones to spawn helmets and rings and since I based all my data on the first floor, the simulations started to become inaccurate after floor 8. This means I have to collect new data for floors further than 8.
Nevertheless I continued making the movie manually, going all the way to floor 24, collecting 2 Iris treasure and Providence. I'm going to redo it from floor 8, but here it is anyways:
http://dehacked.2y.net/microstorage.php/info/1870077650/lufia2-giftmode-test.smv
Recorded using snes9x 1.51 with the following pro-action replay/game genie codes:
Experienced Forum User, Published Author, Player
(198)
Joined: 12/3/2006
Posts: 151
Agreed, that would probably be a better thing to do. In fact, that's what I initially was planning to do. However, I'm not sure if you misread my post, but actually I already have modified the emulator and let it run for a night. And I'm not really sure why you'd want to make a (complex) program to simulate the exact same thing :)
Also, I'll ask for Bisqwit's stance on using Gamegenie codes for unlocking Gift mode.
Also also, I'll be unable to TAS for 5 days as well, starting tomorrow.
