Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
sorry I don't have the original anymore, but when I tried it a second time back when this first happened it worked fine. I must have done something pretty silly somewhere along the way, I just don't know what it was.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
In my opinion the real difficulty is the question of what precedent, if any, publishing a certain game on a certain difficulty has on subsequent submissions of that game.
In the current case of Return of the Jedi, arandomgameTASer chose easy difficulty as subjectively the most entertaining choice. So if someone does happen to come along and submit a run on Jedi difficulty, what would happen to it?
Is arandomgameTASer setting a precedent that should be followed?
Can the 2 runs be published side by side?
Does a judge have to rule on preferred difficulty at the time of acceptance/rejection?
The difficulty guideline doesn't seem to address any of this, especially since the argument it is using is also subjective entertainment value.
In its current state, I see the difficulty guideline as at best determining whether a run goes to moons or vault, but in this case it should be made clear that essentially the same run can be made on 2 different difficulty levels and both are publishable.
My opinion would be that this should be stated as: hardest difficulty is always publishable/preferred, while any other difficulty is publishable if the audience finds it sufficiently entertaining. This would put difficulty level on the same footing as any of the other category choices besides the default any% / 100% and be less ambiguous. It also means 'easy' difficulty TASes would be rejected/accepted on the grounds of entertainment value, not difficulty level, more in line with other categories.
If the community is specifically aiming for this NOT to be the case, and that each game have a preferred difficulty, my suggestion would be making something more clear and less subjective, but I'm not sure how this is possible.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
version 3
emuVersion 9816
rerecordCount 7990
palFlag 0
romFilename Ghosts'n Goblins (U)
romChecksum base64:jDybBBOS+O8G9rDHQyxYnA==
guid 0A767817-53D4-B7F0-18BB-74FACB9C0DFB
fourscore 0
microphone 0
port0 1
port1 1
port2 0
FDS 1
NewPPU 0
comment author Arc v9
|1|........|........||
|0|........|........||
|0|....T...|........||
Well, the published movie certainly does start with a reset. I see Gay's point though and am also not sure how starting from a reset wasn't noticed in the submission, and more importantly, the publication process. If the run relied on Arc's instance of FCEU already having existing values written to memory, I would think this would desync immediately when a publisher went to encode it.
But maybe the fcm file contains a savestate or something, I'm not really familiar with it, but I'll look into it now, along with when the spawn is set.
EDIT: well thankfully this site has really good movie file format documentation:
[1] Even if the header says "movie begins from reset", the file still contains a quicksave, and the quicksave is actually loaded. This flag can't therefore be trusted. To check if the movie actually begins from reset, one must analyze the controller data and see if the first non-idle command in the file is a Reset or Power Cycle type control command.
So it seems that is exactly what happened.
EDIT 2: The key to enemy spawns seems to be at 00C0. It is used to set the X register which effects locations of enemy spawns.
Here is enemy spawn being set at power on:
$8468:BD A7 84 LDA $84A7,X @ $84AD = #$A0 A:06 X:06 Y:03 S:37 P:nvUBdIzc
$846B:18 CLC A:A0 X:06 Y:03 S:37 P:NvUBdIzc
$846C:6D AF 05 ADC $05AF = #$60 A:A0 X:06 Y:03 S:37 P:NvUBdIzc
$846F:99 AF 05 STA $05AF,Y @ $05B2 = #$90 A:00 X:06 Y:03 S:37 P:nvUBdIZC
Here it is after reset.
$8468:BD A7 84 LDA $84A7,X @ $84B6 = #$68 A:0F X:0F Y:03 S:37 P:nvUBdIzc
$846B:18 CLC A:68 X:0F Y:03 S:37 P:nvUBdIzc
$846C:6D AF 05 ADC $05AF = #$60 A:68 X:0F Y:03 S:37 P:nvUBdIzc
$846F:99 AF 05 STA $05AF,Y @ $05B2 = #$90 A:C8 X:0F Y:03 S:37 P:NVUBdIzc
00C0 is 0 at power on, but after resetting the value does not get reset, it is at 3 at this point.
So, it seems that the true enemy spawn at power on is the 'bad' one, oh well.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
http://tasvideos.org/userfiles/info/25373110677269009
OK here is a test WIP of the first 3 levels.
I tried to do it just by watching the youtube encode to get an idea how difficult it is going to be to figure out tricks and just how hard this game is to TAS overall.
So far it seems to be not that bad, I don't think it will be too hard to make steady progress, especially with bizhawk and taseditor.
I put a couple passes into this WIP to at least get a basic level of optimization in it, but it's nothing special.
@Tompa: if you can look that over and see how much you can improve, it will give me an idea of how much effort optimizing will take versus just making rough drafts.
I also went back to level 1 after level 2 and saved some frames just to see how bad the desync will be. I was pleasantly surprised it wasn't terrible, this game is no Umihara Kawase where it was a complete mightmare. So maybe I can rough draft more of the game while you are optimizing Tompa just to get more of the inputs in there. That is most of the work anyway.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
Seems to me like its a memory issue in FCEU 98.16
Everything I tested so far has no impact on zombie spawn.
I found only 2 ways to get the 'good' spawn. One is to start the game and let the first wave spawn in the normal 'bad' spot and then reset the game. The other is to wait on the title screen until the demo starts. If you reset at this point you will also get a 'good' spawn.
All gameplay videos I see online also seem to have the usual 'bad' spawn. The one video that had a good spawn had a non-zero high score on the title screen, so clearly it started from a reset.
DO you have the memory address of enemy position Arc? With that we can trace back to see where it is set and know for sure.
but right now it looks like unless you can verify on console that you can get that 'good' spawn from power on, the spawn from FCEU 98.16 probably is an error.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
http://tasvideos.org/userfiles/info/25327477891633667
First 8 levels, 1/4 of the total done.
I made a mistake in level 1 bonus and had to fix it, which turned out to desync everything else not surprisingly. It was very tough getting the other levels to sync, but enough runs of my script eventually got it working.
Level 8 was the first one I was able to do more or less constructively. I could decide where I wanted openings to be and put conditions in my script to match them. I'm pretty happy with the results.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
Great! Let's make this run happen!
I don't expect this to be a fast or high energy project, so for now this thread should suffice for collaboration. It wiil take me a few days to wrap my head around the game and try to understand what's going on in your wips. I'll put together a test WIP of the first few levels in my usual fashion and then we can start really optimizing it from there. I usually learn the most from iterating runs anyway.
Thanks for the offer to collaborate Tompa, I'm sure this will be a very interactive project, especially since you are already the expert at this game. Hopefully it doesn't go the way of your other attempts. I don't mind revisiting things a few times to get them right, but after that I tend not to look back, so I hope we can keep some forward momentum and not endlessly restart.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
I'm surprised that DKC 101% hasn't been updated in so long. It was one of my favorite games growing up, and I've been wanting to TAS a higher profile game for a while now, so I think I'll be giving it a try!
Does anyone know if there is anything more recent out there? Or if anyone is/was working on anything already?
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
When I try recording a .m64 it always makes the first frame blank, when i need it to be 'down', is there a way to fix that?
But I'm not sure why you can't recreate it, I'm not even using a controller, I just load the save state, press down on my keyboard, and frame advance.
Here is a savestate where mario is already in the sideways state and groundpounding though if it helps.
https://www.dropbox.com/s/sxpwkjp7rh2ofa3/Super%20Mario%2064%20%28J%29%20%5B%21%5D.st2?dl=0
If anyone else can get a .m64 from that save file please let me know how to do it.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
http://tasvideos.org/userfiles/info/25271447164618248
First 5 levels done. The bonus on level 5 still needs a little work since I couldn't get the same timer as Walker Bo, but the first 4 levels are done.
Overall about 8 seconds saved so far. In particular a much better bonus in stage 2, and a much faster stage 5 thanks to my random guessing script make up the majority of the savings.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
So I tried messing around with bob-omb's and coins and such looking for collision detection anomalies, and this happened.
For 1 frame, Mario got snapped to wall horizontally (sorry for the bad pic) while being pushed into the grating by the bob-omb.
Not really sure what to make of it, anyone seen this before? He pops back out normally the next frame.
Here is a savestate, just hold down on the control stick after it loads to recreate this effect.
https://www.dropbox.com/s/9lpfcs66wk3kxkm/Super%20Mario%2064%20%28J%29%20%5B%21%5D.st1?dl=0
You can do other interesting things from that savestate too. If you hold left on the control stick until the bob-omb explodes, you get pushed out of bounds.
EDIT: you can even do a ground pound while in this orientation. and he ground pounds towards the wall instead of the floor (but hits the grating instead)
EDIT2: also, the game thinks mario is under the triangel at -5365, even though he is only at -5211, so it must have put him there at some intermediate step
my only guess is that the game can't resolve what to do with Mario, so it rotates his sprite to at least fit him in the wall for a frame and hopes for things to change. Does anyone know the address for mario's sprite orientation?
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
Awesome!
Every level looked really smooth.
I can rarely watch any TAS to the end, even short ones like this, but something about Umihara Kawase makes it really fun to watch, and I'm impressed at the level of improvement too, well done!
Do you think you have leveled up enough to try your hand at flagitous' run Samsara? I've always wondered if there was a frame to save there, even one, but could never do it. Even TKDL just copied flagitious in all the levels in the any% run.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
I completed the boss fight with 1 upgrade, it takes ~1020 frames from first frame to last in the room.
Compared to ~770 for the original movie
Currently the movie is almost done and I don't think many more frames will be added.
1 upgrade run: 71200
current: 71700
I expected more savings honestly
EDIT: well bad news, I had always assumed the final weapon in the game has set damage, but it seems like that is not the case. It is effected by upgrades you get as well. The last boss takes significantly longer with only 1 upgrade, quickly eating up all the savings I had and then some. Final time is 72000 frames.
Well at least that question is answered.
I'll go back to my other file and give it a check for any additional frames to save, but it looks like it will be pretty much the final version.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
http://tasvideos.org/userfiles/info/25216805130435765
Here is the improvement file, it contains the zips Arc just mentioned in his post above. Saving about 5 seconds
Not having to worry about that boss sure does make things easier.
So now it's time to see if skipping an upgrade is faster. My gut feeling would be that it is, especially since I can also zip through 2-1, since I don't need to collect points.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
Ahh ok I just needed the older version of fceu, thanks.
Well it turns out finding how the game built the levels was pretty easy, but I doubt I'll be able to manipulate it in any constructive way.
But, it does seem to lend itself to the same type of random guessing that Super Pitfall did. My plan is to press random inputs, then count how many open spaces the lower level of blocks has (that is the random part.) Then I can just keep the level with the most openings. This way I can test many possibilities extremely quickly.
I'll need some other heuristics, like if the openings are actually close to the players or in convenient places, but I'll probably be able to eyeball that.
@jlun2: hmmm, I think it gets initialized the same way, it's just dependent on instruction level timing, since it is constantly doing ROR and LSRs and such.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
This is really cool! Well done Arc, under 20 minutes!
New finds like this keep TASing fun, at least for me.
I remember back when I was working on this that the level 3 boss caused me a lot of trouble. He has a couple different patterns and if you get a bad one it costs a lot of time.
There were even some improvements I had to leave out because I just couldn't get the right pattern, but now, that difficult boss is skipped!
So, I'll go over this and implement the improvements I know are there. I think this can stay on the workbench though, it shouldn't take long (level 1 and 3 will be identical still, only level 2 will change.)
Also it might be faster now to only get one upgrade, so I'll test that too.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
I tried posting it just now and I got parser error, so must be something wrong with the file.
I tried just now converting it again and it worked this time, so I must have done something wrong along the way.
Here's the broken one:
https://www.dropbox.com/s/30li0sy4f6fpgkz/ice%20climber2.fm2.bk2?dl=0
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
Hi,
I am trying to import an fm2 movie from fceux 2.2.2. After converting it using bizhawk I tried importing it and got this error:
When I hit ok, it just pops up again (I hit ok at least 20 times and it kept repeating)
The movie is a 2 player movie of ice climber.
Bizhawk 1.11.1
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
Well here is a test WIP. I saved about 30 frames compared to Walker Boh in 1, and ~2 second in mountain 2. I think I'll just stick with all bonuses like the previous one and not worry about all veggies.
There is a lot to consider here. One thing I noticed is an 11 frame rule for improvements in both the level and the score screen. It got me stuck figuring out what was going on for quite a while.
http://tasvideos.org/userfiles/info/25199078778548517
@mtvf1: That all veggies run is pretty interesting, but the FDS version seems like it has much different mountains, and even veggie placement, compared to US version.
EDIT: I can't get the current publication to sync at all, if anyone is able to please let me know how you did it.
EDIT 2: seems like holding 'A' in between levels changes the mountain layout, will have to test to find the best one.
EDIT 3: I tried playing my movie on neshawk and it desynced right away. All the loading frames lined up correctly, but the mountains were different. The game must be doing something every cycle with RNG.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
I'm thinking of working on this game next, but I wanted some input on the goal choice in the original.
Would this be considered '100%' by current standards? Should I include collecting all the veggies?
Any input appreciated here.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
@Radiant: Ah I was supposed to refer to the snes port there but made it general instead. You are right the original PC game is much better. I will correct that.
@mklip2001:
-If you are standing in just the right spot, the game doesn't detect you standing on that square, so you can walk right by it.
-In order to float correctly you have to hold down, and that makes you grab ledges you are near. I tested running in that spot and it took the same time. EDIT: oh yeah, the crouching thing doesn't put you close enough to grab the ledge.
-snakes are instant kills, but I didn't test it so I test it and see, but my guess is Nitsuja would have done so if it was possible. EDIT: nope, nothing happens.
-The glitch doesn't let you walk through doors, I guess they are still considered objects you can interact with, and not just walls.
-The PC version has another level that makes this part make much more sense, its too bad it's left out in the snes port.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
http://tasvideos.org/userfiles/info/25126562015960098
well finally made it to the end.
Not many more improvements overall.
In level 3 I manipulate the ghosts slightly better, saving about 0.5 seconds.
In level 4 I manipulate an enemy better with 2 sword strikes instead of 1, saving about 1 second. Nothing much changed in level 5.
So total about 3 seconds of savings, means that ~16 extra seconds of lag and loading time are in this movie compared to the original.
I still need to go over it some more before submitting, but I can't imagine much will change.
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
http://tasvideos.org/userfiles/info/25115734793474381
Seeing krankdud's suggestions made me go over everything yet again.
I'm glad I did, the result is now back to being faster then the published run!
Please replace the movie file with that one. I will replace the temp encode probably tomorrow.
EDIT:
For easier comparison to the published run, here it is converted to PAL, since its really easy to do:
http://tasvideos.org/userfiles/info/25127482148196260
EDIT2: temp encode updated
Editor, Experienced Forum User, Published Author, Expert player
(3602)
Joined: 11/30/2014
Posts: 2752
Location: US
Krankdud wrote:
Nice to see someone working on this!
On 4-3, there's a quicker method to do the zips that I found some time ago. When you get in the wall, all you need to do is release jump after getting in the wall, hold jump, hit left once, and then ball to zip. Works on each one. Video reference if needed: http://www.twitch.tv/krankdud/c/6950101?t=8m57s
Also I noticed you went really far to the right at the 5-2 boss skip, you only need to hit the trigger for the boss fight and you can go left immediately.
Oh that is neat, I didn't think of doing it that way, you are right it definitely saves time.
Yeah I did lose a few frames on that 5-2 skip part without realizing it.
I am in the process of going over my run again and will put in these improvements. I have found quite a bit of savings already
So I tried messing around with bob-omb's and coins and such looking for collision detection anomalies, and this happened.
For 1 frame, Mario got snapped to wall horizontally (sorry for the bad pic) while being pushed into the grating by the bob-omb.
Not really sure what to make of it, anyone seen this before? He pops back out normally the next frame.
Here is a savestate, just hold down on the control stick after it loads to recreate this effect.
https://www.dropbox.com/s/9lpfcs66wk3kxkm/Super%20Mario%2064%20%28J%29%20%5B%21%5D.st1?dl=0
You can do other interesting things from that savestate too. If you hold left on the control stick until the bob-omb explodes, you get pushed out of bounds.
EDIT: you can even do a ground pound while in this orientation. and he ground pounds towards the wall instead of the floor (but hits the grating instead)
EDIT2: also, the game thinks mario is under the triangel at -5365, even though he is only at -5211, so it must have put him there at some intermediate step
my only guess is that the game can't resolve what to do with Mario, so it rotates his sprite to at least fit him in the wall for a frame and hopes for things to change. Does anyone know the address for mario's sprite orientation?
When I hit ok, it just pops up again (I hit ok at least 20 times and it kept repeating)
The movie is a 2 player movie of ice climber.
Bizhawk 1.11.1