Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Spikestuff wrote:
Uh-- there's a userfile that's faster than this that's been out for a few days now: User movie #638716474753426825Which also has its own similar issue.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
eien86 wrote:
I think you do have permissions to set the "sync confirmed" flag, don't you?
Yes, I do. I was just waiting for the movie to be replaced, since it resets the 'Sync Verified' flag. But I can mark the original movie as sync verified too.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
After some more testing, we identified the Wii system country as the cause for desyncs. This piece of information is not included in the .dtm file, so it must be set manually. Since there is no country setting in the emulator options, I put detailed instructions in the Sync Notes.
Thanks to CasualPokePlayer, the Wii system country is about to be added to the .dtm format and thus should no longer be a cause for desyncs in future Dolphin releases.
https://github.com/dolphin-emu/dolphin/pull/13257
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Success! After following the settings from your screenshots, the movie synced first try with no issue. Thank you for sharing your settings with such responsiveness. In the next few days I will narrow down which settings are required.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Hello TheRandomMaster! Over the past four days I have been trying to sync your movie, unfortunately without success. At first I ran into the same desync as Dimon12321: at frame 93823 the ''2'' press is present in the movie file but it seems to be ignored by the game. With extra waiting time and some hex editing I managed to resync the movie, only to get further desyncs at the end of 3-2 (solved) and at the beginning of 3-5 (still unsolved), again from dropped inputs. Although a resync via hex editing might be theoretically doable, it would take an absurd amount of time and effort with a lot of noticeable extra waiting time here and there.. Potentially more effort and a less pretty end result than '''simply''' redoing the TAS from the 3-2 desync.
Of course, it would be better for everyone if redoing the movie could be avoided. So far 99.9% of the movie seems to run fine, and all deyncs have been caused by some dropped input. There might be an emulator setting that we missed. Could you go back to the setup with which you made your encode and take screenshots of Help > About, Options > Configuration > General / Audio / Wii / Advanced, and all four tabs in Options > Graphics Settings? With the Help > About tab especially we would be able to confirm that the 2407 release was used and not some of the dev builds.
(And as a side note, in the event that we don't find anything and the movie has to be redone, I would be happy to help out - for instance by verifying sync gradually.)
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
I confirm that both the original movie and the movie replacement synced for me.
(libTAS v1.4.6 with EasyRPG Player v0.8. Ubuntu 24.04.1 LTS via WSL 2.)
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
I managed to sync this movie. There was just a lot of blank frames at the very end of it. Here is a possible replacement: User movie #638705097882386416.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Casual encode with sound:
Link to video
For now I have not been able to find the Ruffle version which matches the checksum stored in the movie. Both 2024_11_12 and 2024_12_11 gave me a checksum warning on libTAS. Thankfully this was not an obstacle to sync verify this submission.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Perfect, thank you! The movie synced on my first attempt with no issue.
I was not familiar with this game, and I was happily surprised by how unique the gameplay looked. It seemed simple yet quite non-trivial to optimize, which is a common theme in the games I like to study. Congrats on your hard work, I look forward to your next Flash submissions! Yes vote.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Greetings Mikewillplays. I intend to sync verify your run, but I am having a hard time finding a game file which matches with your MD5 hash. All the versions that I have found feature an extra animation at the start (either Bigdino or Newgrounds) and as a result they do not sync. Do you happen to remember where you found your game file?
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Natetheman223 wrote:
Clearly there is a miscommunication here.
In my Spider-Man 2 TAS from a while ago, I had a situation much like this one - two movie files, one that ends on the final input of the main game for RTA timing, and one that goes on longer to allow the ending cutscene and credits to happen.
The one that ends sooner was used as the submission time, but the movie that plays the credits and ending was used for the publication movie.
Yes, you were right. I was at fault, sorry about that. I was not aware that it could be a thing, but it was clearly stated in the Judge guidelines: Wiki: JudgeGuidelines#PostCompletionInput.
On another note, congrats on getting this published! I really enjoyed the progress made on your original submission.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
"unnecessary inputs" were actually referring to the extra 'z' presses during the final cutscene and the blank frames at the very end.
But indeed there could be a discussion about exactly when the movie should end. In my view it should just be what is needed to make the last piece of credits/congratulations message appear. That's what I believe to be the consensus on TASVideos (see e.g. Post #531227 by eien86 in Thread #25573: #9202: Tegron's Linux Project Kat - Paper Lily Prologue in 07:56.55). But I did not see it anywhere in the movie rules or guidelines, so at the end of the day it is only my opinion.
I can confirm that this one syncs. However, in the final cutscene one textbox was not cleared as quickly as possible, and there were unnecessary inputs at the end. Here is a version which fixes both issues: User movie #638667544697225272.
In later submissions, I would recommend using the Input Editor in libTAS to help clear textboxes as soon as possible.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
This is my first experience with this game and I'm glad that I have watched the whole thing. Resource management looked both interesting and impressive, notably in the latter half of the movie. I especially enjoyed the life/energy trade-offs induced by the vampire form. Easy Yes vote!
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Hi lexikiq. I was able to sync your movie on libTAS v1.4.3. To sync verify it, all I have to do now is determine the game version. Right now there are two versions with identical name, region and version number (see game page Mari0). Only their hashes differ. Do you know which version is correct and which file the hash refers to?
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Mizumaririn wrote:
Thank for making a version of TAS of this game mode that I call complete, including the most updated strategies! And thanks for giving me credit for the strategy used in Goomba board.
Hi Mizumaririn, thank you for your past work! It was quite inspirationaal to have such a solid first half of the run available.
Mizumaririn wrote:
The biggest challenge to further improving this game mode would definitely be trying to reconstruct how a King Boo map is generated.
In my opinion the next step would be to understand more generally how every pseudo-random event is decided from the RNG value, whetever it's dice rolls, coin blocks, candy drops or board elements. Then a script could be created to anticipate which RNG value we should aim for and be able to plan ahead its manipulation. The fact that you found the RNG address and a candidate formula is already a big milestone!
I plan to go back to this game at some point, though not anytime soon. In the meantime, I would be more than happy to follow potential advancements from the sidelines and occasionally help out!
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Here are two possible corrections:
this first one: User movie #638653234548162807 which ends when player control effectively ends, and which is probably what the previous replacement was supposed to be,
this second one: User movie #638653234684537309 which leads to the final message in the credits ("By the way, [...] Thanks for playing!").
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
I can confirm that it syncs on 2.10-rc1 and it desyncs on 2.9.1 (around frame 30400 during puzzle 110).
By the way this also gets a Yes vote from me! I always enjoy when DS TASes take advantage of touch screen pattern recognition.
Experienced Forum User, Published Author, Reviewer, Active player
(284)
Joined: 5/20/2018
Posts: 48
Location: France
Tegron wrote:
Good morning, I am starting to re-encode for this problem, just want to ask one thing, at what point does the game desync in your case? Also, are you doing the same sync steps as it was in Project Kat and have clock_gettime() realtime set?
Also, make sure you don't have another game running while you're coding, with me when minecraft was running it wouldn't play the imputs properly at random times
Hello Tegron, thank you for your responsiveness. For now the furthest syncing point we have reached is frame 32148 (8:55 in your encode), which is the first time that file "retrysave" is attempted to be loaded. Here is an example of the error messages given by the game in the terminal: https://imgur.com/dk2L9PZ (the other messages are the libTAS debug thread logs).
A few additional notes:
The syncing notes & steps of #9202: Tegron's Linux Project Kat - Paper Lily Prologue "true ending" in 07:56.55 were followed. clock_gettime() realtime was checked, the movie was started as paused and then carefully frame advanced for the first few hundred frames (especially at frames 2 and 33, so that the clock disappears at frame 34).
Like with Project Kat Slow Motion had to be set at 12% (i.e 7.5 fps). Otherwise a desync could happen at any point in the movie.
Unlike Project Kat this game has autosaves, which is another cause for desyncs. Some Runtime options were found to avoid desyncs from autosaves (https://imgur.com/XG95WWy) at the expense of a slower frame rate (around 4 fps).
Last but not least and as I mentioned before, game loading from "retrysave" does not work properly. We tried to uncheck Prevent writing to disk but the problem remained.
In short, after some experiments we found a set of libTAS settings under which the movie syncs fine most of the time- though at a very slow frame rate. Unfortunately we have reached a dead end with game loading. So it would be very useful if you could check that the movie syncs for you again and, if it does, share your libTAS settings and any additional piece of information about the game setup (e.g. confirm whether the Prevent writing to disk has to be unchecked).
