Experienced Forum User, Published Author, Player
(70)
Joined: 4/16/2024
Posts: 11
feos wrote:
What's the full workflow to TASing it? How do you grab and run the game? Does it have any kind of redistribution license?
The .sb3 file can just be downloaded by going to the project's editor and hitting File > Save As... . From there, the best option would be the TurboWarp Project Packager (https://github.com/TurboWarp/packager), which converts projects into Electron applications, and then running that in libTAS. Unfortunately, I haven't gotten this working so far as it complains about not being able to connect to socket.
If that is truly impossible right now, the best workflow for running the game would probably be to convert it to a .sb2 file (as of right now, there's https://github.com/RexScratch/sb3tosb2/releases/tag/v1.0, and I will try to document and check the script to see if it's converting the file without compatibility issues), and then converting that to a standalone SWF using https://amys.website/scratch/. From there, the SWF file can be loaded with Ruffle + libTAS as usual.
I should mention a couple of things with the standalone SWF converter: there is source code attached, but the website has a chance of going down so that's not good. This might also be unnecessary if importing projects works while TASing (though to be honest it probably doesn't).
There's also technically the option of just remaking the whole game within the TAS (with Scratch 2 Offline Editor + Ruffle + libTAS)...
Experienced Forum User, Published Author, Player
(70)
Joined: 4/16/2024
Posts: 11
feos wrote:
poipole807 wrote:
I want to submit a TAS of a game obtained through a game library. The thing is that this game library allows for a game to be updated at any time, and there's no version history. The author of the game hasn't been active in years so there's no threat of my game being changed, but is this still a problem? If it is, can I upload the game somewhere else for eternal preservation?
Experienced Forum User, Published Author, Player
(70)
Joined: 4/16/2024
Posts: 11
I want to submit a TAS of a game obtained through a game library. The thing is that this game library allows for a game to be updated at any time, and there's no version history. The author of the game hasn't been active in years so there's no threat of my game being changed, but is this still a problem? If it is, can I upload the game somewhere else for eternal preservation?
Experienced Forum User, Published Author, Player
(70)
Joined: 4/16/2024
Posts: 11
Very nice run! The routing looks very optimal and I'm excited at the possibility of a Sun/Moon TAS with this TAS.
Anyways, after trying to look into the sync issues, here's what I've found:
For anyone wondering why they can't get the TAS to start with an in-game time of 11PM, it's because you need to set your computer's time zone to EDT due to a bug with BizHawk 2.10.
If Froakie is not fainting on frame 15295, start over.
This never happened to me so this should be a non-issue as long as the TAS is started after having closed and re-opened BizHawk.
Make a savestate on frame 78801 and reload it. This fixes a desync in the text mashing.
This is the biggest concern with this TAS. This issue seems to be caused by memory that isn't set properly when a savestate is loaded, causing the loaded savestate to be incomplete. However, I'm not really sure what part of memory isn't being set properly; I tried to not use a temp folder on the movie and swapping out the 3DS folder in BizHawk (thinking that it's a SRAM issue) but it seems to have no effect. Hopefully this can be resolved, as it'd be a huge shame if this movie can't be accepted for technically requiring an illegal memory state.
Experienced Forum User, Published Author, Player
(70)
Joined: 4/16/2024
Posts: 11
Xander wrote:
Also, within the submission they said that they've already found an improvement. And after watching it, within 5 minutes I found a small time save grabbing the last tree in CCM by waiting out the squish animation. I suspect there are more improvements throughout. No vote from me.
I feel like this is being way too harsh. The improvement to the movie was something that could have been used in the 120 star TAS and an IL TAS, and yet despite that, wasn’t found until much later. It’s completely unreasonable for this one person to have found it by themselves.
As for the CCM improvement, this is, in my opinion, way too small to be considered detrimental. Unless you’re an SM64 TASer (like you are), this would plausibly look like the fastest way to grab the tree: you can’t grab the tree while squished, so stalling a bit and side flipping onto the tree when Mario becomes un-squished seems reasonable (and to be honest, I still don’t understand what the improvement you’ve found is supposed to be). Unless you can find a timeloss that is way bigger, this can’t really be considered too suboptimal in my opinion.
jlun2 wrote:
Would it make sense spending some time in Bob-omb Battlefield building up speed, to quickly grab all the trees there later?
That’s also the same question I’m wondering. It seems to me like you can BLJ on the green rock next to the underground cage and go between the trees on the platform where the red coin star spawns, but maybe there’s something I’m missing.
The biggest problem I have with this TAS is that it’s an “all trees” run and not an “all poles” run. For anyone who doesn’t know, there’s no functional difference between a pole and a tree; if you took away every single texture from the game, there would be no way to tell what’s a pole and what’s a tree. To me, the decision to only climb on trees and not poles feels completely arbitrary. Why not go to JRB and climb on the weird pillar things? I think that would be a fun addition to the TAS and not detract from its entertainment or novelty.
Experienced Forum User, Published Author, Player
(70)
Joined: 4/16/2024
Posts: 11
CasualPokePlayer wrote:
Where did you even read that Atari2600Hawk was not acceptable for submissions? All BizHawk cores are acceptable for submissions, as long as they aren't marked as "Experimental" (i.e. an "unreleased" core).
I presumably got confused by the BizHawk repository, which states “In the table below, core names in bold are accuracy-focused and acceptable on TASVideos.” (Atari2600Hawk is not bolded in the table.) It’s correct to say that it’s not directly stated in the sentence, but “core names in bold are […] acceptable on TASVideos.” implies that core names that aren’t bolded aren’t acceptable on TASVideos. I now know that that’s probably not the intent, but I imagine a reasonable amount of people will misinterpret what that’s supposed to mean. Tl;dr, I wish that the wording there would be changed to be a little clearer.
(Part of the confusion was also the fact that I swear I remember reading somewhere that the QuickerNES core is not acceptable for submissions. Guess it must’ve been faulty memory.)
CasualPokePlayer wrote:
A bug report would be appreciated, although just looking at this quickly myself I'm confused on how such a bug could even happen as there have been no emulation changes to Atari2600Hawk between 2.8 and 2.10 in the first place.
The TAS author is saying it doesn’t work in either 2.8 or 2.10, not that there is any difference between them.
(I’m realizing that this is getting a bit off-topic, so I’ll try to avoid continuing on the matter.)
Experienced Forum User, Published Author, Player
(70)
Joined: 4/16/2024
Posts: 11
Spikestuff wrote:
That issue is only present in the Atari2600Hawk core.
Well, this doesn’t really matter for TASing because the Atari2600Hawk core isn’t even an acceptable core for TAS submissions to begin with. It’s a shame that the TAS uses such an old version of BizHawk because of a bug in a core that isn’t even acceptable for submissions. Is the Atari2600Hawk core the default one for Atari 2600 games? I would assume that the more accurate one would be the default one but maybe I’m missing something.
Experienced Forum User, Published Author, Player
(70)
Joined: 4/16/2024
Posts: 11
I attempted to run this game on newer versions of the emulator, such as BizHawk 2.10 and 2.8. However, it appears that errors occurred during gameplay. In the first level, when the character's X-coordinate exceeds 45, it gets forcibly corrected to 29, which seemingly prevents game completion. Therefore, I opted for a more reliable approach: using the same emulator as in previous submissions for this TAS.
So, is this an emulation inaccuracy? Methinks the emulator devs would probably appreciate a bug report about this, especially since it sounds like a regression.