Posts for RetroEdit

Posted: 10/6/2024 3:00 AM
(Edited: 10/6/2024 3:01 AM)

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

CasualPokePlayer wrote:

RetroEdit wrote:
You've loaded a savestate; your rewind buffer is going to be clear.
It does not clear the rewind buffer: https://github.com/TASEmulators/BizHawk/issues/2175

In the test I did before I made that post, I was on latest dev version, rewind was enabled, and I couldn't rewind after loading a savestate. It doesn't work when a movie is loaded, but I guess it could technically work if no movie was loaded.

Posted: 10/6/2024 2:32 AM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Loading a savestate should discard any presses that would be applied the next frame (i.e., the power press). To retain it seems like a misfeature on the user UI end. To me, retaining the power press as a special case here is just obviously unintuitive behavior, (Broadly speaking, maybe it's not that significant because you can load state again to avoid having the power button pressed, but it still is a source of confusion). Rewind... literally can't work in the scenario we're describing. You've loaded a savestate; your rewind buffer is going to be clear. Storing actively pressed inputs in the savestate seems like a non-sequitur: I'm not suggesting savestates contain these presses.

Posted: 10/6/2024 1:16 AM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

CasualPokePlayer wrote:

I'm saying when loading a state it should "forget" that I inputted a hard-reset. Loading a state should rewind to the exact situation I was in before.
Savestates do not hold your current input.

This reply is a bit terse, so I'm confused. From reading this discussion about hard-reset, I'm guessing that the menu option presses it in a way where it's "sticky", where it will will stay pressed even though a literal button isn't being held, in contrast to most other standard emulated controller inputs? From a bit of casual testing, I can see the menu press is custom and independent of other interactions like virtual pad. I also notice the UI is entirely disabled when TAStudio is open, which I find interesting (I can imagine the reasoning). But if it's just an input, is there any reason why it couldn't just be reset when a savestate is loaded? At the very least there should be a way to unpress power if the game is paused and you realize you didn't want to press it? This seems to me an architectural limitation, not an inherent limitation of how power presses work.

Posted: 10/1/2024 5:37 AM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

MUGG wrote:

In Mario & Luigi Superstar Saga, when sleep mode is activated, pressing select+L+R should cause the screen to become white until you press the same combination again. This works in VBA v24 svn480 but on Bizhawk 2.9.1 with mGBA 0.10, there is only 1 frame of whiteness and the game resumes right away.

This is known: https://github.com/mgba-emu/mgba/issues/1482

Posted: 9/25/2024 8:30 PM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Late reply since I've generally been taking a break, but looks like that permission was successfully added and functions properly when I tested it.

Posted: 9/22/2024 9:16 PM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

This TAS is no joke: years in the making, redoing sections as new glitches were found when the TAS just had final boss phases remaining, and just general collaboration with the speedrunning community to make it the best TAS it could be. Was interesting and fun to see the TAS iterate as I followed along with progress updates. Yes vote.

Posted: 9/3/2024 8:21 PM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Recently Reviewer permissions were changed to remove the "Edit submissions" permission. This is a permission I previously had and found useful, both for conveniently viewing submission metadata and correcting irregularities on occasion, including emulator version info and wiki syntax issues in submission notes. Some of that need has been obsoleted by submission cataloging, but it was still useful enough that I noticed when the permission was recently removed. (As for actually Reviewing -- I admittedly have not been very active on that front -- it's more been movie maintenance and cataloging.) I also have the editor permission, and I was sort of hoping the edit submissions permission could be incorporated into it. Editors already have the "Edit Publication Meta Data" permission and it seems somewhat surprising to me that "Edit submissions" would be considered a higher trust permission than that. There was some discussion on GitHub and internally on Discord, but the forum seems to be a more proper venue.

Posted: 9/1/2024 7:46 PM

Post subject: Standard Class & "Cheats"

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

I had a few thoughts from becoming aware of the cheats issue and reading the whole discussion topic. First, I wanted to put cheats aside for a second and address the run's endpoint and general acceptability in Standard Class. There were a few posts earlier in the thread where there seemed to be doubt whether this run counted as a completion at all. To me, given the game's genre of idle game, first ascension seems like a perfectly valid endpoint for fastest completion. And the game explicitly gives a screen titled "Game Complete" at the end. Of course, you can play indefinitely by ascending with boosts so each following loop will be faster, but to me it's analogous to the movie rule that says "If there's no clear ending, end after completing the first full game loop." I'm not sure there could really be another practically achievable Standard Class goal for idle games like this. As for cheats... I'm of two minds even while typing out this post. I do understand TASVideos has certain standards of validity it tries to uphold. There's a long-standing precedent against using cheats in Standard class runs, and as I understand it, this game explicitly labels you a Cheater as a result of using the Konami Code. At first, I didn't find the secrets achievement argument compelling because sometimes secret achievements in idle games are intentionally unfair and not part of the intended progression (e.g., the Shadow Achievements in Cookie Clicker). But comparing the secret achievements to other achivements, they don't seem much different in terms of the role they play in progression. I do think there's a fair argument that these codes are more like secrets than game-breaking cheat codes (especially in light of the separate cheat menu that gives full control over the game). To me personally the primary rationale for the "no in-game codes" rule is to prevent modifications that break the game in ways that are non-comparable to a run without using codes. I don't understand why the format of these particular secrets as codes makes any difference versus other kinds of secrets present in games that can be far more difficult to find and give far more extensive benefits to the player.

Posted: 8/31/2024 4:00 AM
(Edited: 8/31/2024 4:00 AM)

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Sand wrote:

But what I'm more excited about, and what I plan to make a thread at tool-assisted laboratory for, is the rudimentary data-oriented savestate manipulation library I used to script this TAS. All the inputs come from a Lua script, which does things like find the earliest frame at which each input is accepted. ...

This sounds interesting. I've coded something similar a few years back (not uploaded online though; never got around to it. :-( Hopefully eventually.). Having other code of this nature available to reference could be useful -- if I recall, my code also incorporated length predictions to detect if loading times varied from the baseline to look for potential longer term subtle optimizations there.

Posted: 8/24/2024 6:59 PM
(Edited: 8/24/2024 7:01 PM)

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

I am not a judge, but I do want to note this sounds like a very standard software feature in many GB/C and GBA games (and I guess NES and SNES games as well probably) where pressing Start+Select+B+A performs a "soft reset". Funny enough, I've used this exact trick in Game & Watch Gallery 3 for basically the same reason. I personally didn't even think to consider this an in-game code, just as an alternative way of power-cycling, which is usually preferred in general because you have to wait for the boot logo if you just press the power button to reset the game.

Posted: 8/13/2024 1:03 AM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Getting rid of the submission and publication process definitely seems like a radical suggestion from where we sit currently. But I also will say it's something I've wondered about sometimes from my five years here. I've wondered whether the judging process best accomplishes our goals of showcasing and organizing TAS work, and whether it can accommodate an expanded scope and role as for example a TAS record-keeping site. I... find myself struggling to comprehend the full implications such a change would entail, to be honest. I would boil down my present thoughts that judging and publication serve some important roles such as ensuring at the very basic level a movie can be synced by someone other than the original author and posting a high-quality video to YouTube, which is where a lot of the broader audience gets exposure to TASes. Of course, anyone can post their TAS to YouTube and try to get traction, but TASVideos does have the benefit of an existing audience. The rules will never be perfect, but they do provide some constraints and clarity on some common types of movies that are well understood, and the current movie navigation structures are generally well-integrated with this purpose. But that's not to say the site can't evolve... definitely something to think about. I do think whatever direction the site goes, it's important that site staff isn't overworked for what is voluntary work in their spare time. We all should all be given the grace to do our best to take care of ourselves.

Posted: 8/12/2024 2:43 AM
(Edited: 8/12/2024 3:08 AM)

Post subject: Game & Watch Gallery 3

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

feos wrote:

Overall, we publish minimum amount of players, maximum amount of players, and something else that is the fastest. If your improvement idea requires the second player to always be present, it's labeled "2 players" and co-exists with strictly "1 player". If it only involves the second player in specific places, doesn't need a label. So how does the 2-player mode look in that game? Do the players cooperate, or compete, or take turns?

They compete. Most of the time they're each independently playing individually for each of them to have the 20 stars progression, but they have to both enter Judge B together, where they compete. I've already described this in my last post above, but I suppose a video is worth 1000 words; I've now uploaded a video of my full 2-player proof-of-concept here (timestamped with a link to Judge B): https://youtu.be/nHR3Km8GssQ&t=1712s It copies the original submission until the 2-player route diverges. See my prior posts above (1, 2) for further explanation/context.

Posted: 8/7/2024 11:12 PM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

TASVideosGrue is just a bot. It makes a post when a movie is canceled or rejected.

Posted: 7/30/2024 11:46 PM

Post subject: Game & Watch Gallery 3 follow-up

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

I want to mention a few things:

Do the rules actually specify co-op? I know the recent "triviality of N-player-only games" discussion implies there's some internal judging discussion about this, so maybe it's implied or assumed.
There's already an existing GB publication that is very explicitly PvP: [5642] GB 4-in-1 Fun Pak: Chess "2 Gameboys" by BlackWinnerYoshi & Spikestuff in 00:15.02
For GAWG3, 150 stars is probably considered full completion and would require both players anyway. So it becomes a case where some PvP is part of the intentional progression.

For context, the Game & Watch Gallery series has games where the main progression mechanism is stars. You can collect up to five stars in each mode (Classic or Modern)/difficulty (Game A or Game B) combination of each main subgame, generally by reaching 1000 points (200 points per star). When you collect 50 stars in GAWG3, you reach the credits, so that's the standard completion. When you collect all 150 stars in GAWG3, you get congratulated. 5 of the stars are locked in the multiplayer mode of the Judge B subgame and those happen to be among the faster stars. To be honest, I don't think a video would impart enough additional clarity to be worth encoding and uploading right now. I'd also prefer not to spoil the new TAS too heavily. Suffice it to say, it would look very similar to the video in #9128: PiePusher11's GBC Game & Watch Gallery 3 "50 stars" in 38:20.50, but both players play through all the games (technically player 2 has more freedom but still has to collect at least 20 stars to unlock Judge). Judge B is similar to Judge A: most of the optimization comes from RNG manipulation so that the transition between rounds is fast (there's a random speed associated with the foot-tapping and countdown) and so the primary player has a higher number in each round: https://youtu.be/upNPzKuIDgo?t=1757 -- in Judge B, you can only get maximum points in each round if your number is larger, which is RNG; Judge B also omits the foot-tapping animation and just has the countdown. As far as 2-player visually goes, Judge B is mirrored between both screens, since for the second player, their opponent would be the one scoring. The rest of the run prior could be identical between both players, though for presentation purposes, we might add some variety. (I can upload a video if you really think it would help, but it might take me some time to get around to it.)

Posted: 7/30/2024 6:25 AM

Post subject: Difficulty in Standard class

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

I think difficulty should be up to author discretion unless the movies are essentially identical aside from difficulty selection. Harder difficulty is not always more interesting. For instance, in one game I'm interested in, difficulty merely increases the health of enemies and just makes it longer and more tedious to defeat them, not actually more challenging or more interesting from a TAS creation or viewer perspective. For the recent Top Gear 2 submission that prompted this discussion, I watched both the new (Easy) movie and the current (Hard) publication side-by-side, and they are almost entirely identical. But nonetheless, author discretion makes sense to me. I do think it's an interesting case because it's not necessarily an intentional new difficulty branch, but a new improvement that happened to change the difficulty to Easy and also had a faster time partly (maybe completely?) from that difficulty change,

Posted: 7/26/2024 9:06 PM

Post subject: On the triviality of N-Player-only games

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

My first thought looking at our definition of a video game is that PvP games don't pose the task for the game overall to be won, but poses the task of winning to each individual player. But it's easy enough to argue that an "optimal strategy" for each player to win is to be pressing the buttons for all the other players to facilitate that win (or to ask the other players to cooperate to allow you to win). To an outside observer, this may seem odd, but I think it makes sense within the context of the site and the creative goals that happen in speedrunning (though I'm not sure how much real time PvP speedrunning there's been). I do think in terms of game choice, PvP sports or fighting game may offer additional complexities. For instance, I guess we could have "win one match" as a branch (with a better branch name). But does that mean with default settings? Can we set HP/stocks/rounds as low as possible? What about map selection and randomization or allowed items selection? It's not entirely clear to me in this context, and I guess it only matters if people want to submit these specific types of movies anyway. Regarding a related post recently where I asked about the Game & Watch Gallery 3 TAS, the goal is actually identical between single player and multiplayer: have the primary player collect 50 stars to reach the credits. The difference being that allowing linked-GB PvP in one portion of the run happens to allow a faster route for collecting these stars. I will say as far as triviality goes, the PvP component is not trivial at all: it requires extensive RNG manipulation and a decent understanding of player tactics to achieve optimality.

Posted: 7/19/2024 11:37 PM
(Edited: 7/26/2024 9:08 PM)

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

PiePusher11, Fortranm, and I have been talking about improvements to #9128: PiePusher11's GBC Game & Watch Gallery 3 "50 stars" in 38:20.50 and one potential improvement suggested by Winslinator is adding 2-player to get extra stars from Judge B (the multiplayer mode of Judge). I've made a proof-of-concept movie that confirmed 2-player can probably reach the credits a little over six minutes faster than a single-player movie can. My question: would a 2-player movie would obsolete a theoretical single player movie? After reading the relevant rules and looking at existing movies, I'm under the impression that both could be published side-by-side, but I wanted to double-check since multiplayer in this case is a two Game Boys with link cable and there's only a few movies like that. (There are some game details if relevant: a 2-player movie essentially requires the second player solely for the purposes of playing a single game of Judge B. Both players have to collect 20 stars to unlock Judge to play it in multiplayer. The primary player would collect more stars after that for overall efficiency before entering Judge B, while the secondary player needs no further stars. In fact, the second player could literally copy the inputs of the first player basically all the way until Judge B. Then at Judge B, the primary player will win, while the secondary player will intentionally lose to accelerate the primary player. The primary player will go on to finish the game, while the secondary player no longer has anything left to do. From a presentation perspective, it's unfortunate the secondary player's gameplay is largely irrelevant.) See also: related follow-up post.

Posted: 7/14/2024 12:49 AM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Splitting stars between games is an interesting thought and probably worth pursuing. There is an additional cost to adding an extra game of having to menu into an extra game, but PiePusher11 confirms it's still probably faster to split stars to avoid the slow later stars in Flagman in a 2p movie. I'm still not sure if I want to do a 2p movie, but I'm definitely considering it. I was able to port over the current movie to the 2p movie and sync all the way up to the Judge. As far as menuing goes, I found another interesting wrinkle: I was able to create a new movie that finishes the game and saves ~108 frames by soft-resetting to enter Gallery Corner. I also independently discovered that Peach's cutscene length varies, so I think entering Gallery Corner at the end is probably the fastest in this TAS. RNG manipulation is also something to consider in menu routing, but it's difficult to know exactly how significant it is for each game without more research.

Posted: 7/10/2024 11:58 AM

Post subject: Egg improvement

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

I have an idea for reducing the number of bob-omb spawns in Egg Modern, but it crucially depends on whether recreating the RNG necessary for the rest of the movie to sync is possible. I'll make a follow-up edit to this post in a few hours about my findings.

Posted: 7/2/2024 5:40 AM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Yeah, definitely could benefit from a photosensitivity warning.

Posted: 6/8/2024 3:44 AM
(Edited: 6/8/2024 3:52 AM)

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

BigBoyAdvance wrote:

Game & Watch Gallery 4 (USA) seems to fail linking together (Boxing 2P mode) in the newest GBAHawk. Is there anything I'm doing wrong? It's my first time using this emu for GBA link play.

I actually have tested all four multiplayer modes (Boxing Modern & Classic; Donkey Kong 3 Modern & Classic) in this game and they seem to work fine. I can think of one likely issue: G&WG4 is a single-pak game, so you need to go into the multi-disk bundler and load one copy of the game (which will be the primary GBA) and one copy of the GBA BIOS (which will be the secondary GBA).

Posted: 1/15/2024 12:44 AM

Post subject: Feedback

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

I only watched the first half, but it was an enjoyable watch. The captions are a nice addition and offer additional context and personality while watching. Avoiding the lock screen battles is a good strategy. I can't speak to all the movement intricacies, but it generally seemed to make sense within the constraints of avoiding battles and I imagine the submission notes provide additional details. Overall, a fast-paced TAS and smooth watch.

Posted: 1/3/2024 3:48 PM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Nominating: nymx, enderpal7, Merl_

Posted: 1/1/2024 7:52 PM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Alyosha: the work on GBA TASes in tandem with work on GBAHawk has been good to see.
Reseren: the Wario Land 4 TASes were chaotic and entertaining.

Posted: 10/23/2023 6:21 AM

Quote

RetroEdit

Any

Editor, Experienced Forum User, Reviewer, Player (169)

Joined: 8/8/2019
Posts: 152

Short and sweet. It's definitely cool that something like this could exist for the Atari 2600. Initially, I was surprised that level 8 was the easiest difficulty and level 7 was the hardest, but it seems like level 8 being the easiest difficulty was probably an oversight by the developers. Not much else to say here. I do also think it'd be cool to see the level 7 difficulty TASed; with such long move lengths, there would possibly be an additional optimization aspect of finding a winning game where the AI doesn't have to think too hard.