Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
KusogeMan wrote:
Slamo, Fortran, DrD2k9, DigitalDuck, Tompa, Chanoyu, oceanbagel don't TAS any of these, they also do not TAS racing games either. Nobody in this voting TASes racing games,
Yet.
It's worth noting that while I haven't submitted a TAS for a racing game, I have a couple that I haven't submitted because I'm not happy about how optimal they are.
I'vedefinitelyspeedransome,infactI'mintodoubledigits. So I don't think it's out of the question that I might want to TAS one or more of these in the future.
It's not really relevant though, because my argument isn't "I don't like fighting games or racing games so we shouldn't allow runs of them", it's "I don't see why fighting games or racing games need special rules, we should allow the same set of runs we allow for all games". There's a difference between starting from zero, and starting from a saved game. TASVideos tends to prioritise the former. If the latter allows the game to be played faster, then that's great - we should accept it. But it shouldn't obsolete the run starting from scratch.
I don't think DS Castlevania: Order of Ecclesia "Albus mode" by mtbRc in 18:03.21 should obsolete DS Castlevania: Order of Ecclesia by mtbRc in 34:38.32, even though the former completes the game much faster than the latter, because the former requires having played a significant portion of the game first in order to achieve.
I don't think Linux Undertale "Neutral ending, newgame+" by LukeSaward in 36:04.20 should obsolete Linux Undertale "Neutral ending" by OceanBagel in 48:51.95, even though the former completes the game much faster than the latter, because the former requires having played a significant portion of the game first in order to achieve.
As such, I don't think Wii Mortal Kombat: Armageddon "Arcade, Kreate-a-Fighter" by KusogeMan in 06:38.42 should be considered as already obsoleting Submission #9025: KusogeMan's Wii Mortal Kombat: Armageddon "Arcade" in 07:01.67, even though the former completes the game much faster than the latter, because the former requires having played a significant portion of the game first in order to achieve. The latter would be a separate branch, just like the two previous examples.
Edit by feos: URLs without https:// get parsed as relative links, fixed.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
Chanoyu wrote:
I really don't quite get what's so bad about having seperate publications for fresh starts and save starts in fighting games (and racing games, and other games where the game is still played from a 'new game' state - but in an unlocked (hard) mode). I do get the concern of it needing to be clear, also in the obsoletion chain, of what the actual fastest completion is. I guess you would generally be mostly interested in the save start obsoletion chain (which could possibly include fresh starts as well).
The problem in my mind is more that if, for a certain game, the save start is really competitive, and then someone decides to make a fresh start TAS that is, from an optimization standpoint, already obsolete. I would hope that such TASes, and taking the problem a bit broader, those TASes that are aimed more at getting a publication than showcasing the best of TAS, just don't get made, but probably that will not be the case. I can imagine the rules should give leeway for rejections in such cases, even if there is no previously published movie in that specific branch.
Movies that have issues with optimisation can be and often are rejected even when there is no previously published movie. Just look at the recent Segapede drama for an example.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
For fighting and racing games that have additional fighters/cars that you can replay the game with, it makes sense to me if the branches are labelled something along the lines of "initial character/car" (which would start from a fresh power on) and "any character/car" (which would start from a save). If the fastest character/car is accessible from the very start of the game, then naturally that would be the only branch, and/or would obsolete the any character/car branch; but I imagine in most cases the fastest character/car will be locked at the start of the game.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
KusogeMan wrote:
TL;DR The slower runs are here because you wanted to enforce a specific ruleset that doesn't respect the ruleset from the game's specific speedrunning or competitive community. Not because the community values worse and slower runs. Branching will let people find the slower irrelevant runs for no good reason.
What if a game has multiple communities with conflicting rulesets? Splits happen all the time because of differences in opinions on what the rules should be.
How many games' speedrunning or competitive communities time runs from power on? Should we stop doing that too?
Let's take a look at the Mega Drive Road Rash games on SRC as an example. Road Rash 1 has categories for Level 1, Level 1 & 2, and a custom "Rash Mode" category that isn't recognised by the game at all, but no category for actually... beating the game. Road Rash 2 has each Level individually (but only using specific passwords, meaning you could beat these times in a full game run), and Level 1 & 2 and Rash Mode again, and still no full game category. Road Rash 3 has each Level individually, and, finally, a full game category (marked as "100%"). They don't even agree on timing rules between games.
If I were to submit a Road Rash TAS, could I just do Level 1 & 2 and call it a day, because that's what the Road Rash "community" has agreed on, even though it doesn't even come close to beating the game? But then it wouldn't be good enough for Road Rash 3, which doesn't have that category, and would require either stopping earlier, or actually beating the game.
What if the categories changed? Do we have to reject TASes that were technically excellent and completed the game because some other people not part of the TASVideos community changed their minds on what counts as a category?
This is one of the larger problems I have with the user-driven speedrun sites in general, and I greatly appreciate what TASVideos does in terms of trying to create consistent rules that apply universally to all games. I really don't think "here's what other people do for this specific game" should dictate what's done on the site or encourage game-specific rules.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
Spikestuff wrote:
DigitalDuck wrote:
In any case, it's effectively just NG vs. NG+, and it seems wrong to me that a NG+ run would obsolete a NG run. Just my two cents.
Only that it isn't. The only difference in terms of "new content" for a fighting game is that some titles has a rival character that shows up halfway as a mini """boss""".
A new character is technically not new content, for a fighting game. Therefore it should just obsolete.
Given that you described them both as new, the only logical conclusion from this is that characters aren't content, which is a weird thing to say about fighting games.
But ignoring that, in many cases NG+ does not add new content, it merely allows you to play using your character's state from a previous playthrough. What new content is added in Sonic 3 & Knuckles "newgame+" that wasn't there in Sonic 3 & Knuckles "Sonic 100%"? It's NG+ because it allows you to replay under conditions that aren't available when you first play the game.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
DrD2k9 wrote:
I don’t like any save-anchored run obsoleting a run based on a clean start.
I don’t see a point in having a save anchored run be slower than the clean run unless it introduces a significant amount of new/different content. I personally don’t see a different move set (from an unlocked fighter) as qualifying as significantly new/different content in terms of gameplay. So for fighting games, my perspective is as follows:
A clean start run can only be accepted to standard using the optimal base character.
A save anchored run using an unlocked character can only be accepted to standard if that unlocked character is faster than using the fastest base character. This would not obsolete but be published along side the clean start run.
If multiple unlockable characters are more optimal than any of the base roster, then only the most optimal unblocked character is eligible for standard publication as a save anchored movie. Unlockable characters can obsolete each other’s save anchored publications, but cannot obsolete the clean start run.
If no unlockable characters are faster than the fastest base character and no new content is introduced, then there’s no reason to have a save anchored branch in Standard publication for that fighting game.
The branch name needs to indicate that the run is save anchored without using the fighter’s name in order to prevent potential confusion that the site would accept all character runs.
This seems like the most reasonable way of handling things to me, if the unlockable characters aren't significantly different gameplay-wise to the base characters.
To me every save-anchored run carries the implication of all the additional time it takes to get the game to produce that save file. If you've just bought the game and started it, you have access to the base characters only. Having access to unlockable characters is only possible if you've already completed the game with a base character at least once, and one would assume that completing the game twice takes longer than completing it once.
I see more reason in not allowing unlockable characters to standard than I do in having them obsolete base characters. The best defence I can think for having both as separate branches is that they're both "fastest completion", but with a difference of timing method - standard characters are fastest completion from absolute zero (no save, power on), whereas unlockable characters are fastest completion from power on, but having already played for a significant amount of time and created saves beforehand. But then that leaves the question of whether a run that prioritises IGT over TAS time should obsolete, or exist in a separate branch from, one that prioritises TAS time over IGT.
In any case, it's effectively just NG vs. NG+, and it seems wrong to me that a NG+ run would obsolete a NG run. Just my two cents.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
nymx wrote:
I have finalized my judgement of this run. The only issue that I found, came from watching Faddy's run against yours. There are times that I see Faddy ahead (early in the game) and you catching back up. Then you are ahead, but get behind again. Eventually, you are growing in your lead and this never happens again. Can you please explain this?
They're using different routes. In the TAS when crossing the clouds for the first time, I pick up the cloth duster and move it to the left side and deathwarp to avoid backtracking across the clouds, which saves a few seconds later. Faddy collects the cloth duster on the second visit to the clouds and has to make an extra screen's worth of journey as a result.
Faddy's route was based on an earlier version of the TAS, after I discovered horizontal item warping but before we'd decided the best places to perform deathwarps to make full use of the two spare lives. As a result, there are no deathwarps in Faddy's run.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
GJTASer2018 wrote:
One concern I have about this is that you have the "flying hearts" cutscene twice, once after rescuing Daisy and the second time once you collect the 30th coin. I know the second one technically occurs after the last input, but couldn't you still skip the first one by getting all the coins you need before rescuing Daisy?
Even if you have 30 coins when you rescue Daisy, you still have to go and meet her at her hut, so nothing is skipped.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
The issue with the Dandanator Mini is that it's a homebrew peripheral, and as such can be made to do pretty much anything. If it's just about loading times, we have SD card readers that just dump a snapshot of the game straight into memory - why bother with the cartridge emulation?
If converting from tape to cartridge via a homebrew peripheral is allowed, can I just edit a tape file to remove the loading screen? There's less changes to code in doing that, and plenty of people did that back in the day.
Personally I'd rather stick to "official" releases hardware-wise, and as close as we can get to that software-wise as well. As much as it's annoying seeing every microcomputer encode flooded with comments about the loading, I'd rather that than what's effectively starting with a savestate.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
LogansGamingRoom wrote:
very interesting stuff! do you think a fully optimised TAS could be made with CPCHawk yet?
When I was TASing the Spectrum version of Dizzy: The Ultimate Cartoon Adventure I gave the CPC version a go at the same time. While it technically ran fine, the timing was clearly wrong - the game was running at about half speed (and not the emulator, as the music was running at normal speed).
I wouldn't recommend it in its current state.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
I was going to make an encode but I couldn't seem to find the same version of the game used here, and couldn't make it sync with any of the four versions I have. That's a me problem though.
One thing I will say is that I notice this was done on a 48K, but I think this game would probably run slightly faster on a +2A? Was that something that was considered?
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
nymx wrote:
Because there is no feedback...there is one thing I'm wondering about. The re-record count.... 693?
I've wondered this on previous submissions, but feedback and my review never have shown anything standing out. Are you "slow recording" with the keyboard and then going back to clean them up with TASStudio? Not that I'm saying that a problem exists, but I have been curious.
Pretty much. I usually have TASStudio open the entire time but I'm playing through with the keyboard and frame advance and using TASStudio to jump back to previous frames, and I don't think any of that contributes to re-record count? It's usually only when I'm trying to time a run past a guardian or I'm using branches to test different routes that I'll actually use the piano roll.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
InputEvelution wrote:
The duration of time it takes for Windows XP to actually finish booting up and take you to the desktop is rather inconsistent between different time-and-dates in libTAS (with half of the outcomes being a full 17 seconds longer!). I've documented the duration and outcome of some of these possible starting times in a spreadsheet here: https://docs.google.com/spreadsheets/d/1_j1zkYPPHR4jLLW3bybvnjsN6K7sV_vWZHLm7X5SwaE
Every "long" load occurs during daylight savings time, and every "short" load occurs outside of it. I wonder if that's relevant.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
Technoturnovers wrote:
I tried the first stage again, forcing center fuel spawns, and I didn't gain any time. hhhhhhhhhhhhhurrrrrrgh. I literally entered the rocket at the exact same frame as I did in the current submission.
EDIT: To elaborate, a lot of that was because of bad enemy spawns, which are another RNG-based frustration, except this time I don't even know when the roll occurs for selecting the spawn location. The fastest route from the middle fuel spawn to the rocket has you go directly through a location where recently-spawned enemies will also be passing through, with absolutely no clearance to either side of them and forcing you to piss away time by waiting a hovering. The longer routes I take in the current submission actually allow me to largely avoid this issue, because I have MUCH more flexibility with where I choose to drop in to the lower part of the stage and being able to actually shoot enemies.
So I've taken a quick look at the game and I think this is the relevant code, starting from $6614 :
ld a,($5dcc)
cpl
and $0f
ret nz
ldir
call $65db
ld ($5d39),a
ret
This bit of code runs only in the period between one block of fuel being added and the next spawning. $5dcc appears to be the game timer, so it looks like fuel can only spawn every 16 ticks. It then calls a function at $65db and stores the value in $5d39; it looks like $5d39 is the fuel x-position. So let's look at the code at $65db:
ld hl,$65e9
ld a,($5dce)
and $0f
ld c,a
ld b,$00
add hl,bc
ld a,(hl)
ret
This is indexing into the array at $65e9, which contains the following values:
8, 32, 40, 48, 56, 64, 88, 96, 120, 128, 136, 192, 224, 8, 120, 128
These look like the possible drop positions. They seem to be placed so they don't land on the very edge of a platform, or anywhere near the rocket. 136 appears to be the optimal position, but I don't think you'd lose any time on 128 because there's a 16-tick rule in place anyway.
The index is determined by the last four bits of $5dce, which is the random number we're looking for. This number is also rolled by enemies spawning and generally existing so it's not going to be very controllable, but the smoke effect when you first launch into the air also rolls this, as well as explosions caused by killing enemies, and shooting in general without hitting anything.
If you set up a watch on these places (for BizHawk and +2A you'll want BANK 5 and subtract 0x4000 from all of these - e.g. the fuel x-position is 0x1D39) it should help things go a bit faster. I'm not great at making efficient scripts but this should be enough to go on for someone who wants to take on that task.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
Dimon12321 wrote:
So, what happens after you touch the bed? Does Willy go all the way to the left on his own, and teleports from Bathroom to Central Cavern?
That's right. The movie file ends with the final jump towards the bed, everything else is done by the game itself. Reaching Central Cavern is considered the end of the game.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
nymx wrote:
Also, for ending input early...I do see that manipulating the last fuel object, to drop directly over the rocket should be performed. This should save a good chunk of frames. Additionally, if RNG is controlled better...you might get more than one fuel object to drop directly over the rocket without having to do any extra inputs. This would be even more impressive.
If I remember rightly, fuel can't fall directly onto the rocket, it has to be touched and dropped by Jetman to count; in addition, you do have to enter the rocket at the end of every level, and since the last level's enemies home in on you, you can't just stand outside the rocket and wait.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
Nice run! It looks relatively optimised to me, I think some scripting is required to get the most out of the RNG given both fuel and enemy spawns utilise it heavily.
A quick note for judges on goal choice: strictly speaking, levels 9-16 are identical in gameplay to 1-8 (minus RNG), and there's an argument that the run should stop there; however, the rockets themselves are labelled U1 for levels 1-4, U2 for levels 5-8, U3 for levels 9-12, and U4 for levels 13-16 before looping back to U1. I'd argue that the game itself considers this to be one loop and therefore 16 is the correct goal choice.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
Technoturnovers wrote:
There's even a [romhack|https://highriser.itch.io/jetpacrx] for the tape version called Jetpac RX that enhances the gameplay and graphics, although it might be challenging to TAS since running it relies on tape manipulation features that BizHawk might not have.
It's not tape manipulation features; the game is presented as a pair of .pok files (instructions for an emulator on which memory locations to change). To run this in BizHawk (and real, unmodified hardware) you'd have to create a tape (or .tap file) that loads the patched game, and the text files provide full instructions on how to do that.
So you'd just use that .tap file in BizHawk like normal.
Experienced Forum User, Published Author, Active player
(291)
Joined: 12/20/2022
Posts: 32
Location: UK
Patashu wrote:
Manic Miner
Jet Set Willy
Dizzy
Treasure Island Dizzy
Done.
As for my personal wishlist, there are lots of games I'd like to see TASed and it's certainly beyond my abilities to do them all, even if I had infinite time.
Miner Willy series:
Manic Miner (1 loop, Max Score)
Jet Set Willy (any%, WRITETYPER)
Jet Set Willy II (any%, 100%)
Jet Set Willy II+e.22 (any%, 100%)
Jet Set Willy Nightmare Edition
Manic Panic
Jet Set Willy Altered Reality
Manic Jet Set Willy
Dizzy series:
Dizzy: The Ultimate Cartoon Adventure (any%, Max Score)
Treasure Island Dizzy (any%, Max Score)
Fantasy World Dizzy
Magicland Dizzy
Spellbound Dizzy
Dizzy: Prince of the Yolkfolk
Crystal Kingdom Dizzy (any%, All Perfects)
Crystal Kingdom Dizzy Remake
Wonderful Dizzy
Fast Food
Kwik Snax (any%, All Perfects)
Panic Dizzy
Dizzy Down the Rapids
Other classic Spectrum games I'd like to see/do:
The Trap Door (any%, Max Score)
Jetpac
Lunar Jetman
Destructo
Chaos: The Battle of Wizards
Lords of Chaos
Spindizzy
Ant Attack
Skool Daze
Back to Skool
Equinox
Highway Encounter
Alien Highway
Cauldron
Cauldron II: The Pumpkin Strikes Back
Quazatron
Strike Force Cobra
The Way of the Exploding Fist
Firelord
Uridium/Uridium Plus
Other modern Spectrum games I'd like to see/do:
Jumpin' Jupiter (+ Prelude)
Last Train to Tranz Central (Part 1, Part 2)
Just a Gal
Bonnie and Clyde
Cocoa and the Time Machine
Cocoa 2: Twenty Four Hour Parsley People (Episode 1, Episode 2 any%, Episode 2 100%, Episode 3)
Valley of Rains
DRIFT! (any%, Max Score)
DRIFT! 2K22 (any%, Max Score)
The Dark Redux (+ Lost Pages)
Travel Through Time (Story, EVA)
Lava
The Swarm is Coming
Angels
Lava Birds
Metal Man Remixed
Rubinho Cucaracha
Hakkenkast
Don Quioxte
🔺🟢🟦
I'd also like to see Tau Ceti and The Sentinel, but anything that can be considered a completion for these are very long (although I guess shorter than Desert Bus Max Score).