Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
Docs on Java.Random seem to just indicate it's just some PRNG. I could imagine that it itself is deterministic. However, if the game were to use it with multiple threads (which probably isn't deterministic) or if the initial seeding itself was done in some non-deterministic manner (e.g. it was just say done by querying system time, or whatever other way to get an initial seed). Assuming it's not some threading screwing up results, it's probable you could easily patch away anything non-deterministic with RNG.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
We're currently working on a new input file in the Team 3 discord that'll combine the best strats together (which really is just slapping Team 4's Stage 6 inputs onto Team 2's movie, plus some minor improvements on Stage 6's boss).
The SOB shown in Darkman's chart as a note won't be close to this final movie. It's slightly misleading, as it combines incompatible strategies together to have a faster than possible movie under the contest rules. More specifically, Team 2 uses death to save time on Stage 5, while Team 4 does so on Stage 4. You have a limited number of lives so you can't do both, and Team 2's Stage 4 boss fight strategy does not work if death is used to save time (mainly as it requires the Assault Rifle weapon). Stage 4's boss is also where Team 2 ultimately pulls ahead as once Stage 5 is finished, Team 2 will be 15 frames ahead (until they fall behind on Stage 6's boss, where Team 4 takes the lead).
Darkman425 wrote:
No continues allowed which for this game I doubt would save time anyways
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
I'll clarify some points about my post. It is not to undermine the trust in staff. Staff using other users movies to improve on their own movies to gain an advantage in this contest would be incredibly silly and would be a breach of trust.
However, other contestants should know that staff at the very very least *would* know about this uploading. It is placed in the same section that other staff only updates are posted, so staff would be somewhat expected to check that area. Staff might also end up clicking the links accidentally, or even nearly do so, simply by not being vigilant enough with links from that staff only section. I'd rather have it too that we don't have blind trust in staff, there are numerous places you can upload a movie which will not give staff even the knowledge that you made an upload of your movie. Hidden means hidden from public view, not staff view, and staff don't need to go digging through your profile to know if you uploaded or not.
EDIT: Also a disclaimer, my team submitted at least 4 weeks ago, we're definitely not at all going to cheat here (if not just due to the trust issues) and there's no way for us to have done so.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
Invariel wrote:
CasualPokePlayer wrote:
Memory wrote:
the only thing staff members would actually have access to is all userfiles including those hidden, so you'd need to find a different way to store them.
and at least know that 2 teams have done this without even having to view the site.
Also, just to clarify, I hope that by "have done this" you mean "uploaded files to userfiles" and not "abused staff privilege to look at input files for this contest".
"have done this" refers to uploading to the userfiles, which staff members would know without even need to look at the site (due to that being announced offsite).
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
Memory wrote:
the only thing staff members would actually have access to is all userfiles including those hidden, so you'd need to find a different way to store them.
I'd like to remind all teams of this. 2 teams have so far uploaded them to userfiles. As half the teams have staff members, this means half the teams have access to those movies, and at least know that 2 teams have done this without even having to view the site.
The file is being parsed correctly, it's just the file has garbage length_sec/length_nsec fields (which are required to be set correctly to figure out the time in the case of variable framerate, which this movie uses).
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
Captain_Forehead wrote:
Never mind. I just confirmed the movie I uploaded to the submission is correct. The information that was gathered from it shows as incorrect for some reason. The TAS is almost an hour and a half long, and definitely has more re-records to it. Not sure why it's showing differently.
The movie file can be downloaded, extracted (it's just a zip), and manually examined to verify the site is looking at it correctly.
Doing so will show that the movie file does in fact have a frame count of 44712, and the re-record count in the header is in fact 24279. 44712 frames at 60 FPS is 12:25.20. The site is not parsing the movie it has wrong.
So, it would be a fact that this movie seems to be wrong. Why it's wrong could be a few possible reasons:
1. The server mangled the movie file, somehow (this would be seriously concerning, but this is frankly extremely unlikely).
2. You uploaded the wrong movie and the correct movie is elsewhere on your PC and you just did a mistake when re-checking to see if the movie submitted was the correct movie.
3. You uploaded the correct movie. The correct movie is wrong, and you don't have the correct movie on your PC.
For 3... well assuming you didn't do a mistake then that seems to be the most likely possibility. By "the correct movie is wrong" I mean while it is the movie you have on your PC and what you intended to upload, it's not complete, due to some event making you lose your progress (granted, since you have an encode, this is also just unlikely). I would hope then that you just made a mistake in re-checking that movie :|
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
* Emulator used: BizHawk 2.2.1
There's something odd here, while the submission says the movie was made in 2.2.1, the movie file reports it being made with 2.9.1. Although it would at least seem the movie syncs fine in 2.2.1 and immediately desyncs with 2.9.1?
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
The IGT disparity issue goes down a bit of a rabbit hole here:
https://www.speedrun.com/titol1/runs/y8eq6qdy
This is the SRC WR for same category this TAS competes in. For our intents and purposes, it is a record a TAS must at least match in order to be accepted for publication (and there's no funny exceptions here for records posted after the TAS; this SRC record is 2 years old).
The resulting IGT in this SRC WR is 1:14.375. That is .1 seconds faster than the IGT in the (current) TAS submission, which is 1:14.475. This is a disparity of 4 frames, to be precice.
Why does this occur? Is it an inaccuracy in Ruffle making this TAS lose IGT? Is it something special the SRC WR does?
Well, no on the first part. The TAS is simply suboptimal, on two counts.
For the first 3 stages (which are done glitchless), there is just a case of losing 4 frames, which roblox8192 has demonstrated in their own glitchless TAS (NB: you can cut off the 4 frames used to delete the save and it syncs regardless). Stage 1 loses 1 frame, stage 2 loses 1 frame, and stage 3 loses 2 frames.
However, an oddity shows up here: the SRC WR is in fact 2 frames slower than this submission with the first 3 stages. Subsquently, it would be slower by 6 frames compared to roblox8192's own TAS with the first 3 stages.
This oddity is explained all the way at stage 28. The SRC WR here has the flash window stretched, so the runner is able to click out of bounds here and finish the stage faster with the main glitch used in this run. This can be simulated in libTAS with out of bounds mouse coordinates anyways (which we've published several movies in other flash games abusing this). So while it is something special the SRC WR does, it can be applied to the TAS. This ultimately results in the SRC WR gaining 6 frames compared to this submission, and so is 4 frame faster overall compared to this submission.
Going off of this, we could say that at least there is 10 frames that can be shaved off this submission, and for our purposes there is an existing record which beats this submission, thereby making this submission unacceptable.
EDIT: I mused a little with making forgoing the out of bounds mouse clicking as a separate category, like other games with out of bounds mouse click glitches. This ends up being problematic as using the out of bounds mouse clicking is itself allowed in Standard for Fastest Completion. So this TAS can't go to Standard's Fastest Completion. Standard also has No Major Skips and Glitchless as possible goals. Which are also barred due to the main glitch used to practically skip nearly every stage being used. So such a separate category would end up being in Alternative. However, even then, there's just not much actually different here, out of bounds mouse clicks can hardly be called a major skip in this context (only being applicable in one stage, and that would be in conjunction with the main glitch abused), not to mention that there is only a 6 frame difference (although that's assuming the SRC WR is optimal, it's maybe some more with TASing, but doesn't really change the point). So I would say it would just not qualify as an Alternative goal, and so it would need to be judged under Fastest Completion, which is comparable against the SRC WR I linked.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
The WinXP+PCem combo just wouldn't be viable at all here (and really this applies for most modern-ish PC games):
This is an emulator. An emulator for a really old PC. It's not like libTAS being an "emulator" simulating an extremely powerful machine (with its ability to just lie about time). Rather, you will be dealing with running the game at however fast that really old PC can run it (which for Hollow Knight sounds like it would not be able to maintain full speed at all, so you can imagine that, assuming you were running PCem itself at full speed, you'd still be running the actual game at 5-10 fps or whatever, and that's going to be reflected in an encode.
Hollow Knight, looking at its Steam page, already has a Windows 7 requirement, which cuts off support with the WinXP+PCem combo.
Hollow Knight also, on it's steam page, seems to have a D3D10 requirement. This also cuts off WinXP, which only supported at most D3D9. Even then, the best GPU that PCem supports (3DFX Voodoo 3 3000) only supports up to D3D6.
There is also a likely chance that even the 32 bit version (which apparently exists for HK?) wouldn't work, as it's highly probable it is built assuming SSE/2 support. With more modern compilers (e.g. more modern MSVC / gcc) the developer needs to add additional special flags to actually make a compiler spit out something not making these assumptions.
2 could in theory be worked around by using some modern Linux distro. But then you are swapping out 3's D3D10 requirement with some OpenGL requirement. The best GPU that PCem supports only supports up to OpenGL 1.1. For context, this predates support for shaders (also similar case with D3D6; D3D8 was the first D3D to support shaders). This itself is going to eliminate practically any modern game that uses the GPU as they are going to be using shaders, not the ancient fixed function pipeline ancient GPUs used (which is even possibly outright unavailable for newer GPUs, so even more reason for modern games to use the more modern APIs which that Voodoo card fundementally cannot support).
5 could in theory be worked around by forcing software rendering. Which just makes 1's issue get turned up to an instance level. Keep in mind that software rendering is going to be happening on that emulated CPU. An emulated CPU that only has 1 core (and no hyperthreading magic), and is only running at 450MHz, and predates even the basic SSE extensions that have been present for over 20 years.
Simply be, PCem is very much not a magic bullet once we talk about any modern-ish PC game. Especially not Unity games (along with most modern-ish PC games), which would all use shaders, and so fundementally would not work here.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
You turned on the DIP switch, messed around with things, then turned off the DIP switch. Then you started a new movie. That movie will start the DIP switch off. It will also start off without SaveRAM (i.e. what stores MAME's NVRAM).
This is something that is probably better solved if you could flip DIP switches with TAStudio, but without that, you'd just have to start from SaveRAM after doing your test menu changes. Do File > Movie > Record Movie... then pick the "from SaveRAM" start option, then open TAStudio.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
You have to edit the DIP switches when outside of a movie, then you can create a new movie with those DIP switch edits. Loading a movie will just load that movies sync settings, overwriting whatever sync settings you currently have.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
If they do not work in regular MAME then it will not and never will work in BizHawk's MAME. If you really want to TAS those games, use libTAS + that custom MAME version.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
drkspace wrote:
Personally (completely biased), I think it should be published since it'll be the longest TAS by a wide margin.
You probably missed the Desert Bus maximum score submission undergoing publication, which is over 33 days (so your submission would be a solid 2nd place if published).
Regardless, the issue I point out is due to the goal of this movie. It is not a "Standard" class goal. Standard goals do not have to worry about entertainment in order to be published, if you do a TAS and it is with a standard goal it will be published (assuming the game is publishable in the first place, which official non-AO games will always meet, and the TAS is optimal/beats all records/etc).
If you do a TAS and it does not have a Standard goal, it will be judged under the "Alternative" class. These don't have the goal restrictions that Standard has, but the TAS must be deemed "entertaining" by the audience. The esoteric nature of this submission's goal also hurts it due to this Alternative class guideline:
Your goal choice should be clearly defined and sensible, especially to those who do not know your chosen game. Assume that the audience is seeing the game for the first time.
When the goal is not Standard, and the run is not entertaining by our community, it falls under our "Playground" class. This class doesn't actually entail any publication, in its current form it's just left as a submission which has a flag indicating it's in the Playground.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
KusogeMan wrote:
isn't the cloning glitch useful? from what i recall it's faster to catch pokemon than it is to evolve, so i dont get why you have so many multiples of some of them,you could catch a wingull later... this opmitzation happens a lot in the r2 frlg tas as an example.
the more i watch the less i like it, why didn't u just speedrun the game and cloned rare candies, i'm either clearly missing something or you didn't do any additional research
To be fair, if they allow glitches, they would have to have a constrained allowance of glitches (i.e. only allow the Battle Tower Duplication Glitch) to prevent the pomeg berry from making things even more boring. With the pomeg berry and all the fun that entails you can just outright complete the dex entirely, by just constantly corrupting to get eggs which hatch into the Pokemon you need. Which is even worse with Professor Oak Challenge rules, which would make it so once the 4th gym is complete you have to obtain the entire rest of the dex right then and now and can't complete the 5th gym until you do so. Not to mention glitches which can allow for skipping gyms kind of makes the challenge itself somewhat moot, if the requirement of not completing a gym can be cheesed with glitches to outright skip gyms.
Also, not sure if you just didn't read the description, but they can't just speedrun the game and clone Rare Candies, if not just due to wanting to be glitchless, but that not work as well as you might think due to the "Professor Oak Challenge" rules, which stipulate that you must obtain all the dex entries for Pokemon before you complete the next gym. So the Battle Tower Duplication Glitch can only be effective against Pokemon which are only obtainable after the 8th gym (or is it only Pokemon which are only obtainable after E4/Champion, is E4/Champion considered a "gym" under the Professor Oak Challenge? I don't know, still doesn't change the point).
To be fair again however, not reading the description would be somewhat proof of the esoteric nature of the goal presented here (although you still missed the "Glitchless" part of the goal here). "Professor Oak Challenge" makes no sense for anyone who doesn't know the series, as "Professor Oak" is just the professor in Kanto region, who is never even mentioned in Emerald, so it requires knowledge of the game series itself. Not to mention "Professor Oak" as a character doesn't even have this kind of challenge, it's a fan created challenge. Professor Oak in the Kanto games just gave you the Pokedex and wanted you to complete it, but not with any of the restrictions placed in with by the fan created challenge (the fan created nature of the challenge could be seen as a parallel to a Nuzlocke challenge).
To complicate this even further, this TAS places an additional challenge on top of the Professor Oak challenge. A "Living Dex" stiuplates that you need a member of every single species, evolved, mid-stage evolution, and unevolved, in your party/boxes/breeding house. Also by that, it is somewhat an independent challenge which has to be accomplished alongside the Professor Oak Challenge (i.e. you can very well do a Living Dex without Professor Oak Challenge or do the Professor Oak Challenge without a Living Dex requirement). Although this TAS, I assume, would tie them together as to mean the Professor Oak Living Dex Challenge needs to complete the Living Dex as much as possible before each gym.
Ultimately however, "Living Dex" does make some sense if you know the game a little (Dex is short for Pokedex, the thing that tracks all Pokemon seen/caught, and Living means that you have each Pokemon Living in your party/boxes/breeding house), but still it is a fan created term here and not entirely clear to someone who doesn't have knowledge of what it means exactly (granted, the fact it makes sense with some knowledge means it's not too esoteric here).
Considering too the kind of boringness that this TAS has, that at least has been seen so far in responses, it's probable too that this TAS might not be publishable per se, and would just go to Playground.
Emulator Coder, Experienced Forum User, Judge, Published Author, Experienced player
(609)
Joined: 2/26/2020
Posts: 698
Location: California
Try rebinding the controls here?
I suspect what's actually happening here is one of the modifier keys is "stuck" (Ctrl can do this sometimes iirc?) causing the inputs to actually be something like Ctrl+Up etc. Try pressing Ctrl Alt and Shift too and see if that fixes issues.
