Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
I suspect the correct category for an ABC run that had to be viewed in realtime is 70 star. That'd allow you to skip the very slow stars and stars that required an A press. (It'd still need one A press due to BitFS, but that isn't technically a star :-P)
Probably the best run would be a minimum-A-press 120 star run, but fastforwarding through the boring parts. This is pretty much what pannenkoek's playlists on YouTube do already, though.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
VCutM early might possibly save time in 120 star even if the key skip doesn't. You can get everything in VCutM in one trip, can't you? Leaving the level afterwards would be slower, but I think it might still save time on castle movement overall.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
This seems like the sort of thing that would make for a good YouTube video or the like, rather than a submitted TAS. Especially if the game has no RPG mechanics, it seems like something that can be put up IL style as a "look what you can do" type of thing.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
I guess we're going to have to obsolete that run then :-P
(Actually the whole TASVideos Pro joke did fairly badly at guessing what sort of runs would be possible in the future. I don't think it anticipated the SMB3 run that wins from the title screen, for example.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
IMO take all the time you need on the submission text. The submission texts are a major part of the entertainment I derive from TASes, because knowing what's going on helps you appreciate the encode much more.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
If you could carry the jiggies and notes over, then you'd probably have to stop using FFM, as it wouldn't make sense to use the glitch for a suboptimal use and the optimal one would be boring.
Luckily, there's no known way to do that.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
This run leaves me feeling a bit confused. We see at the start of the game that there's basically a glitch that lets you teleport anywhere as long as you're moving fast enough. Then it isn't used past BitFS. I assume that there are actually large restrictions on what you can do with an overflow jump, and that just isn't visible in the actual game.
Nonetheless, I still enjoyed it. Not as much as the 70 star run, but it's interesting to see what sort of tricks let you cut down the time as much as possible.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Kles wrote:
To say they "failed horribly" is definitely selling them short. RNG calculation was too computationally expensive back then (both due to low processor speeds and the fact that the Mersenne Twister was not developed until 1997), and for the vast majority of players, a cycling system is more than sufficient.
An LFSR is very computationally cheap even on older consoles. (An LCG can be even cheaper, on a console with hardware support for multiplication.) There are other RNG designs than the Mersenne Twister.
Meanwhile, there are Final Fantasy games which have random tables for which some of the results literally can't happen, as there's no seed that produces them. Additionally, the probabilities of various low-probability events are a long way away from where the developers intended them because the RNG is biased.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
There used to be an emulation problem in which the target that tests to see if you have Fire Eggs in order to allow access to the next part of the overworld didn't work.
I suspect it's been fixed by now, but it explains why there aren't any older TASes of the game.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
In many older Final Fantasy games, random events are on a fixed cycle of 256 possible results (basically, the devs tried to write an RNG and failed horribly). This gives you huge control over pretty much everything that happens in the game once you're aware of it, especially because many of the games have an easy way to throw away unwanted results to fast-forwards to the useful ones.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
I think the main difference here is that what gets copied in the DKR run is a major part of the game's logic (whether or not the endgame is open), whereas what gets copied in the BK run is comparatively minor (what moves the character's have access to). This doesn't change the category itself, but it does change how entertaining the resulting run is. In DKR, it feels pretty much as though the runner is cheating in order to complete the bulk of the game. In BK, the same trick doesn't achieve nearly as much, and the bulk of the game still has to be played, just with a few more movement options than before.
I used to be a teacher, and this situation is a bit like marking someone's programming assignments. Students are allowed to use code from elsewhere, as long as they give credit for it. If they copy the whole code from somewhere else, they get no marks because they didn't do anything on their own. If they copy code from elsewhere but still do most of the work, they can get very high or even full marks (if the external code does a job that they weren't being assessed on).
In DKR, what you're basically doing is recompleting a completed game, because so much of the effort required to complete the game is being copied over. Meanwhile, the glitched newgame+ in BK feels a lot more like an actual newgame+ category (preserving the capabilities of the character but resetting everything else is the most common newgame+ definition in games that add them intentionally), plays like one, and most of the gameplay of the game is still intact.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
To elaborate on what Samsara is saying for people who missed the previous incident: basically, you'll save everyone (including yourselves) a lot of trouble if you make a run that starts from a blank cartridge and sets up the save files appropriately for FFM on the same emulator you plan to use for the main movie (recording as you go), and then save the recording and resulting SRAM somewhere safe, starting the submitted movie as a save-anchored movie from the SRAM in question. There's no need for this preliminary recording (the verification recording) to be optimized in any way; it could just be someone playing realtime. Its purpose is to make sure that the save file you're starting from is something that's achievable via normal gameplay, and not having it tends to lead to a huge row.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
My opinion on FFM is that it classifies a game as "glitched newgame+", rather than being in the more common "newgame" category that starts from a blank cartridge. However, obtaining newgame+ status via a glitch is fairly interesting to look at, and if it makes for a more entertaining run in other ways, it may make for a better category than newgame any%.
One thing to be wary of is that the time spent loading the second save file after a reset and setting up the glitch would probably count against your run time total, at least the way that TASes are normally timed; this is a much shorter time period than that needed to create the save file in the first place, but it's still going to be quite relevant. If the glitch still saves time despite this (and because newgame+ also lets you skip the intro cutscene, it almost certainly does), it's definitely worth adding it into the run even though it requires a category change. If this setup costs more time than the trick saves in the first place, the glitch is going to be quite hard to defend from a categorisation point of view.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
OK, now I've watched the run.
I wasn't as entertained as I thought I would be. For one thing, although the level design in the hack seems very good for casual play, I dislike the music. There's not much a runner can do about that, but it started the run off on the wrong note.
My main criticism is that the run looks more like a realtime playthrough than a TAS, not so much in terms of optimization, but in terms of entertainment. There are plenty of time periods where nothing much is happening and the runner misses opportunities to make things more interesting through risky play; the strategies in this run are in general very safe. A realtime speedrun typically uses the safest of several possible strategies if they're close in time; a tool-assisted speedrun should use the most dangerous if the times are identical. This is particularly noticeable in boss fights; during periods of time where the boss can't be hit, the runner finds a safespot and stands there, rather than actively dodging enemy projectiles. (3-Castle is a good example of this.)
Despite looking good for casual play, the level design is often kind-of awkward for speedruns, too. The autoscrollers get tiring after a while, and several levels early on have sections which on a speedrun are just running without much variety or obvious difficulty.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
With respect to the choice of version, it strikes me that using the latest available patch is often a good way to define "glitchless". How do you know whether something was intended or not? Well if the developers fixed it in a later patch, it probably wasn't intended.
(Of course, this relies on there being no glitches that make their way through all the patches without the developers fixing them, despite the fact that they want to.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
OmnipotentEntity wrote:
The run just finished. RTA time is 2:46:38. TAS time isn't released yet.
Is there any update on the progress for commentary for this?
I've been waiting for the commentated version, but want to know whether I should keep waiting or whether I should just watch the uncommentated encode.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
This is a well-known category in NetHack (sufficiently so that I had to write a hack of it once for a tournament that made the kill count explicitly tracked by the game), called "extinctionist". People don't speedrun it, though, because it requires way too much grinding to make all the enemies appear (most methods of generating enemies can only generate finitely many so the game eventually runs out, leading to a natural end point for the category, but it takes ages).
The NetHack community being what it is, people have even done pacifist extinctionist (where you arrange for every monster in the game to die indirectly but don't kill any yourself).
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
I was working on something like this a while ago, but at a different level of abstraction (hooking at the system call level, rather than the library call level). My version would likely be more accurate, but yours is likely easier to write.
There's some functionality that you mark as missing that I may have solutions to already. For example, one of the things I was working on was placing the game within an isolated environment so that things like save files were easily detected, and things like process IDs (which some games use as input into their random number generators) would be the same on every run. (I don't think it would translate exactly, though.)
The way you're trying to go about savestating could possibly work in theory, but is necessarily fragile (many games use simpler allocators) and is also very complex. There's a simpler approach using the program's memory map tables (savestating at the level of mmap/sbrk, rather than malloc). Some savestating programs, such as CRIU, already exist on Linux. You might want to get into contact with dwangoAC, who was looking at CRIU and related programs in the context of TASing. (He's likely busy with GDQ stuff at the moment, but may be able to give you some pointers as to how to get started with that.)
The main problems I was running into were to do with threading (I wanted to handle it deterministically in all circumstances). I'd just about reached the point of deciding that a kernel patch would be helpful for that when I got distracted by other things and stopped.
There are some fundamental differences between our models for running the program, e.g. I assume you're allowing the program to connect to the screen directly, whereas I was using xvfb (and hadn't reached the point of simulating clicks yet). However, I doubt there's anything insurmountable there.
I'll be idling in #tasemu on Freenode over the next few days if you want to talk to me about anything.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
The voting question is about one specific aspect of the run.
As such, if you want to comment on something that isn't covered by the voting question, you have to do so in the forum thread. (I semi-frequently do this, normally arguing that a movie should be published even though I voted No.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
IIRC the TGM people were considering giving this GDQ a miss, and they pretty much need to be there for TGM to work.
If they are, though, my suggestion is to make a TGM TAS that goes for something 100%-style (I'm not sure if you can get all achievements in one run but if you can you should), and race that against the humans playing any%. I have a suspicion that TASbot will still win, but it'll make a hypothetical race more interesting.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
SDA has different rules for yelling "Time!" than TASvideos does. People wouldn't get the joke there unless it was explained in advance, which would ruin it. I think you'd have to physically disconnect the controller for the joke to work, but that has problems of its own (spurious button presses and the like).
It's an idea that I'd like to see work, but I don't see a way that it does.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Although I think the console virus idea is a good one, I don't think we'd be likely to properly do it justice without more planning. As such, saving it for a future GDQ may be for the best. For similar reasons, messing around on the NES Mini is probably a bad idea (as we won't know what's even possible until the console is released).
The N64 shenanigans sound like a good thing to make the centrepiece of the block, assuming you can find something sufficiently crazy that will be a hit with the crowd. (I know there are people who want to see an ACE on a 3D console.)
I'd recommend that most or all of the games you play are well-known ones; a common criticism of TAS blocks is that people often can't easily distinguish humanly possible play from TAS insanity unless the games are well known or the results (ACE, etc.) are so insane that the game doesn't matter. Mario Kart 64 is a good example to this end, because the game is widely known, and even if you don't know it it's clear that something ridiculous is happening. I'd recommend looking for a category with good variety (in particular, you don't want to be using the same glitch in multiple courses).
Scribblenauts is an excellent idea as far as crowd involvement goes. One issue is that it's not really showing off anything that can be done non-TAS, and it's also not really a speedrun, so it might be a poor fit for the event. (Besides, the correct solution is always "handcuff the Starite to a vending machine" :-P)
The other ideas are vague enough that I can't really comment, other than the general advice above.
(I also note that there's been a suggestion that it might be worth skipping a GDQ if there isn't enough material. One of the problems with TASbot at GDQ is that it tends to set the bar for itself very high, and blow minds to the extent that people can't comprehend anything better. Then it isn't, and they get disappointed.)
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Well, someone watching me play, say, Pokémon HeartGold just by watching the screen wouldn't see much of what's going on, because I'm moving around in RNG space in addition to moving on the map. (I play on hardware, but have an effective memory watch set up via doing tests every now and then to determine the current RNG value.) Having a display of my current RNG goal and the current RNG seed would make it a lot easier to figure out what's going on.
Or as a simpler example, there are FPS runs where runners spend all the time looking at the floor to minimize lag, even if there are no glitches exploited. The encode makes it very unclear what's happening there. The "correct" way to view the run would likely be with a camhack.
Editor, Experienced Forum User, Published Author, Player
(44)
Joined: 7/11/2010
Posts: 1022
Is a speedrun equivalent to an encode of a speedrun? Or are, say, the values that appear in memory part of the speedrun too?
I think you can easily make the argument that with these game-end-glitch runs, much of the reason they're boring is that the encode is focusing on what appears onscreen and through the speakers, but those aren't what the run is actually focusing on, and thus don't give a good view of what's going on.