Posts for TiKevin83


1 2
13 14
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Check the paste for updated LUA scripts for both GBHawk and Gambatte, you should probably use Gambatte right now because it appears to be a tad bit more accurate.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Also apologize for the off topic, but I think I'm getting some crashes in Gambatte right now related to save states which I think might be being caused by my oversight of porting this commit from gambatte-speedrun with Gifvex's updates. https://github.com/Dabomstew/gambatte-speedrun/commit/1b5c520a55e0319915625138e26989aadcea3f48 And there are a few more accuracy related commits since then that don't affect Yellow. https://github.com/Dabomstew/gambatte-speedrun/commits/master
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Gambatte continued syncing all the way to the 1 hour mark at the first Koga Juggler, so whatever it's doing for mode 3 timing is accurate in my estimation. I haven't written any further yet :)
Post subject: New Desync
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
I continued using GBHawk for finishing this project out of curiosity and ran into another desync at around the 52:45 mark. A Horn Drill missed Eevee where it hits on GBHawk. I'm investigating whether it could be due GBI and the verification pipeline but am not quite so good at narrowing down emulation bugs. I'll probably want to test Gambatte for it as well. http://tasvideos.org/userfiles/info/47098121238689520
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Yeah I think in general it will make sense to do long TASes on Gambatte unless something new is found that it can't emulate. I should also mention I'm getting a lot of audio problems with GBHawk right now, it seems to speed up and slow down a lot which in turn makes the pitch sway wildly.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
GBHawk is now extremely accurate, I used the new build today to begin rewriting the Yellow TAS past a previous desync and it worked, new record longest syncing GBC TAS at about 46 minutes. Here's some documentation of how that works. https://pastebin.com/DXiDT9ZT I do have 2 things to mention now that sync is so accurate: GBHawk appears to emulate the visual effect of an LCD disable differently from other emulators and console. For example during Pokemon map transitions the LCD is disabled and console and emulators show a white screen during this time frame, where GBHawk appears black. This got me a bit freaked out about another desync but I was able to write past it so it appears to just be a visual difference. Second, it's a lot slower than Gambatte when the fps throttle is turned off, which makes writing these new hyper-accurate TASes a lot longer, but if the slowness is related to the improved accuracy we might have to live with that.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
I believe it may need to be tweaked slightly to play back on BizHawk 2.2.2 since it's the only version with the necessary accuracy patches.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Thanks again for everyone's input and thank you GoddessMaria for your encouragement! Thunderaxe, there are a few vectors I can think of to improve on this TAS. One would be to do 3 Pidgeotto fights in the forest instead of fighting the 2nd bug catcher, but this requires some extremely precise RNG and I'm not even sure if it ends up being faster. Another would be to use a proper bot to find more efficient RNG setups (I found everything in this movie by hand). Lastly, there might be a way to handle bag management slightly better. If you skip getting the early Ether it makes several menus more efficient and then you don't have to choose to use it on Horn Drill. You can instead get the Max Elixer on cycling road after biking down the far left side. This might end up being worse though because of the increased time to scroll down when you need to use it.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Chamale, The RTA community recently re-allowed instant text for Red glitchless, if I were to TAS Red under the same ruleset that I did for Yellow I would use instant text. The glitchless term is a good label for RTA runs using this ruleset, but I could understand why it would make less sense to call the branch glitchless here.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Isn't Brock through Walls an Arbitrary/Undefined Code Execution setup? Since the walk path for the player is not defined for the case where you are to the NPC's right.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Hi Feos, Thanks for your jnput. I think I'm following what you're saying, we could drop "glitchless" to avoid confusion. My primary purpose using the term here was it's familiarity to the RTA community. Zowayix, You could go down a rabbit trail with that as well because you have to infer the intended use of memory addresses from what the code actually uses them for, but yes I agree that those intentions are more clearly defined than story progression. The use of the word "glitchless" is certainly strange from the TAS community's perspective because RTA abuses so many glitches that don't result in ACE or memory corruption, so I can definitely see a need to be careful with terminology here.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Hi Chamale, Thanks! Yes, there may be a way to optimize the options and tackle animations by a few frames, and you are spot on that it might not mean anything because the RNG for getting Nido's stats with very little delay is so precise in this run.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Hi Fortran! It would be a bit different. The entire intro through Brock is different due to having Squirtle but you'd likely still swap to Nidoran, You'd be able to stay as Nidorino for Misty because of already having learned Poison Sting, it would be more complicated to choose when to take damage for red bar, you wouldn't be worried about manipulating color lag, and a lot of the late game enemies have different AI and different level Pokemon which could change move uses.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
I disagree with the assumption of difficulty in defining glitch. The "glitchless" category used in the RTA community is based on a very straightforward definition of "no arbitrary code execution." On the other hand, trying to follow the intended sequence of the game would also necessitate beating Erika before Koga before Sabrina before Blaine, which has never been suggested as a requirement for glitchless Pokemon runs before by anyone, classic or no classic. I do understand the desire to not publish competing classic and non-classic runs that have very similar category definitions if they both existed, but for the moment there is no classic category TAS.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Hi Lapogne, Double Kick shakes are 55 frames faster than the flash animation from horn attack and tackle. This drops to 10 frames faster when double Kick is not very effective, but being in red bar saves 16 extra frames only on the shake animation. Combined with 7 frames lost to switching moves, it's about 19 frames faster to use double kick instead of Horn Attack there or 26 compared to switching to Tackle. These timings are not perfect but they are accurate enough for decision making. More on that here if you're interested: https://pastebin.com/VPURUUak
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
2 more things, Here is a link to the old JPN cart TAS I mention in my submission text: https://www.youtube.com/watch?v=l01wqxsPHB4 Also, another aspect of the red bar timesave is that it saves time on some of the move animations, especially when they are super effective or not very effective. Gifvex has compiled a list of timings of move animations with animations set to off, off+red bar, on, and on+red bar. This led us to realize that having animations on can be faster in places as well, which is not relevant anywhere in Yellow but can be relevant in Red/Blue TASing. In my opinion, the nuance of these frame optimizations is well worth listening to the sound. Gifvex's timings: https://pastebin.com/cTAUc5aS Corresponding list of moves: https://glitchcity.info/biglist.htm
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Hi Thunderaxe, Thanks for your input! For your first point I will refer to my Legend entry on Glitchless. The use of Poke Doll on Marowak in glitchless Pokemon speedrunning has been uncontroversial for years in the RTA community, and we only recently split glitchless and glitchless (classic) as a means of integrating the SRC leaderboard with the JPN cart community's leaderboard. The easiest way of explaining the category difference is that "glitchless" allows S+Q (not relevant for TAS) and allows exploiting oversights that cannot lead to ACE (like Poke Doll on Marowak) whereas "glitchless classic" disallows an arbitrary set of non-ACE-causing glitches to experience more of the storyline. In my opinion, the glitchless category is actually less arbitrary because it is much easier to draw a line as to what causes ACE vs what is an essential part of the story. For example, why does classic not require you to go catch Mewtwo after the Elite 4? It's clearly a major story element as seen in Pokemon Origins. In any event, this is a TAS of glitchless, not glitchless classic, so poke doll is not used here. I might TAS classic some time, but it requires rewriting from about Ether on. Not getting red bar would lose probably ~4 minutes between lvl up jingles, learning Thrash in red bar, enemy death cry skips, and text box jingle skips. If I refrained from taking advantage of it in the interest of entertainment, it would be trivial to use it to improve on this TAS. Also, the vast majority of comments from YouTube, Discord, Twitter, etc. have found the frequent crafting to survive at 1HP very entertaining. About Earphone3: in early January I discovered a trick called "fast options" where pressing left or right on the same frame as A to enter the options menu, both at the new game screen and in-game, causes the left or right press to apply to every option as they load without causing any timeloss. Conveniently, The left press sets every option we need for the run, though it extraneously sets Earphone3. Because the sound settings do not change any frame timings later in the run, it's safe to leave it on instead of losing frames taking time to scroll down and set it back to mono. Also, Earphone3 down mixed to Mono in post production sounds the same as the game's Mono output, so I just downmixed to Mono for entertainment value for YouTube.
TiKevin83
He/Him
Ambassador, Experienced Forum User, Moderator, Published Author, Site Developer, Player (120)
Joined: 3/17/2018
Posts: 348
Location: Holland, MI
Thanks Spikestuff, I don't know of any others who have done Pokemon TASes lately, but there are other ideas, especially for a proper Yellow NSC TAS!
1 2
13 14