Pokémon Trading Card Game

(The horrendous tutorial is now slightly more than a third of the run; please skip it still. Go to 7:26 in the video)
What's happening, forum? It's only been a month since I finished my previous Pokemon TCG TAS, but... on the surface, it's 10 seconds slower than the old TAS. Now, that's just because of emulator inaccuracy, and doesn't matter whatsoever, but... It really bugged me, and the only way to improve my time was to start over. So I did!
The main reason for the improvement is that, with help from Aroymart and Mountebank regarding accuracy, I created a Lua script that simulates RNG/deck shuffling in order to easily quickly test every possible frame for what you need the opponent to have. This was especially a godsend for leader duels, and for testing different paths of duel order.
With this script in hand, and the experience I gained from the last TAS, I was able to cut out 3 deck edits, and fixed the routing mistakes and subpar Amy/Rick/Mitch luck, to end up saving 19 seconds.
You run the script once the decks are loaded, but before they're shuffled (so during the shuffle animation). It gives a lot of results you probably won't want to use, because it includes frames that could work if you reshuffle your hand (which is generally a pain to edit around). There's probably some false positives with the player reshuffles as well; I'd rather have to sort through unworkable frames, than have it leave out good ones!
For 4-prize duels: https://pastebin.com/LS681932
For 6-prize (opening hand cheating) duels: https://pastebin.com/AsxQzADp

Game objectives

  • Does aim for fastest time, even as a secondary goal
  • Heavy luck manipulation
  • Still get put in Vault because it's impossible to follow what's going on, but hey, at least I pause the game three less times now
  • Emulator used: BizHawk 1.11.4 (syncs on BizHawk 2.0)

Comparison

19 seconds saved is pretty significant, considering a large percentage of this game is non-input!
All of the duels after the Lightning Club are different hands and have different wait times, so the input is completely new past there. Overall, I think my wait times for duels were lower, aside from Murray in the Psychic Club.
SegmentPrevious TASNew TASMain reason
Intro2759127591 (+0)Literally no change
Lightning61096238 (+129)Longer edit #1
Grass 154435440 (-3)No change
Water68296468 (-361)-1 edit, better Amy
Ronald 113331300 (-33)Better coin flip
Science39393748 (-191)-1 edit, better Rick
Fire41143860 (-254)-1 edit
Rock39303757 (-173)-1 trip
Ronald 213081339 (+31)Worse coin flip
Psychic26322683 (+51)No change (slow frame)
Fighting27602379 (-381)-1 edit, better Mitch
Grass 224522727 (+275)+1 edit (swapped with Fighting)
Dome92199075 (-144)Faster frames/menuing
Total: -1101 frames

Deck edits

When I started the last TAS, I was expecting to do edits every few duels, so the first 2 edits only being used for 2 duels each didn't seem so bad. I generally took the first good frame I found for a duel, because they took a long time to find. This time, with Lua script in hand, being more open about club order, avoiding the few routing mistakes, and being very experienced on how to TAS this game, I was ready to maximize the re-use of PlusPowers and other cards and I hoped to go from 6 edits to just 4, particularly improving the early game, and hopefully save 10 seconds (deck edits take roughly 5 seconds each).
For full edit info, check this Pastebin: https://pastebin.com/xMdJzQTL

Deck Edit #0 - Jennifer, Nicholas

With the Lua script, I was quickly able to check all permutations of these first three Lightning Club duels, trying to find an order where I could quickly beat all 3 without an edit, and hopefully get a second PlusPower before the first edit. There was none. So, all 4 Lightning Club duels use the same frames as before.

Deck Edit #1 - Brandon, Isaac, Heather, Kristin, Brittany, Michael

To go from 6 edits to 4, I knew I needed to maximize how many duels each edit is for. There's 28 duels total, so each edit needs to be used for 7 on average. With this edit, I was at 8 duels done. Most of them are easier duels (leaders are generally harder), but it was still good pace.
I only have 1 PlusPower in the deck at this point (since I couldn't find better booster packs in edit #0), which makes it much harder to stretch the edit out for a long time. Luckily, since the Grass Club duels can be done in any order, I had a lot of options, and was able to find this order that works, reusing the same PlusPower to beat Brandon's Eevee, Brittany's Tangela and Michael's Mankey, plus a couple uses of the one Double Colorless Energy.

Deck Edit #2 - Sara, Amanda, Joshua, Amy, Ronald 1, Joseph, Rick, Jessica, Ken, Chris, Gene, Ronald 2, Murray, Mitch

Well, never mind 4 edits! Being able to so quickly check so many frames made it possible to make this edit work for practically the entire game! 14 duels in one edit... wow!
In my quest for minimizing deck edits, I decide to use Farfetch'd against Amy. Amy would take all 4 PlusPowers to beat with a 20 damage Pokemon, while using Farfetch'd only takes 2. Locking up all 4 PPs to one duel seemed like a much worse idea than taking the coin flip.
After the Water Club, I found a nice 2 Porygon frame for Rick, saving 3 PP uses over the previous TAS which settled for 1 Porygon. The Joseph fight also uses the same two PlusPowers I used against Amy; at this point, I had done 7 duels and only used 3 PlusPower slots.
I carried on through the Fire, Rock, Psychic and Fighting Club, checking as many permutations as I could to find frames that would work without an edit. Oddly, I use Machop and 2 PlusPowers against Gene instead of a Paras and DCE like I normally would; it's just what was available. The Murray took a little over a second to find, but he's the hardest leader to find a viable frame for. I use a Professor Oak and Item Finder against Mitch, but it was a necessity to go for 3 edits, and it's my only usage of them.

Deck Edit #3 - Nikki, Dome

The Dome is hardest, because you have no more variance in duel order to get different RNG, and you have few options in each duel, especially against Jack (who needs his Articuno, and a Ditto). My deck is also much weaker than it was in the previous TAS, since I've only done 2 edits instead of 5; most of the first half of the deck is useless filler, while with more edits, you'd have gotten rid of most of the filler by now.
It took a few different attempts, using different Mitch/Nikki frames, to find one where I can do a short edit to complete it. I spent hours trying to make other Domes work, but just didn't have the Machops/Staryus/Energy able to fall into place where I needed them, without removing 40 cards or something.
In this one, I only remove 13 cards, and the last 11 I add are all Trainers, which is fantastic. You want to minimize switching between card types as much as possible. In particular, adding Staryus and Machops is very slow, as Water/Fighting types are very far away from Trainer/Energy.

Possible improvements

Honestly... none? I took care of all the improvements from last time (better club routing, better Amy/Rick/Mitch frames, no additional enemy actions (Energy Removals, Bills, etc), more booster manipulation, one less usage of Computer Search and Item Finder each), and three edits is definitely the minimum number of deck edits without losing time, with Lightning+Grass being the only beginning that can make it happen. Plus, since I tried different club orders with my script, I can say with a relative amount of certainty that this is the best possible order to do the game.
The most I can think of is going first against Ronald 2 (~30 frames lost), slow Murray wait time (~30-60 lost), still two Farfetch'd usages (~45 each, but he's so useful) and one Item Finder usage (~60 lost). All of that added up is less than 5 seconds, and was all part of getting down to 3 deck edits; none of it is mistakes.

Conclusion

VBA is dead!
May as well use the same funny thumbnail (45973 this time):
Note for encoder (probably Spikestuff): Please don't use the Gambatte palette; it uses very weird colors (esp. for reds/oranges) and generally looks incorrect compared to console. (Vivid or VBA Accurate is AOK)

Masterjun: Judging.
Masterjun: Accepted to Vault as an improvement to the previous run.
Spikestuff: Publishing.

TASVideoAgent
They/Them
Moderator
Joined: 8/3/2004
Posts: 14776
Location: 127.0.0.1
This topic is for the purpose of discussing #5597: anonymous user's GBC Pokémon: Trading Card Game in 21:30.96
Post subject: Please don't turn out like Clock Tower.
Spikestuff
They/Them
Editor, Expert player, Publisher (2254)
Joined: 10/12/2011
Posts: 6324
Location: The land down under.
You mother-- claiming it again.
WebNations/Sabih wrote:
+fsvgm777 never censoring anything.
Disables Comments and Ratings for the YouTube account. These colours are pretty neato, and also these.
Post subject: Re: Please don't turn out like Clock Tower.
Blazephlozard
He/Him
Banned User
Joined: 2/27/2013
Posts: 175
Location: Ohio
Spikestuff wrote:
You mother-- claiming it again.
pls no pink charizard
Active player (250)
Joined: 12/13/2016
Posts: 352
Good improvement, still Meh vote on entertainment.
Judge, Skilled player (1275)
Joined: 9/12/2016
Posts: 1645
Location: Italy
Why does Ronald's overworld sprite have a completely different hairstyle? Yes vote
my personal page - my YouTube channel - my GitHub - my Discord: thunderaxe31 <Masterjun> if you look at the "NES" in a weird angle, it actually clearly says "GBA"
Editor, Skilled player (1936)
Joined: 6/15/2005
Posts: 3239
Oh nice. You managed to use just 3 deck edits! It would indeed be difficult to improve on this. Better luck manipulation maybe, but as is typical of TASes, the amount of effort required per second gained tends to grow exponentially with the revision number of the TAS. What's your next TAS project? Is it Pokemon Card GB2?
Blazephlozard
He/Him
Banned User
Joined: 2/27/2013
Posts: 175
Location: Ohio
FractalFusion wrote:
Oh nice. You managed to use just 3 deck edits! It would indeed be difficult to improve on this. Better luck manipulation maybe, but as is typical of TASes, the amount of effort required per second gained tends to grow exponentially with the revision number of the TAS. What's your next TAS project? Is it Pokemon Card GB2?
nah, I don't think I'd be happy trying to do that project at the optimization level I'd want it at, there's so many more variables and deck requirements. I was thinking an updated Dynamite Headdy TAS. I don't think there's much to improve over emu's excellent work, but there are some new tricks/skips used in real-time runs
Player (42)
Joined: 12/27/2008
Posts: 873
Location: Germany
Nice work again! I am amazed that you managed to finish it so quickly. Yes vote for suggesting the vivid palette :P
Post subject: Re: #5597: Blazephlozard's GBC Pokémon Trading Card Game in 21:30.96
Editor, Emulator Coder, Site Developer
Joined: 5/11/2011
Posts: 1108
Location: Murka
TASVideoAgent wrote:
Note for encoder (probably Spikestuff): Please use the Vivid color palette. I think the TASvideos encode used the default "Gambatte" palette, and I don't know exactly what it's supposed to be, because Charizard isn't pink, he's orange
A Gameboy player is not a Gameboy.
Post subject: Re: #5597: Blazephlozard's GBC Pokémon Trading Card Game in 21:30.96
Blazephlozard
He/Him
Banned User
Joined: 2/27/2013
Posts: 175
Location: Ohio
natt wrote:
TASVideoAgent wrote:
Note for encoder (probably Spikestuff): Please use the Vivid color palette. I think the TASvideos encode used the default "Gambatte" palette, and I don't know exactly what it's supposed to be, because Charizard isn't pink, he's orange
A Gameboy player is not a Gameboy.
He is also orange on my Game Boy Color tho. The entire game has an incorrect look with the Gambatte palette, that was just the best example I could find. The Vivid palette looks pretty much identical to what the game looks like on a GBC or GBA, and I just tested both. The Gambatte palette does not look accurate to any version of the game I've ever seen.
Player (42)
Joined: 12/27/2008
Posts: 873
Location: Germany
In my humble opinion, the policy is completely misguided with respect to the pallettes. The fact is, people do not watch TASes on a real gameboy, they watch them on their monitors, which uses a color encoding which is completely different from the gameboy screen. It is just common sense that if you take your beautiful wallpaper and print it, the end result will not be as vivid as it is when you look at it on the computer screen, that's just how different color spaces work. The idea that we should avoid applying corrections just because it is not the most immediate conversion to the image on a platform where the movie is not being watched makes no sense.
Judge, Skilled player (1275)
Joined: 9/12/2016
Posts: 1645
Location: Italy
Encoder Guidelines wrote:
As set out in our introduction page and parts of the Movie Rules, our runs are intended to appear as though they could be played on the original hardware (if not actually capable of being played back on the original hardware). Thus, one core guideline for our encodes is to appear, as closely as possible, as though the run was played on the original hardware.
my personal page - my YouTube channel - my GitHub - my Discord: thunderaxe31 <Masterjun> if you look at the "NES" in a weird angle, it actually clearly says "GBA"
Dwedit
He/Him
Joined: 3/24/2006
Posts: 692
Location: Chicago
The GBA SP v2 and Game Boy Player will output the true RGB colors that the game specifies, as will the Super Game Boy. What exactly are the palette choices? There's probably one that washes out the colors and makes white gray to try to simulate the original Game Boy Color, then there's maybe a brighter version of that, then there's probably also a True RGB color mode.
Post subject: Re: #5597: Blazephlozard's GBC Pokémon Trading Card Game in 21:30.96
Editor, Emulator Coder, Site Developer
Joined: 5/11/2011
Posts: 1108
Location: Murka
It's a tricky question, as the GBC's LCD is not backlit, and so color output depends on viewing environment. Using the Gambatte preset will give a color output that, when viewed on a properly adjusted computer monitor, will appear to the human eye as reasonably close to how real GBC hardware appears in average lighting conditions. If there's a slightly more accurate palette that the community would like to see used, I'll gladly add it to Bizhawk, but the idea that a real GBC screen could output a gamut anywhere near sRGB (what the Vivid preset gives) is patent nonsense.
Post subject: Re: #5597: Blazephlozard's GBC Pokémon Trading Card Game in 21:30.96
Blazephlozard
He/Him
Banned User
Joined: 2/27/2013
Posts: 175
Location: Ohio
natt wrote:
It's a tricky question, as the GBC's LCD is not backlit, and so color output depends on viewing environment. Using the Gambatte preset will give a color output that, when viewed on a properly adjusted computer monitor, will appear to the human eye as reasonably close to how real GBC hardware appears in average lighting conditions. If there's a slightly more accurate palette that the community would like to see used, I'll gladly add it to Bizhawk, but the idea that a real GBC screen could output a gamut anywhere near sRGB (what the Vivid preset gives) is patent nonsense.
Well, I'm playing the game right now, on real GBC hardware, in average lighting conditions, and it looks nothing like the Gambatte palette, and does look like the Vivid palette but with less backlight. Game Boy Player looks like Vivid, with adjustable brightness/contrast based on your TV settings. The "VBA Accurate" palette is a bit better, especially with reds/oranges, but it is still overexaggerating how "washed out" a GBC screen is. All versions of the game should be outputting the same color data; from there, the only thing that should be adjustable is brightness/contrast. How could orange turn into a color closer to purple/pink? I'm really not trying to cause issues here, I just don't understand how you can say "Gambatte is how the game looks on a GBC" when it simply is not. You will never find a screenshot of the game, or even a camera picture of a screen (which I'll try to get one of), where the game's colors look closer to Gambatte than Vivid.
Dwedit
He/Him
Joined: 3/24/2006
Posts: 692
Location: Chicago
As I just mentioned, you can stick a GBC game cartridge into a GBA SP (second version with backlit screen), and it will output sRGB video. As for what an actual GBC screen looked like, you either had the right angle to see the screen reflecting some yellowish indoor lighting, or you had the Worm Light shining a bluish white LED on the screen.
Blazephlozard
He/Him
Banned User
Joined: 2/27/2013
Posts: 175
Location: Ohio
Alright, I don't have a great camera but I took a photo of my Game Boy screen. I wanted this to be as fact-based as possible. http://i.imgur.com/Prd7n9a.png The palettes are: Gambatte, a photo, Vivid, VBA Accurate (Old), VBA Accurate, GBA "VBA Accurate (Old)" is, uh... Maybe if you dunked your Game Boy into a bathtub it'd end up looking like that? And "GBA" looks like Vivid with a terrible threshold filter over it. So forget them. Of the other three, comparing to this photo, VBA Accurate seems to be the winner, looking at the Charizard's shade of orange, the contrast between the blue outline and yellow text of the Pokemon logo, and the tan background (compared to Vivid's yellowish-tan). Under ideal lighting conditions, I do think a GBC screen's color shades look closest to Vivid's (in addition to Vivid being the most accurate to GBP and 3DS VC, which are certainly official hardware, though not original). And I don't agree with the concept of purposely using ugly colors to try to look more "accurate" (shouldn't GB TASes be encoded with a yellow-green filter then?). But I couldn't capture a photo of ideal lighting conditions, and VBA Accurate doesn't have the major color-shifting issues of the Gambatte palette, so if Vivid's some kind of issue to you guys, I'm completely fine with VBA Accurate being used. Watching the game in VBA Accurate, I think "this looks kinda dark" instead of "this looks wrong".
Patashu
He/Him
Joined: 10/2/2005
Posts: 3999
I think for this to be accurate you would have to ALSO take pictures of your monitor of each emulator colour palette, so that it's not screenshots vs picture but picture vs picture.
My Chiptune music, made in Famitracker: http://soundcloud.com/patashu My twitch. I stream mostly shmups & rhythm games http://twitch.tv/patashu My youtube, again shmups and rhythm games and misc stuff: http://youtube.com/user/patashu
Judge, Skilled player (1275)
Joined: 9/12/2016
Posts: 1645
Location: Italy
Blazephlozard wrote:
http://i.imgur.com/Prd7n9a.png The palettes are: Gambatte, a photo, Vivid, VBA Accurate (Old), VBA Accurate, GBA
Woah I really like VBA Accurate, it feels like an actual GBC to me... So nostalgic.
my personal page - my YouTube channel - my GitHub - my Discord: thunderaxe31 <Masterjun> if you look at the "NES" in a weird angle, it actually clearly says "GBA"
Blazephlozard
He/Him
Banned User
Joined: 2/27/2013
Posts: 175
Location: Ohio
Patashu wrote:
I think for this to be accurate you would have to ALSO take pictures of your monitor of each emulator colour palette, so that it's not screenshots vs picture but picture vs picture.
that's way too meta I'm really really sorry if I was rude at all about this, I feel scared like "omg the encoders hate me now" or something haha, I didn't think it'd be a big deal. I think it's safe to say Gambatte palette uses provably inaccurate colors, that's all I knew. I'm kinda liking VBA Accurate too the more I look at it, like now Vivid's looking too bright in comparison instead of Accurate being too dark, haha. I just knew Gambatte was wrong, I hadn't given VBA Accurate a fair shake since I've always used vivid
Editor, Emulator Coder, Site Developer
Joined: 5/11/2011
Posts: 1108
Location: Murka
Blazephlozard wrote:
(a picture)
Vivid has way more saturated yellows and blues than your photograph. I agree that the red hue seems off. From other images, I've seen that vivid greens are way different on vivid vs real hardware.
Blazephlozard wrote:
(shouldn't GB TASes be encoded with a yellow-green filter then?)
Doh hoh hoh... That's actually a good question. I don't know.
Post subject: Movie published
TASVideoAgent
They/Them
Moderator
Joined: 8/3/2004
Posts: 14776
Location: 127.0.0.1
This movie has been published. The posts before this message apply to the submission, and posts after this message apply to the published movie. ---- [3485] GBC Pokémon: Trading Card Game by anonymous user in 21:30.96
Site Admin, Skilled player (1234)
Joined: 4/17/2010
Posts: 11251
Location: RU
FWIW, vba accurate old's color in the center looks the closest to the photo, but I haven't seen the actual gbc screen IRL, so I'd believe that accurate old is wrong. Vivid is just way too saturated, as is gba. Gambatte is just about some arbitrary colors, especially in comparison to vba accurate. So the latter looks best to me.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.