Submission #9128: PiePusher11's GBC Game & Watch Gallery 3 "50 stars" in 38:20.50

Game & Watch Gallery 3 is a collection of 10 classic Game & Watch games - Egg, Greenhouse, Turtle Bridge, Mario Bros, Donkey Kong Jr, Flagman, Judge, Lion, Spitball Sparky, and Donkey Kong Jr II. The first 5 games also have "Modern" remakes, which retain the spirit of the original games with some slight gameplay differences (and also feature Mario characters)

First, I'd like to talk about the category of this run. This is 50 Stars, which is the requirement to see the credits for the first time. Each game and mode gives one star every 200 points, up to a maximum of 5 at 1000 points. It doesn't matter where the stars come from, as long as you get 50 total.

So, the way I determined which games to play was by doing a low-optimized TAS of every game and mode (basically I just played at half-speed and rewound if things didn't go well). After testing every game and mode, I did a more optimized TAS of the slowest game that "made the cut" and the fastest game that didn't. In my initial cut, Greenhouse Classic Hard was the slowest that made the cut, and Donkey Kong Jr Modern Hard was the fastest that didn't make the cut. Upon further testing, Donkey Kong Jr Modern Hard ended up being much faster than Greenhouse Classic Hard, so I'm glad I tested this further.

While doing research for this TAS, I did come across a player named beco who did an RTA run of this category ( https://www.youtube.com/watch?v=sx0NGBpRMEE ). Obviously I did my own testing for what is fastest, but they did have one interesting routing decision I would've never thought of myself - They got 2 stars from Flagman A and 3 stars from Judge, splitting up that group of 5 stars between two games. Their reasoning makes sense - most games speed up as they go on, but Flagman A takes longer and longer as it goes on, so they want to minimize the time spent there. However, if I went with that route, I'd have to add in another game to make up for those 5 stars lost (with Donkey Kong Jr Modern Easy). After testing both routes, despite Flagman A taking longer and longer as it goes on, it's still faster than adding in another round of Donkey Kong Jr Modern.

As you collect coins, the object Toad brings to Peach gets more and more valuable, so up until you're delivering stars, it's always going to be worth it to get all the coins. It's at this point I'd like to bring up my method for determining whether it's worth going out of your way for points. I first time how long a cycle takes if you don't get any coins, uninterrupted, and get a frames-per-point as a baseline from that. Then I time how long a cycle would take while getting all the coins, and figure out the frames-per-point on that. As it turns out, it's very nearly identical getting the coins versus not getting the coins. However, this is short-lived anyway, as now we need to introduce the Peach cycle. Starting after 500 points on Easy and 300 points on Hard, Peach will back away and come back periodically. As it turns out, this cycle almost exactly corresponds to an uninterrupted back-and-forth that gets both sets of coins! So for the most part, I don't really even have to worry about the Peach cycles. The point routing of this game works out in such a way that on the final cycle, you only need 5 coins, and don't need to return to Mario, so I collect all 5 coins before delivering to Peach, and exit out of the game after delivering the final star to Peach. This point routing works out exactly the same for both Easy and Hard.

In this game, cookies are rolling down four paths to Yoshi. The cookies will go from uncooked, to cooked, to burned as they get lower and lower. (The way the game recognizes uncooked vs cooked vs burned is simply how far Yoshi's tongue has to extend to eat it. If you eat a "cooked" cookie as early as possible, it will actually still appear visually as uncooked) Uncooked cookies are worth 1 point, cooked cookies are worth 2, and burned cookies are worth 0. Additionally, cooked cookies fill up the corresponding meter at the bottom of the screen, and when those are full, a Yoshi Cookie will appear. If you eat a cooked Yoshi Cookie, it will be worth 10 points, and Yoshi will lay an egg, and after a couple more of them, the eggs will hatch. The baby Yoshi will take up one of the paths and eat cookies for you for a brief time. Baby Yoshi eating cookies are worth 5 points each. So, you'd think optimizing this game comes down to simply eating each cookie as early as possible after it gets cooked. That's most of it, but not entirely. Now's a good time to bring up the speed increase present in this game. Throughout each 100-point section, the speed will slowly increase, and it'll actually slow back down when you cross over the 100-point barrier. So you'll notice when I approach each 100-point barrier, I also try to delay eating those cookies until as late as possible, so that the faster speed lasts as long as possible. There are some times where this isn't always the case, though. For example in the 699-700 point transition on Easy, I had a Yoshi Cookie coming that would push me over 700 points. Since Yoshi Cookies reset the playfield, it's still worth it to go as fast as possible in situations like that.

Ah, this was a fun one! In this game, your speed isn't capped, so you can move as fast as you can mash the D-pad. For this game, everything is cycle-based, based on the swinging key. Any points I'm able to get while still making the same cycle is obviously worth it. Doing a similar frame-per-point analysis, I determined that if you skip a cycle, it's worth it as long as you get 7 points or more from it. Here are the point values you can get: -Jumping over an enemy: 1 point -Dropping the fruit on an enemy on the top row: 3 points -Dropping the fruit on a bird: 5 points -Dropping the fruit on an enemy on the bottom row: 9 points. From this, we can determine that it's worth it to drop the fruit on enemies if it either a) hits a top-row enemy and a bird, or b) hits a bottom-row enemy. Like several other of the Classic games, if you play deathless up to 300 points, all point values are doubled until you take a death. This doesn't change the math for what enemies are worth skipping a cycle to kill, but it's worth noting. This doubling of points surely contributed to why this game was fast enough to make the cut for this run.

I picked Hard because more enemies will spawn, and killing enemies with the fruit are worth 5 points each. Doing a frames-per-point analysis on this game, it's worth it to kill an enemy with the fruit if you go out of your way by 110 frames (a little under 2 seconds) or less. If you kill multiple enemies, the window for it being worth it is even larger. For this game, we finally have to talk about RNG manipulation. I have to heavily manipulate the RNG so that I'm able to weave in between the enemies while slowing down as little as possible, and when playing on Hard, the number of "bad" options heavily outweighs the number of "good" options. Additionally, the only way I was able to figure out to manipulate the RNG in this game is by doing things faster or slower, so essentially, every time I want to manipulate RNG, I need to slow down a little bit. This is probably the biggest point for potential future improvement I see in this TAS. Other than that, this game is fairly straightforward, just get to the end of each of the three screens as quickly as possible, while getting as many points as you can from killing enemies with the fruit.

After finishing Donkey Kong Jr, the remaining games need to be unlocked from the Gallery Corner. So, I have to go to the Gift section of the Gallery Corner to get Flagman, Judge, and Lion. I will not be playing Lion in this TAS, as it is incredibly slow, it's only unlocked by happenstance. Judge is another seemingly simple game, but optimizing it is another story. Like I said under Donkey Kong Jr, the only way to manipulate RNG in this game is by wasting frames. Now, I'm manipulating RNG for two different things here. First, and more obviously, I want to have the higher number (or have them equal). But also, there are four or five different speeds that each round can go by. The timeloss from the fastest to the second-fastest is 24 frames, so it's very important to manipulate it to get the fastest speed every time too. Because this is so RNG-heavy, this is another game that could potentially have room for improvement in the future if a better way to manipulate RNG is discovered.

This game is incredibly simple. For Game A, just remember the pattern they give you and repeat it back. For Game B, hit the number they throw up as quickly as possible. In a TAS, there's not much to say here.

And that's it! This was a really fun TAS to work on. A lot went into the routing and optimizing of this TAS, and I hope you enjoy watching it as much as I did making it!

nymx: Claiming for judging.