About the game:
Oekaki Logic (J) (NP) is essentially a Mario Picross clone (of which there actually seem to be quite many, like Oekaki Logic 2, Ochan no Oekaki Logic, and Mario's Super Picross with all of them for the SNES).
There seem to be 2 different game modes (represented by a polar bear on the left and a brown bear on the right in a menu screen) among which one can choose. The polar bear game mode seems to be equal to ''5 misses (incorrect tile being coloured in) end an image drawing attempt, but this means one at least gets feedback on which tiles are incorrect. The brown bear seems to correspond to a harder mode in which no feedback is provided, but alongside, one can colour in incorrect tiles as often as one wants (but only once the correct set of tiles is coloured in, an image is completed).
There's images of different sizes to deal with: 5x5, 10x10, 15x15 and 20x20. The images that need to be drawn are fixed (as opposed to randomized), and pictures of the image solutions exist and can be found at the following links. Oekaki Logic image solutions:
Easy level solutions:
http://www.neoseeker.com/oekaki-logic/faqs/180410-easy.html
Normal level solutions:
http://www.neoseeker.com/oekaki-logic/faqs/180416-normal.html
Hard level solutions:
http://www.neoseeker.com/oekaki-logic/faqs/180413-hard.html
Expert level solutions:
http://www.neoseeker.com/oekaki-logic/faqs/180411-expert-1.html
http://www.neoseeker.com/oekaki-logic/faqs/180412-expert-2.html
Insane level solutions:
http://www.neoseeker.com/oekaki-logic/faqs/180414-insane-1.html
http://www.neoseeker.com/oekaki-logic/faqs/180415-insane-2.html
Here's a zip file that contains the images if anyone else feels intrigued and wants to try out some own manual routing of further images (or maybe it could even be feasible to solve the smaller images with a cleverly designed brute search that takes out equivalent alternate options and moves for which one can prove that they cannot lead to or be part of an optimal route):
https://cdn.discordapp.com/attachments/218520158963105792/534426355329269771/Oekaki_Logic_image_screenshots.zip
- - -
About TASing the game:
The main part of the optimization lies in a sort of Travelling Salesman Problem for routing the best way of drawing the images, which actually isn't quite as easy due to an additional restriction for movement over the tiles. Namely one always starts first at the top left corner of an image, and now regarding the mechanic to move the cursor around from 1 square to another, it does not always hold that the cursor will move exactly in the (combined) direction (which can be horizontal, vertical or diagonal) that one is holding at a given frame, but there's a slight difference that makes it interesting: For any image drawing gameplay frame (that is not a lag frame of which there's rarely one if at all to my knowledge so far), it holds that the (combined) direction that the cursor will move to is the sum of exactly those directional inputs (among Up, Down, Left, Right) that exist/are held in the current frame but were not pressed in the previous frame. As example,
frame 0: (no input)
frame 1: >
frame 2: v+>
frame 3: ^+<
frame 4: <
would move the cursor right in frame 1, straight down in frame 2, diagonally up-left in frame 3, and nowhere in frame 4 (except that one cannot move past the solid image boundaries). This directional stun (when a direction is held over multiple frames) is due to the game waiting a bit until it will start moving in the held direction automatically (slowly, from a TASing perspective), if a direction is held long enough. This means that diagonal movement is ''dangerous/potentially risky'' since it covers 2 directions at the same time to start charging towards and be stunned for the next frame. This also means that one preferably wants to go in ''roundabout paths'' (mostly consisting of horseshoe-turns , stairs, and Z-turns), since going straight lines is penalized with stun times. Colouring or not colouring a square that one is moving over in a path is just determined by pressing A on that frame or not, so for finding the fastest paths one can take the correct colouring of tiles that need to be coloured in as given. The drawing with an A press happens after executing the directional input if both happens at the same frame.
Upper bounds for the number of steps (or frames) needed for images of different sizes (by colouring an entire image space preferably with as few steps as possible):
5x5: At most 29 frames/steps (with the green dot referring to 1 waiting frame, so that one has free movement in any direction afterwards)
10x10: At most 108 frames/steps
15x15: At most 239 = 225 + 14 frames/steps
20x20: At most 400+18 frames/steps
Here's some routing attempts that I did for some of the images:
https://cdn.discordapp.com/attachments/218520158963105792/534234385189306384/Oekaki_Logic_route_approximations_Easy_3to16Expert14Normal1to39.zip
And to get an idea of what the routing can look like (for larger images than just 5x5), two examples from the zip (of which some might be improvable) as visualization:
Easy stage, level 7 in 85 steps:
Expert stage, level 14 in 211 steps:
In some zigzag scenarios, one can decide to either wait one frame and continue in the same direction and draw in another tile later on the path, or to draw it in earlier without having to wait, but to then move over the same filled out tile again later.
I wonder how many TASers might feel like taking some of the images and trying to route them. Maybe with collaboration, we could cover all images with excellent routes for a TAS.
