“Mario’s Super Picross” is a nonogram puzzle-solving game developed by JUPITER and published by Nintendo for the Super Famicom on the 14th of September, 1995. Originally only released in Japan due to the poor international sales of its predecessor “Mario’s Picross” earlier that year, it would eventually be released in PAL regions on the Wii Virtual Console in 2007, and then in North America through Nintendo Classics in 2020.

This game first came to my attention in March last year, where I decided to give it a try on my Switch and quickly got hooked, completing all of the puzzles in the span of about a month. A while after that, my friend vichisuki suggested that I could potentially try TASing the game, and I decided to give it a shot, initially starting work on a 2-player TAS in August 2024. I quickly learnt that optimisation was much more complex than I had initially assumed, causing progress to slow down and eventually stall around January while I focused on other projects.

In April, UndercookedRice submitted a 1-player all puzzles TAS of Super Picross to the site, which reignited my interest in TASing the game. After it was published and I had some more free time on my hands, I noticed that many of the optimisation techniques I had previously used working on the 2-player TAS could be applied to save time. I subsequently started creating the 1-player any% TAS you see now, working on and off over the course of about 6 months before finishing up a few weeks ago.

There are a few quirks of how player input is handled in this game that make optimisation less trivial than it initially seems. Both the A and Y buttons can be used to mark squares, and their input is handled independently from one another. As a result, alternating between A and Y allows the player to chisel up to 60 times a second instead of 30. Consequently, the D-Pad becomes the main bottleneck, as we’re unable to move in the same cardinal direction two frames in a row. Using zigzagging movement patterns wherever possible is crucial for saving time and minimising wasted frames, and diagonal movement becomes more risky due to using two directions at a time.

To make planning my movement between squares easier, I opted to use graph paper for my routing, using a pen to draw the solution and a pencil to draw my planned route. This helped me to identify mistakes and efficiency problems and adjust things accordingly before I even started inputting anything into TAStudio, and it also helped me better visualise ways to tackle some of the more complex puzzles and break them down into smaller sections.

Additionally, the hammers that appear when you mark squares follow animation cycles (8 frames for Player 1 and 7 frames for Player 2), and those need to finish before the game can start the end of puzzle animation. This particular mechanic was first discovered by mohoc when he was giving me feedback on my first 2-player TAS WIP back in October last year. I later discovered that it’s possible to interrupt and restart the hammer animation by using B or X to cross out a square, which is useful for ensuring they finish as soon as possible after a puzzle is completed. In a lot of cases I was able to set things up such that I didn’t have to waste any frames getting the hammers to finish on time, though this wasn’t always possible.

The timing used above uses the selection of “No” to the hint roulette as the starting frame, and the puzzle timer disappearing as the ending frame. On top of the puzzle-by-puzzle improvements, I also reset the game between completing each level, as this saved time over simply watching the congratulatory cutscenes from Mario play out. This comes out to about two and a half minutes saved in the puzzles themselves, and about one and a half minutes saved in menuing.

All of my routing was done by hand, and so it’s possible that some puzzles could be further improved by using a custom solving program or brute-force to find the optimal solutions instead. However, I didn’t have the necessary programming abilities or experience to create such a program myself, and learning how to do so might well have taken me as long as it took to make this TAS, if not longer, so I opted to solve them manually here.

I really enjoyed working on this project, and I’d like to do some more Picross TASing in the future, either in the form of a 1-player all puzzles improvement or a 2-player any% TAS. For now though, I’m probably going to take a break and focus on some other projects for a while.

(frame 96089)

nymx: Claiming for judging.

nymx: I remember when I judged [6548] SNES Mario's Super Picross "1 player, all puzzles" by UndercookedRice in 1:54:43.10. At the time, I didn't even give it a second thought that it could be unoptimed. This version just shocked me! Well, there is no argument here...this is an excellent and unexpected improvement or more than half of the original TAS.

Accepting as "Standard" as "Fastest Completion".

fsvgm777: Processing.