#simon tatham

[rauko]

Newton Fractals of Complex Polynomial Functions ❁

𝒵𝓃+1 = 𝒵𝓃 – 𝑝(𝒵𝓃) / 𝑝'(𝒵𝓃)

𝑝(𝒵) -> Fig. 1–5: 𝒵³–1 | 𝒵⁴–1 | 𝒵⁵–1 | 𝒵⁶–1 | 𝒵¹⁰–1

[mitzvahmelting]

Thank you so much for recommending that puzzle collection! I've been obsessed with Net and Slant recently, so I'm really glad I can play them offline now ^v^

Hell yeah!!

For those who aren't in the know, this anon is talking about Simon Tatham's Portable Puzzle Collection, which is a collection of open source puzzle games that have been ported to every platform ever.

Do you miss the times when you could sit at your computer and open a lightweight, utilitarian puzzle game without any ads or splashy effects? Do you miss the little puzzle games that came packaged with your 2000s flip phone or mp3 player? This puzzle collection fills that niche perfectly. You can find a port of it that works offline and ad-free on Android or iOS, and you can also play it in your browser at the link above.

Personally, my addiction is the game Signpost. By addiction, I mean that I've probably played upwards of 100 hours of it in the last few years. I haven't really touched any of the other games in the collection because I love Signpost so much. Highly recommend.

[andmaybegayer]

Simon Tatham added the new monotile to the Loopy game and did a write-up on dynamically placing aperiodic tilings. I have never even tried to beat the Penrose Loopy variant, I stopped at hexes.

[taperwolf]

Maybe a majority of my gaming time is spent playing little puzzle games written by Simon Tatham. (That site has the collection for unix, windows, and web, although I most often use a port for Android titled "Simon Tatham's Puzzles" in Google Play.)

One of the games is internally titled "Pattern", which implements a puzzle variously called elsewhere "Picross", "Nonograms", "Hanjie puzzles", and a dozen other names, and usually involves filling in a grid with black and white squares, based on numerical hints as to the length of black lines in each column or row, to reveal a hidden picture. Originally a puzzle in print, there are two basic ways that folks have implemented this for computers; the first involves including in your program a big list of puzzles and stopping when the user has solved all of them.

Tatham used the second solution, which is to create arbitrary solutions for each new game. You now don't run out of puzzles, or have to worry about storing them, but you sacrifice the part about revealing a (n intentionally) hidden picture.

Now, I say "arbitrary" and not "random" for a reason. The revealed pattern of black blobs has a level of structure to it — it's not a random field of static — but neither does it resolve to a recognizable source image. You can play pareidolia with it all day if you want, but saying "Oh, this is Gonzo the Great as a centipede centaur" isn't authoritative.

Which is why I'm wondering about using some kind of web search to find low-resolution images, dither and/or threshold them, and use that to produce the puzzles. You'd likely get a lot of boring images — one of the things I remember from running JWZ's webcollage was how much of the imagery on the web was just text — and there's little to no chance you could use the knowledge that it's supposed to be a picture to help you solve the puzzle, but I just wonder if it'd be in any way different to reveal the "original" at the end.