An astrolabe, from one page to seven posts

On 8 October a new page went up on yLookbook, my design lab. It draws a planispheric astrolabe: an orange spark follows a TikZ listing and draws the instrument stroke by stroke. When it's done you can drag the pierced rete on top, and the sky turns with it and time moves on. There are five plates, Oxford, Paris, Toledo, Baghdad and Tokyo, each cut for its own latitude. A column on the left keeps reading out what the instrument shows: sidereal time, the Sun's height, the part of the ecliptic rising in the east.

The next day the page grew a film. Nothing was redrawn. The film reads the Oxford plate from the same listing, all sixty lines unchanged. The turning you do by hand on the page is handed to time in the film, which runs from just after midnight on 5 October to sunset on 10 October.

On 10 October I signed the film off and sixteen versions came out of one export: the full film horizontal and vertical, a vertical highlight and a set of nine stills, each in Chinese and in English, plus covers in four ratios and two languages. The same day it went out seven times on six platforms: once on Bilibili, a video post and an image post on Xiaohongshu, once on Douyin, the full film and a Short on YouTube, and once on X. In my ledger all of that is one work, number 05 in the Drafting Sheets series.

In my notes this chain has four layers.

At the bottom is the source, something that can be recorded, captured or explained. The astrolabe page is a source; it lives in yLookbook and is registered there. Above it is the work, one unit of publishing: the film and its seven posts count as one work. Works made the same way add up to a series. Above the series is the engine, a machine that can turn out film after film, computing the picture, the timing, the sound and every version the same way each time.

I set myself two numbers for when one layer becomes the next. Three works of the same kind become a series with its own collection on each platform. More than five, and I consider pulling the method underneath out into an engine. I don't do that lightly.

Drafting Sheets was counted into being exactly like that. At the start of October I made three scale films with the same drawing method: a road bicycle at one tenth, a watch movement ten times enlarged, a camera lens at full size. On 3 October the third one made it a series, and it took the name of the book it already had on yLookbook. By then the method had drawn six or seven things, a logo, a server and a mainboard among them, so the same day the drawing code was split out of yLookbook into a library of its own, yPlotter. Six days later the code that builds miniature cities for the model cabinet went the same way, as yDiorama.

What I want from an engine is mostly certainty. When I know a complicated kind of picture will come out right every time, I can choose subjects for it, write a story around it and plan a long series on it. That's what let the astrolabe go from page to film in a day. How the pen moves, how time is kept, how the picture stays consistent from frame to frame: yPlotter had answered all of that long before, and that day I only had to think about what to draw and where to let the sky stop.

A series' specification lives with the code that produces it. The cover layout for Drafting Sheets, the closing credit and the pattern for Chinese and English titles are written in a file called series.md in yLookbook, next to the drawing code. The AI reads it before every episode. On 6 October I picked one of three cover samples, and every episode since has been made to it. yPulse, which handles distribution, only registers series and works; it doesn't care how they were made.

The numbering on the site and on the platforms is separate too. On yLookbook the astrolabe is volume 16 of Drafting Sheets; in the platform collections it's No.05. Volumes are numbered in the order things were made, collections in the order they actually went out, and only some of what gets made goes out at all.

The plainest benefit of the layers is that when a work runs into trouble, the source is untouched. The San He and San Yuan compass pages were made on the same day, 7 October, and each grew a film. The vertical cut of the San Yuan film went up on Xiaohongshu three times. The first two never entered the feed, so I deleted them and posted again, the third time under a new title, "An ancient compass", and it had 1,556 plays within a day. The posts came and went; the San Yuan page stayed where it was.

And a source can give more than one work. The astrolabe page has five plates. The film used only Oxford; the other four haven't become anything yet.