5 ms·

Reverse typesetting: reflowing page layouts where you don't have knowledge of the typesetting structure, i.e. a scanned physical book or PDF paper. Naive rules-

Reverse typesetting: reflowing page layouts where you don't have knowledge of the typesetting structure, i.e. a scanned physical book or PDF paper. Naive rules-based heuristics based on the dimensions of bounding boxes and gaps. Point is to reflow things for resizing to eink readers. (Specifically the size that fits in my pocket which I carry around. User #1 is me). Building in Common Lisp and targeting an Emacs mode for interactive execution with manual feedback.

Foreignborn 2y ago
i don’t quite understand, what makes it reverse typesetting? my understanding is your typesetting books for responsive eink readers.
- perihelions 2y ago
  You're inferring the structure of the document from the printed result. If typesetting takes a set of layout directives and outputs a page, this is taking a finished page and guessing what layout directives could create it. Then you can take that inferred structure and reflow the page in a new layout.
  - froh 2y ago
    so like ocr but not recognizing characters and words but recognizing the layouted structure and transforming it into content markup and layout markup?
    - perihelions 2y ago
      That's a way to view it! The reason I'm not falling back on OCR is because the general case is full of things, like math equations and inset graphics/diagrams, that can't be OCR'd. The only robust way to deal with those is to treat them as graphical atoms: "this bounding box can be moved around, but should not be split up into pieces".
perihelions 2y ago
Err, here's a visual explanation of what I mean by this, from my REPL: https://ibb.co/album/MDw79y?sort=name_asc https://ibb.co/album/MDw79y?sort=name_asc (The source example is from David Tong's physics lectures notes, that were featured on HN last week — https://news.ycombinator.com/item?id=43763223 https://news.ycombinator.com/item?id=43763223 )

Foreignborn 2y ago

i don’t quite understand, what makes it reverse typesetting? my understanding is your typesetting books for responsive eink readers.

perihelions 2y ago
You're inferring the structure of the document from the printed result. If typesetting takes a set of layout directives and outputs a page, this is taking a finished page and guessing what layout directives could create it. Then you can take that inferred structure and reflow the page in a new layout.
- froh 2y ago
  so like ocr but not recognizing characters and words but recognizing the layouted structure and transforming it into content markup and layout markup?
  - perihelions 2y ago
    That's a way to view it! The reason I'm not falling back on OCR is because the general case is full of things, like math equations and inset graphics/diagrams, that can't be OCR'd. The only robust way to deal with those is to treat them as graphical atoms: "this bounding box can be moved around, but should not be split up into pieces".

perihelions 2y ago

Err, here's a visual explanation of what I mean by this, from my REPL: https://ibb.co/album/MDw79y?sort=name_asc https://ibb.co/album/MDw79y?sort=name_asc (The source example is from David Tong's physics lectures notes, that were featured on HN last week — https://news.ycombinator.com/item?id=43763223 https://news.ycombinator.com/item?id=43763223 )