Linocut

Pictures written in code, for games on cauldron_2d. What tuning_fork is to sound, this is to sprites: notation in Elixir, pure render functions, a disk cache, and a way to look at the result.

Installation

def deps do
  [
    {:linocut, "~> 0.2"}
  ]
end

ship =
  Linocut.sprite("""
  ..RR..
  .RWWR.
  RRWWRR
  """, R: "#c33", W: :white)

headings = Linocut.headings([{15, 0}, {-8, 8}, {-8, -8}], 64, size: {16, 16}, fill: "#e8e8e8")
red_team = Linocut.recolour(ship, %{{204, 51, 51, 255} => {51, 51, 204, 255}})
spin = Linocut.Strip.new(headings, 30)


`Linocut.sprite/3`	a picture with a palette legend, `.` and space transparent
`Linocut.outline/2`	a filled or stroked shape from its points, `x` right and `y` up
`Linocut.headings/3`	the shape at every heading, `0` pointing right, turning counter-clockwise
`Linocut.recolour/2`, `tint/3`, `scale/2`	derivations on any raster
`Linocut.Derive`	a picture from another by a line on its keys: `from knight; colour C #c02020; flip h; over hat 0 -2`
`Linocut.Pic`	the `.pic` file: a palette, a blank line, the picture; frames separated by `---`
`Linocut.Strip`	frames at a rate, looping
`Linocut.Cache`	rasters rendered once, kept under `~/.cache/linocut`
`Linocut.Preview`	a sheet of many rasters, terminal output, PNG
`Linocut.Quantise`	a picture reduced to a small palette, chosen by median cut over its opaque pixels
`Linocut.Sheet`	a sprite sheet cut into frames, found as bands of rows and runs of columns between blank ones
`Linocut.Frames`	frames of one animation lined up on one canvas, standing on the same floor, and written as a `.pic` with a shared palette
`Linocut.Pose`	a figure's joints, and the OpenPose skeleton a ControlNet reads them as
`Linocut.Resample`	a picture redrawn at another size, by area or nearest pixel
`Linocut.Seam`	a picture made to repeat without a visible join, its far edge faded into the near one
`mix linocut.import`	turns an image file into a `.pic` at a chosen size and palette
`mix linocut.sheet`	cuts a sprite sheet into the frames of one `.pic`
`mix linocut.join`	joins separately drawn pictures into the frames of one `.pic`
`mix linocut.pose`	writes OpenPose skeletons for an animation cycle, one PNG a frame

Every result is a FrenchCurve.Raster, which Cauldron2D.Atlas.put/4 takes as it is.

Deriving

A .pic is keys, and its keys mean something: C is the cloak whatever colour it is. So a second view is a line, not a second drawing:

from knight; colour C #c02020            the same knight in red
from knight; flip h                       facing the other way
from knight; over hat 0 -2                with a hat on
from walk; frame 3                        one frame of an animation

Linocut.Derive.derive/2 runs such a line against a function that gives each name's text; Cauldron2D.Atlas.load/3 reads the same lines from the fifth field of atlas.txt and builds the derived arts after the drawn ones. The operations work on the text of the picture — reversing a row, transposing the grid, rewriting a palette line — so nothing is rendered until the atlas is built, and a derived picture is a picture like any other.

Looking at it

mix linocut.view ship.exs
mix linocut.view ship.exs --scale 3 --columns 8

mix linocut.view takes a script whose last expression is a raster, a list of them, or a strip, and shows it in this terminal as pixels or braille. The example scripts are in the repository.

License

MIT

Next Page → Changelog