Skip to content
furniture for programmatic typesetting & animation in DrawBot
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
examples
furniture
.gitignore
LICENSE
README.md
setup.py

README.md

furniture

Layout, animation, and typesetting utilities for drawing and animating in drawBot.

Basically just code I reuse every time I make an animation (or anything) with drawBot.

Tutorial

At the moment this library has extremely limited functionality, and the best documentation can be found in this article, which is a tutorial on using the callback-style animation pattern that furniture.animation provides for.

Installation

pip install furniture

(Or pip3 install furniture if your pip does not point to pip3.)

This library is meant to be used in conjuction with DrawBot. It can be used within DrawBot itself (in the integrated editor), but to get the most out of furniture, you’ll also need a module installation of DrawBot.

There are instructions here in the DrawBot github repo on how to do a module installation, but here’s a slightly modified version that has worked on my machine.

> git clone https://github.com/typemytype/drawbot
> cd drawbot
> pip3 install PyObjC
> pip3 install fontTools
> python3 setup.py install

You can verify a module installation of DrawBot and furniture by checking them in a Python REPL:

> python3
>>> import drawBot
>>> import furniture

Why does this exist?

furniture.animations

Doing animations in drawBot is awesome, but it also requires a lot of boilerplate and — when you make a long animation — the render process can be slow and memory-intensive (since the rendered frames are kept in memory in preparation for the video compilation at the end). So instead of rendering all your frames within a single drawBot context, using furniture.animation you can set up the animation in such a way that it can be rendered frame-by-frame from the command-line. This is much faster than rendering all your frames within the app, and also means you can render frames in parallel on multiple cores of your machine. (Well, only sort-of at the moment, it requires running multiple Python processes.)

Caveat If you know of a better/alternative library for this, please let me know!

Why is it called furniture?

Because it helps lay-out type, kind of like furniture.

furniture.geometry

I really love slicing & dicing rectangles with the style of code that furniture.geometry provides. (More on that below, but really it's just some functions for dividing/insetting/offsetting simple rectangles that can be used directly with drawBot primitives, because the Rectangle class implements iterable access. Incidentally, I've since found that this code is quite similar to the arrayTools module in fontTools.misc, which you can see here, and can be used on Rectangle objects: https://github.com/fonttools/fonttools/blob/master/Lib/fontTools/misc/arrayTools.py

Features

  • furniture.geometry provides a simple Rect structure for slicing & dicing rectangles quickly and easily (loosely based on the use of CGGeometry in AppKit programming)
  • furniture.animation provides a simple Animation object for parameterizing animations via a single frame-wise callback that operates in a stateless fashion (meaning any frame of your drawing can be rendered at any time). That is, you build an Animation object by giving it a draw function, which in your code would look like def draw(frame): and within that function you get the context frame object (an AnimationFrame) that has properties like frame.i (index of the current frame), as well as frame.doneness (a 0-1 float that gives the "doneness" of the animation as a function of its length, which is an argument provided to the original Animation constructor) — as below:
  • furniture.vfont provides a single function at the moment, scale_to_axis for scaling a 0-1 value to an axis as provided by DrawBot’s listFontVariations function.
from furniture.animation import Animation
from drawBot import *

def draw(frame):
    fill(random(), random(), random())
    rect(*frame.page.take(frame.doneness, "minx"))

animation = Animation(draw, length=100, burn=True)
animation.storyboard(frames=[0, 1, 50])

The burn=True there just adds a little seconds / frame index / render date box in the lower right-hand corner of the video, for easier debugging if you need to nudge things around once you’ve viewed them in After Effects.

If you run that code in DrawBot itself, you'll see the frames specified in .storyboard, i.e. frames 0, 1, and 50. If you save this code in a file called "example.py" and run the code from a standard (command-line) Python process, i.e. python -c 'import example; example.animation.render();', this will render pdfs of every one of your frames into a folder called frames.

Viewing animation output

Though the written frames can be ffmpeg'd into a video, I've found that possibly the best way to get a quick and easy preview of your rendered work is to grab a copy of Adobe After Effects (free if you have a CC subscription), then start a project and import (⌘i) the first image in your rendered frames folder (i.e. 0000.pdf) into your project, making sure to select "Image Sequence" from the cryptic "Options" option in the import dialog. Once you've imported this "image sequence," you can drag it to to the timeline area and it will create a sequence for you with all the correct settings. Then you can create a composition from that sequence, and, when you’ve rendered new frames, you can purge the After Effects memory (via Edit > Purge > All Memory) and — voila! — you’ve got a previewable/steppable animation.

Why render PDF and not PNG? I've noticed some artifacting in variable fonts when cutting png images directly from DrawBot with certain fonts, but the same artifacts are not present in PDFs, and remain invisible even when After Effects renders PDFs down to mp4s via the Adobe Media Encoder pipeline.

Caveat! It's easy to get the frame rate for the imported image sequence incorrect, since the default frame rate for all imported sequences is set in Premiere's Preferences -> Media -> Indeterminate Media Timebase and After Effects’ Preferences -> Import -> Sequence Footage -> frames per second. Since I'm often combining images and video shot at 23.976, I keep my "indeterminate media timebase" at 23.976, though if you're doing video-free animations, you can use a saner fps, like 24 or 30, or something slower for a funkier feel.

Documentation

At the moment this library has extremely limited functionality, and the best documentation can be found in this article, which is a tutorial on using the callback-style animation pattern that furniture.animation provides for.

You can’t perform that action at this time.