Inside `3b1b/manim`: A Practical Start with Mathematical Animation

3b1b/manim is the animation engine behind many of 3Blue1Brown’s explanatory math videos. Its recent jump of +74 GitHub stars today reflects an enduring developer interest: turning equations, geometry, and abstract ideas into precise visual demonstrations.

The project is Python-first, but it is not a general-purpose video editor. You describe scenes as code, create mathematical objects, and control their transformations with a timeline-based animation API. That makes it especially useful for educators, researchers, and developers who want reproducible visuals instead of manually edited media.

A minimal setup looks like this:

python -m venv .venv
source .venv/bin/activate
pip install manimgl
manimgl example_scenes.py SquareToCircle

A typical scene subclasses Scene, creates objects such as Circle, Square, or Text, and then calls methods like play() and wait(). The important architectural idea is that the scene is declarative at the object level but procedural in time: you define what exists, then explicitly animate how it changes.

For a first experiment, create a file named demo.py:

from manimlib import *

class Demo(Scene):
    def construct(self):
        square = Square()
        circle = Circle()
        self.play(ShowCreation(square))
        self.play(Transform(square, circle))
        self.wait()

Render it with:

manimgl demo.py Demo

Before using it in production, keep these constraints in mind:

• Rendering is compute-heavy. Complex scenes, high resolutions, and repeated previews can consume significant CPU, memory, and disk space.
• API compatibility matters. The original 3b1b/manim codebase differs from Manim Community Edition, so tutorials and imports are not always interchangeable.
• Visual correctness is your responsibility. Code can render successfully while producing confusing mathematical explanations, poor pacing, or unreadable labels.

The strongest feature of manim is its reproducibility: animation becomes versioned Python code. That makes mathematical storytelling testable, reviewable, and easier to iterate than a purely manual workflow.