Installation

Algan is available in Python 3.10 through 3.13. Algan can be installed from PyPI with pip, like any other Python package:

pip install algan

This will install a working version of Algan for CPU, however depending on your operating system it may be missing some features such as GPU acceleration, LaTeX type setting, speech synchronization, and extra 3-D asset formats. To make sure you installation is fully functional select the tab for your operating system below and follow the instructions there.

Never used Python before? Start here.

The instructions below assume that you are fimiliar with basic programming terminology, so they are spelled out here:

Python is the programming language you will write your animations in, and Algan is a package you add to it: Python comes first, and the pip command that installs Algan comes with it. Your operating system’s tab below starts with installing Python.

A terminal is where you type the commands on this page. Open one with:

  • Windows – press Start, type PowerShell, press Enter.

  • macOS – press Cmd-Space, type Terminal, press Enter.

  • Linux – your desktop’s terminal application, usually Ctrl-Alt-T.

“Run this command” means: copy the line, paste it into that window, press Enter, and wait for it to finish before running the next one. Every code block on this page has a copy button in its top-right corner.

A virtual environment is a folder that holds the packages for one project, so that two projects can use different versions of the same package without fighting. python -m venv .venv creates one, and activating it points python and pip at that folder instead of at the whole machine. You activate it once per terminal window – if you close the terminal and come back tomorrow, activate it again before running your scripts. Nothing about it is permanent: deleting the project folder deletes the environment with it, leaving Python itself untouched.

Important

If you run into trouble, do not spend the evening on it: ask on our Discord server or open an issue on the GitHub issue tracker.

Tip

Already use uv, Poetry, conda or another environment manager? Nothing here is special: uv add algan, uv pip install algan or poetry add algan work just the same. The instructions below use the venv module that comes with Python, so that they work on a machine with nothing extra installed.

Don’t have Python yet? Install it first.

Download Python 3.13 from python.org/downloads and run the installer. Tick “Add python.exe to PATH” on the first screen.

Check it from a fresh PowerShell window:

py --version

1. Install Algan

Make a folder for your animations, create a virtual environment inside it, and install Algan into that environment:

mkdir alganimations
cd alganimations
py -m venv .venv
.venv\Scripts\activate
pip install algan

Your prompt now starts with (.venv). It will not in a new terminal: run .venv\Scripts\activate again each time you come back to the project.

Note

If activating fails with “running scripts is disabled on this system”, PowerShell’s execution policy is blocking the script. Either allow it for this window only ( Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass ) or use cmd, where the command is .venv\Scripts\activate.bat.

2. Check the installation

algan check

This prints the Algan and PyTorch versions, the device renders will run on, and where output and cache files will go. If it runs, you can make videos.

3. Enable your GPU (recommended)

If algan check says [INFO] Running on CPU even though you have an NVIDIA card, that is expected: the PyTorch published on PyPI for Windows is built without CUDA. Replace it with the CUDA build from PyTorch’s own package index:

pip install --force-reinstall torch --index-url https://download.pytorch.org/whl/cu128

cu128 is CUDA 12.8. If your driver is older, pick the build that matches it on pytorch.org/get-started/locally (choose Stable, Windows, Pip, Python, then your CUDA version) and use the index URL it gives you. Then run algan check again – it should report [OK] CUDA acceleration and your card’s name.

For an AMD card, follow AMD’s PyTorch-on-Windows instructions.

Algan runs on the CPU perfectly well if you have no supported GPU; it is just slower.

Optional: mathematical formulas

Prose works out of the box: Text typesets with your system fonts, through Pango, which is installed with Algan on Windows.

Mathematical formulas are a different matter. Tex and MathTex are typeset by LaTeX, which is a separate installation: get the MiKTeX distribution and let it install packages on the fly. algan check reports whether it found one, and What Algan needs from LaTeX says what Algan actually asks of it.

Optional: speech and 3-D models

  • Speech Generation works out of the box, through the Windows SAPI5 voices. Aligning animations to a pre-recorded audio file additionally needs pip install "algan[audio]".

  • 3-D models in glTF, GLB, OBJ and PLY are supported out of the box. FBX needs pip install "algan[fbx]" plus an assimp DLL on your PATH.

Requirements

Apple Silicon (M1 or newer) running macOS 13 Ventura or later. Intel Macs are not supported: Algan’s kernel compiler publishes arm64 wheels only, so pip install algan cannot resolve on an Intel machine.

Don’t have Python yet? Install it first.

macOS ships a Python that is best left alone. Install your own 3.13 from python.org/downloads, or with Homebrew:

brew install python@3.13

1. Install Algan

Make a folder for your animations, create a virtual environment inside it, and install Algan into that environment:

mkdir alganimations
cd alganimations
python3 -m venv .venv
source .venv/bin/activate
pip install algan

Your prompt now starts with (.venv). It will not in a new terminal: run source .venv/bin/activate again each time you come back to the project.

2. Check the installation

algan check

This prints the Algan and PyTorch versions, the device renders will run on, and where output and cache files will go. It should report [OK] Apple Silicon MPS acceleration available: your GPU is used automatically, through Metal, with nothing to configure.

Optional: mathematical formulas

Prose works out of the box: Text typesets with your system fonts, through Pango, which is installed with Algan on macOS.

Mathematical formulas are a different matter. Tex and MathTex are typeset by LaTeX, which is a separate installation: get the MacTeX distribution and follow the standard installer. It is a large download; the much smaller BasicTeX plus the packages Algan uses also works:

brew install --cask basictex
sudo tlmgr update --self
sudo tlmgr install standalone preview dvisvgm

algan check reports whether it found a LaTeX installation, and What Algan needs from LaTeX says what Algan actually asks of one.

Optional: speech and 3-D models

  • Speech Generation works out of the box, through the macOS speech synthesizer. Aligning animations to a pre-recorded audio file additionally needs pip install "algan[audio]".

  • 3-D models in glTF, GLB, OBJ and PLY are supported out of the box. FBX needs brew install assimp and pip install "algan[fbx]".

Requirements

x86-64 with glibc 2.27 or newer – Ubuntu 18.04+, Debian 10+, Fedora, RHEL 8+, Arch, and anything of similar vintage. ARM (aarch64) machines are not supported: Algan’s kernel compiler publishes no aarch64 wheel.

Don’t have Python (or python3-venv) yet? Install it first.

Most distributions ship a suitable Python 3. Debian and Ubuntu split the venv module into its own package, and the next step needs it, so install that too:

sudo apt install python3 python3-venv     # Debian / Ubuntu
sudo dnf install python3                  # Fedora
sudo pacman -S python                     # Arch

If your distribution’s Python is older than 3.10 or newer than 3.13, get a supported one from deadsnakes, pyenv or uv python install 3.13, and use it in place of python3 below.

1. Install Algan

Make a folder for your animations, create a virtual environment inside it, and install Algan into that environment:

mkdir alganimations
cd alganimations
python3 -m venv .venv
source .venv/bin/activate
pip install algan

Your prompt now starts with (.venv). It will not in a new terminal: run source .venv/bin/activate again each time you come back to the project.

Note

This pulls in around 6 GB, far more than on the other two platforms, because PyPI’s Linux PyTorch bundles the entire CUDA runtime. If you have no NVIDIA GPU and would rather not carry it, install the CPU build of PyTorch first and Algan will use it, for around 1.5 GB in total:

pip install torch --index-url https://download.pytorch.org/whl/cpu
pip install algan

2. Check the installation

algan check

This prints the Algan and PyTorch versions, the device renders will run on, and where output and cache files will go.

With an NVIDIA GPU and a driver new enough for the bundled CUDA runtime, it reports [OK] CUDA acceleration and your card’s name – there is nothing to install for that, unlike on Windows. If it says [INFO] Running on CPU, your driver is likely too old: check nvidia-smi, then pick the PyTorch build matching your CUDA version at pytorch.org/get-started/locally and install it with the index URL given there. For an AMD card, install a ROCm build of PyTorch the same way.

Algan runs on the CPU perfectly well if you have no supported GPU; it is just slower.

Optional: text and mathematical formulas

Linux is the one platform where text needs something installed first. Algan draws prose with Text and formulas with Tex; on Windows and macOS Text uses system fonts through Pango, which is installed with Algan there, but Pango publishes no Linux wheel and so is not. Until you install one of the two backends below, neither class can draw anything.

LaTeX is much the easier of the two here, and it covers both classes, so install it first:

# Debian / Ubuntu -- the same set Algan's own CI installs
sudo apt install texlive-latex-base texlive-latex-extra \
                 texlive-fonts-recommended latexmk

texlive-latex-extra pulls in texlive-latex-recommended, and the dvisvgm converter that turns LaTeX output into glyph outlines arrives with texlive-binaries, so neither needs naming separately.

On Fedora (dnf) or Arch (pacman), install the equivalent TeX Live packages for your distribution. If you would rather not work out the mapping, the complete distribution (texlive-scheme-full on Fedora, texlive-meta on Arch) always works – it is a multi-gigabyte download, which is the only reason not to recommend it first. See What Algan needs from LaTeX for the small set Algan actually uses.

With LaTeX alone, Text typesets through LaTeX’s text mode: it works, but it cannot use your system fonts and ignores the font, weight and slant arguments. To get those, install the pango extra as well. This is the one part of Algan that compiles from source on Linux – which is why it is opt-in here and automatic elsewhere – so install a compiler and the development headers first:

# Debian / Ubuntu
sudo apt install build-essential python3-dev libpango1.0-dev pkg-config
# Fedora
sudo dnf install gcc python3-devel pango-devel pkg-config
# Arch
sudo pacman -S base-devel pango

pip install "algan[pango]"

Optional: speech and 3-D models

  • Speech Generation needs a system speech engine, which Linux does not ship by default:

    sudo apt install espeak-ng    # Debian / Ubuntu
sudo dnf install espeak-ng    # Fedora
sudo pacman -S espeak-ng      # Arch

    Aligning animations to a pre-recorded audio file additionally needs pip install "algan[audio]".

  • 3-D models in glTF, GLB, OBJ and PLY are supported out of the box. FBX needs pip install "algan[fbx]" plus the native assimp library (sudo apt install libassimp5, sudo dnf install assimp, sudo pacman -S assimp).

Your first animation

With the environment activated, put the following code into my_first_animation.py inside your project folder:

from algan import *

square = Square().spawn()

Scene.save_video("example")

and run it:

python my_first_animation.py

If there is now a video at algan_outputs/example.mp4 beside the script, your installation is complete! Continue on to Getting Started to learn how to use Algan.

What Algan needs from LaTeX

Algan’s default TeX template is deliberately small. It needs only the standalone, babel, amsmath and amssymb packages, plus the latex and dvisvgm binaries. Any distribution providing those renders every Tex and MathTex Algan builds on its own, which is why a minimal distribution such as TinyTeX or BasicTeX is enough.

The wider package list, for custom Manim templates

The list below is what the Manim templates Algan can be pointed at may reach for. Install it only if you supply your own TexTemplate with extra \usepackage lines:

amsmath babel-english cbfonts-fd cm-super count1to ctex doublestroke dvisvgm everysel
fontspec frcursive fundus-calligra gnu-freefont jknapltx latex-bin
mathastext microtype multitoc physics preview prelim2e ragged2e relsize rsfs
setspace standalone tipa wasy wasysym xcolor xetex xkeyval

Installing from source

Everything above installs a released Algan for writing animations. To work on Algan itself, or to run a version newer than the latest release, clone the repository and install it from source instead, as described in Development.