Pype Setup


The general approach to pype deployment is installing central repositories on a shared network storage which can be accessed by all artists in the studio. Simple shortcuts to launchers are then distributed to all workstations for artists to use. This approach ensures easy maintenance and updates.

When artist first runs pype all the required python packages get installed automatically to his local workstation and updated everytime there is a change in the central installation.


Automatic workstation installation and updates will not work in offline scenarios. In these case pype install --force --offline command must be triggered explicitly on the workstation


Python 3.6+

Pype requires Python 3.6 or later to be installed on each workstation running Pype.


If you want to use pype with Blender, you need to upgrade your python to 3.7 or higher.

Windows version of Python can be easily grabbed at Install location doesn't matter but python executable should be in PATH environment variable.


On linux it is somehow different and all depends on linux distribution in use.

Some linux variants (for example Ubuntu) need python-dev variant of python package that includes python headers and developer tools. This is needed because some of Pype requirements need to compile themselves against python during their installation. Please, refer to your distribution community to find out how to install that package.

sudo yum group install "Development Tools"

Python 3.6 is not part of official distribution. Easiest way is to add it with the help of SCL - Software Collection project. This has advantage that it won't replace system version of python.

sudo yum update
sudo yum install centos-release-scl

Now you can install python itself:

sudo yum install rh-python36

To be able to use installed version of python, you must enable it in shell:

scl enable rh-python36 bash

This will enable python 3.6 in currently running bash only!

Check it with:

python --version
Override Python detection

You can override autodetection of Python. This can be useful if you want to use central network python location or some other custom location. Just set PYPE_PYTHON_EXE environment variable to point where you need.


Pype needs site-wide installation of MongoDB. It should be installed on reliable server all workstations (and possibly render nodes) can connect. This server holds Avalon database that is at the core of everything, containing very important data, so it should be backed up often and if high-availability is needed, replication feature of MongoDB should be considered. This is beyond the scope of this documentation, please refer to MongoDB Documentation.

Pype can run it's own instance of mongodb, mostly for testing and development purposes. For that it uses locally installed MongoDB.

Download it from mognoDB website, install it and add to the PATH. On Windows, Pype tries to find it in standard installation destination or using PATH.

To run mongoDB on server, use your server distribution tools to set it up (on Linux).


To be able to deploy Pype, git is need. It will clone all required repositories and control versions so future updates are easier. Git is however only requirent on admin workstation for global studio updates.

See how to install git.

To access private repositories, you'll need other optional stuff like ssh key agents, etc.

PowerShell (on Windows only)

PowerShell is now included in recent versions of Windows. Pype requires at least version 5.0, included in Windows 10 from beginning and available for Windows 7 SP1, Windows 8.1 and Windows Server 2012.

If you want to know what version of PowerShell are you running, execute in PowerShell prompt:


If you need to install PowerShell or update it, please refer to: Installing powershell on windows

Xcode CLT (on Mac only)

Pype need Xcode Command Line Tools installed to provide its tools and be able to install its dependencies via pythons pip command. Those will be downloaded and installed automatically if needed.


Linux headless server

If you need to run Pype's ftrack event server on headless linux server, be aware that due Qt dependencies, you'll need to install OpenGL support there even if server doesn't have any real use for it.

Studio Setup

Pype location

Before you install Pype, first clone pype-setup repository to place you want Pype to be. In studio setting you probably want that destination to be on shared network drive so all users can access it.

production and development branch

We recommend to maintain two versions of Pype. The first is production branch - the one your artists use everyday for their work. The second one is development version you should use for testing new features and as development sandbox. Development branch can point to a different Avalon database and use its own ftrack event server. More on that in Pype Configuration

Shared Network Drive
├─── pype
├─── prod
└─── dev

To prepare this structure, you can use:

cd /shared_drive/pype
git clone --tag 2.4.0 prod
git clone --tag 2.4.0 dev

Specify your version after --branch or --tag option.


It is possible to distinguish dev and prod by changing Pype icon color to orange. To do so, you have to create config.ini file with content:


And put the file to:

Shared Network Drive
├─── dev
└─── pype-setup

You should always use tags to checkout to get specific release, otherwise you end up with develop branch that can be unstable.


By default, both branches will use the same virtual environment. Be careful when modifying your requirements in dev version because then it will influence prod version as well. To be safe, modify PYPE_ENV environment variable before using dev Pype commands.


To install Pype you need to get first to it's root directory in powershell.

If you have Pype on network, you should mount it as drive and assign it some consistent letter. You can also mount this network drive via junction point feature. As admin run from shell:

mklink /d "C:\pipeline" "\\server\pipeline"

Then your network drive will be available transparently at C:\pipeline.

cd Z:\pype\production\pype-setup

Now you can run installation itself:

.\pype.bat install

What it basically does is: 1) Create python virtual environment on path: C:\Users\Public\pype_env2 on Windows or /opt/pype/pype_env2 on Linux or /usr/local/pype/pype_env2 on Mac. This can be overridden by setting PYPE_ENV to different path. Logic behind this is that this directory on Windows can be shared between users on one machine - it only stores Pype dependencies, not any personal setting or credentials.

2) Then it will install all python dependencies defined in pypeapp\requirements.txt into this virtual environment.

Default installation will use your internet connection to download all necessary requirements.

Offline installation

You can also install Pype in offline scenarios:

pype install --offline

This will use dependencies downloaded into pype-setup/vendor/packages rather than pulling directly from the internet. Those packages must, however, first be downloaded on a machine connected to the internet using:

pype download
multiple platforms

pype download will only download packages for currently running platform. So if you run it on Windows machine, only windows packages get downloaded (along with many universal ones). If you then run pype install --offline on Linux machine, it will probably fail as Linux specific packages will be missing. In multiplatform environments we recommend to run pype download on all used platform to combine all necessary packages into vendor/packages.

multiplatform caveat

There can be problems with libraries compatibility, when using multiplatform environments. For example if using PyQt 5.12, there seems to be no problem on Windows, but using it on Centos Linux 7 will cause problems because Centos ships with some older dependent libraries that will not work with aforementioned PyQt version.

Forcing Installation

Sometime it is necessary to force re-install Pype environment. To do this:

pype install --force


pype install --force --offline

in offline scenarios.

This is useful if Pype is misbehaving as first line of debugging. You can of course just manually delete PYPE_ENV directory and run pype install again.


After Pype is cloned and installed, it is necessary to deploy all repositories used by Pype. This must be done on a computer with Internet access.

pype deploy

This command will process all repositories specified in deploy/deploy.json and clone them into repos/ directory.

pype deploy --force

will deploy repositories, overwriting existing ones if they exists and setting them to state specified in deploy.json.

customizing deployment

You can customize your deployment to some extent. Everything specified in deploy/deploy.json is considered as default and can be overridden by creating your own deploy.json in sub directory.

├─── pypeapp
├─── deploy
│ ├─── deploy.json
│ ├─── deploy_schema-1.0.json
│ ├─── my_studio_override
│ │ ├─── deploy.json
│ │ └─── deploy_schema-1.0.json
│ ...

In such configuration, deploy/my_studio_override/deploy.json will take precedence over the default one.

To validate if Pype deployment is ok, run:

pype validate

Structure of deploy.json

There are a few features in deploy.json that needs to be explained in further detail.

Here is a list of keys used and their function:

  • PYPE_CONFIG - path to Pype configuration repository.

  • init_env - these are environment files in Pype configuration repository that are loaded immediately after Pype starts. They define basic functionality.

    "init_env": ["global", "avalon", "ftrack", "deadline"]

    For example, if you don't use Deadline but you need Muster support, change deadline to muster. Pype will then load {PYPE_CONFIG}/environments/muster.json and set environment variables there.

  • repositories: this is list of repositories that will be deployed to repos/. There are few options for each repository:

    • name: name of repository will be used as directory name
    • url: url of the git repository
    • branch or tag: this specify either branch - it's HEAD will be checked out, or tag - commit tagged with specified tag will be checked out.
  • pip: these are additional dependencies to be installed by pip to virtual environment.

  • archive_files: archive files to be unpacked to somewhere. For example ffmpeg installation or anything else we need to extract during deployment to some place.

    • extract_path: path to where this archive should be extracted
    • url or vendor: this is url of source to be downloaded or name in vendor/packages to be Used
    • md5_url optional url for md5 file to validate checksum of downloaded file
    • skip_first_subfolder will move everything inside first directory in archive to extract_path.

Offline Deployment

In offline scenarios it is up to you to replicate what pype deploy does. The easiest way to go is to run pype deploy on machine, get everything in repos/ and move it to your studio install location:

cd pype-setup
tar cvzf pype_repos.tgz repos/

do the same for things deployed to vendor.