Studio Config

All of the studio specific configurations are stored as simple JSON files in the pype-config repository.

Config is split into multiple sections described below.

Anatomy

Defines where and how folders and files are created for all the project data. Anatomy has two parts Roots and Templates.

warning

It is recommended to create anatomy overrides for each project even if values haven't changed. Ignoring this recommendation may cause catastrophic consequences.

Roots

Roots define where files are stored with path to shared folder. You can set them in roots.json. It is required to set root path for each platform you are using in studio. All paths must point to same folder!

{
    "windows": "P:/projects",
    "darwin": "/Volumes/projects",
    "linux": "/mnt/share/projects"
}

It is possible to set multiple roots when necessary. That may be handy when you need to store specific type of data on another disk. In that case you'll have to add one level in json.

{
    "work": {
        "windows": "P:/work",
        "darwin": "/Volumes/work",
        "linux": "/mnt/share/work"
    },
    "publish": {
        "windows": "Y:/publish",
        "darwin": "/Volumes/publish",
        "linux": "/mnt/share/publish"
    }
}

Usage of multiple roots is explained below in templates part.

Templates

Templates define project's folder structure and filenames. You can set them in default.yaml.

Required templates

We have a few required anatomy templates for Pype to work properly, however we keep adding more when needed.

work:
  folder: "{root}/{project[name]}/{hierarchy}/{asset}/work/{task}"
  file: "{project[code]}_{asset}_{task}_v{version:0>3}<_{comment}>.{ext}"
  path: "{root}/{project[name]}/{hierarchy}/{asset}/work/{task}/{project[code]}_{asset}_{task}_v{version:0>3}<_{comment}>.{ext}"

publish:
  folder: "{root}/{project[name]}/{hierarchy}/{asset}/publish/{family}/{subset}/v{version:0>3}"
  file: "{project[code]}_{asset}_{subset}_v{version:0>3}<.{frame}>.{representation}"
  path: "{root}/{project[name]}/{hierarchy}/{asset}/publish/{family}/{subset}/v{version:0>3}/{project[code]}_{asset}_{subset}_v{version:0>3}<.{frame}>.{representation}"

Template groups work and publish must be set in all circumstances. Both must have set keys as shown folder, holds path template for the directory where the files are stored, file only holds the filename and path combines the two together for quicker access.

Available keys

Context key	Description
root	Path to root folder
root[\<root name>]	Path to root folder when multiple roots are used. Key <root name> represents root key specified in roots.json
project[name]	Project's full name.
project[code]	Project's code.
hierarchy	All hierarchical parents as subfolders.
asset	Name of asset or shot.
task	Name of task.
version	Version number.
subset	Subset name.
family	Main family name.
ext	File extension. (Possible to use only in work template atm.)
representation	Representation name. (Is used instead of ext except work template atm.)
frame	Frame number for sequence files.
output
comment

warning

Be careful about using root key in templates when using multiple roots. It is not allowed to combine both {root} and {root[<root name>]} in templates.

note

It is recommended to set padding for version which is possible with additional expression in template. Entered key {version:0<3} will result into 001 if version 1 is published. Explanation: Expression 0<3 will add "0" char to the beginning(<) until string has 3 characters.

Date-Time key	Example result	Description
d	1, 30	Day of month in shortest possible way.
dd	01, 30	Day of month with 2 digits.
ddd	Mon	Shortened week day name.
dddd	Monday	Full week day name.
m	1, 12	Month number in shortest possible way.
mm	01, 12	Month number with 2 digits.
mmm	Jan	Shortened month name.
mmmm	January	Full month name.
yy	20	Shortened year.
yyyy	2020	Full year.
H	4, 17	Shortened 24-hour number.
HH	04, 17	24-hour number with 2 digits.
h	5	Shortened 12-hour number.
hh	05	12-hour number with 2 digits.
ht	AM, PM	Midday part.
M	0	Shortened minutes number.
MM	00	Minutes number with 2 digits.
S	0	Shortened seconds number.
SS	00	Seconds number with 2 digits.

Optional keys

Keys may be optional for some reason when are wrapped with < and >. But it is recommended to use only for these specific keys with obvious reasons:

• output, comment are optional to fill
• frame is used only for sequences.

Inner keys

It is possible to use value of one template key inside value of another template key. This can be done only per template group, which means it is not possible to use template key from publish group inside work group.

Usage is similar to using template keys but instead of {key} you must add @ in front of key: {@key}

With this feature work template from example above may be much easier to read and modify.

work:
  folder: "{root}/{project[name]}/{hierarchy}/{asset}/work/{task}"
  file: "{project[code]}_{asset}_{task}_v{version:0>3}<_{comment}>.{ext}"
  path: "{@folder}/{@file}"
  # This is how `path` key will look as result
  # path: "{root}/{project[name]}/{hierarchy}/{asset}/work/{task}/{project[code]}_{asset}_{task}_v{version:0>3}<_{comment}>.{ext}"

warning

Be aware of unsolvable recursion in inner keys.

group:
  # Use key where source key is used in value
  key_1: "{@key_2}"
  key_2: "{@key_1}"

  # Use itself
  key_3: "{@key_3}"

Global keys

Global keys are keys with value outside template groups. All these keys will be available in each template group with ability to override them inside the group.

Source

# Global key outside template group
global_key: "global value"

group_1:
  # `global_key` is not set
  example_key_1: "{example_value_1}"

group_2:
  # `global_key` is iverrided
  global_key: "overridden global value"

Result

global_key: "global value"

group_1:
  # `global_key` was added
  global_key: "global value"
  example_key_1: "{example_value_1}"

group_2:
  # `global_key` kept it's value for `group_2`
  global_key: "overridden global value"

Combine Inner keys with Global keys

Real power of Inner and Global keys is their combination.

Template source

# PADDING
frame_padding: 4
frame: "{frame:0>frame_padding}"
# MULTIPLE ROOT
root_name: "root_name_1"
root: {root[{@root_name}]}

group_1:
  example_key_1: "{@root}/{@frame}"

group_2:
  frame_padding: 3
  root_name: "root_name_2"
  example_key_2: "{@root}/{@frame}"

group_3:
  frame: "{frame}"
  example_key_3: "{@root}/force_value/{@frame}"

Equals

frame_padding: 4
frame: "{frame:0>3}"
root_name: "root_name_1"
root: {root[root_name_1]}

group_1:
  frame_padding: 4
  frame: "{frame:0>3}"
  root_name: "root_name_1"
  root: {root[root_name_1]}
  # `example_key_1` result
  example_key_1: "{root[root_name_1]}/{frame:0>3}"

group_2:
  frame_padding: 3
  frame: "{frame:0>3}"
  root_name: "root_name_2"
  root: {root[root_name_2]}
  # `example_key_2` result
  example_key_2: "{root[root_name_2]}/{frame:0>2}"

group_3:
  frame_padding: 4
  frame: "{frame}"
  root_name: "root_name_1"
  root: {root[root_name_1]}
  # `example_key_3` result
  example_key_3: "{root[root_name_1]}/force_value/{frame}"

warning

Be careful about using global keys. Keep in mind that all global keys will be added to all template groups and all inner keys in their values MUST be in the group. For example in required templates it seems that path: "{@folder}/{@file}" should be used as global key, but that would require all template groups have folder and file keys which is not true by default.

Environments

Here is where all the environment variables are set up. Each software has it's own environment file where we set all variables needed for it to function correctly. This is also a place where any extra in-house variables should be added. All of these individual configs and then loaded additively as needed based on current context.

For example when launching Pype Tray, Global and Avalon envs are loaded first. If the studio uses also Deadline and Ftrack, both of those environments get added at the same time. This sets the base environment for the rest of the pipeline that will be inherited by all the applications launched from this point on.

When user launches an application for a task, its general and versioned env files get added to the base before the software launches. When launching Maya 2019, both maya.json and maya_2019.json will be added.

If the project or task also has extra tools configured, say Arnold Mtoa 3.1.1, a config JSON with the same name will be added too.

This way the environment is completely dynamic with possibility of overrides on a granular level, from project all the way down to task.

Launchers

Considering that different studios use different ways of deploying software to their workstations, we need to tell Pype how to launch all the individual applications available in the studio.

Each software need multiple files prepared for it to function correctly.

application_name.toml
application_name.bat
application_name.sh

TOML file tells Pype how to work with the application across the board. Icons, Label in GUI, Ftrack settings but most importantly it defines what executable to run. These executable are stored in the windows and linux subfolder in the launchers folder. If application_name.toml defines that executable to run is application_name, Pype assumes that a .bat and .sh files under that name exist in the linux and windows folders in launchers. Correct version is picked automatically based on the platform Pype is running on.

These .bat and .sh scripts have only one job then. They have to point to the exact executable path on the system, or to a command that will launch the app we want. Version granularity is up to the studio to decide. We can show artists Nuke 11.3, while specifying the particular version 11.3v4 only in the .bat file, so the artist doesn't need to deal with it, or we can present him with 11.3v4 directly. the choice is mostly between artist control vs more configuration files on the system.

Presets

This is where most of the functions configuration of the pipeline happens. Colorspace, data types, burnin setting, geometry naming conventions, ftrack attributes, playblast settings, types of exports and lot's of other settings.

Presets are categorized in folders based on what they control or what host (DCC application) they are for. We're slowly working on documenting them all, but new ones are being created regularly as well. Hopefully the categories and names are sufficiently self-explanatory.

colorspace

Defines all available color spaces in the studio. These configs not only tell the system what OCIO to use, but also how exactly it needs to be applied in the give application. From loading the data, through previewing it all the way to rendered

Dataflow

Defines allowed file types and data formats across the pipeline including their particular coded and compression settings.

Plugins

All the creator, loader and publisher configurations are stored here. We can override any properties of the default plugin values and more.

How does it work

Overriding plugin properties is as simple as adding what needs to be changed to JSON file along with plugin name.

Say you have name validating plugin:

import pyblish.api


class ValidateModelName(pyblish.api.InstancePlugin):

    order = pype.api.ValidateContentsOrder
    hosts = ['maya']
    families = ['model']
    label = 'Validate Mesh Name'

    # check for: 'foo_001_bar_GEO`
    regex = r'.*_\d*_.*_GEO'

    def process(self, instance):
      # pseudocode to get nodes
      models = get_models(instance.data.get("setMembers", None))
      r = re.compile(self.regex)
      for model in models:
            m = r.match(obj)
            if m is None:
              raise RuntimeError("invalid name on {}".format(model))

This is just non-functional example

Instead of creating new plugin with different regex, you can put:

"ValidateModelName": {
  "regex": ".*\\d*_.*_geometry"
}

and put it into repos/pype-config/presets/plugins/maya/publish.json. There can be more entries like that for how many plugins you need.

That will effectively replace regex defined in plugin during runtime with the one you've just defined in JSON file. This way you can change any properties defined in plugin.

loader and creators

Similar way exist for Loaders and Creators. Use files create.json for Creators, load.json for Loaders and publish.json for Pyblish plugins like extractors, validators, etc.

Preset resolution works by getting host name (for example Maya) and then looking inside repos/pype-config/presets/plugins/<host>/publish.json path. If plugin is not found, then repos/pype-config/presets/plugins/global/publish.json is tried.

Per project plugin override

You can override plugins per project. See Per-project configuration

Schema

Holds all database schemas for mongoDB, that we use. In practice these are never changed on a per studio basis, however we included them in the config for cases where a particular project might need a very individual treatment.

Per-project configuration

You can have per-project configuration with Pype. This allows you to have for example different validation requirements, file naming, etc.

This is very easy to set up - point PYPE_PROJECT_CONFIGS environment variable to place where you want those per-project configurations. Then just create directory with project name and that's almost it. Inside, you can follow hierarchy of pype-config presets. Everything put there will override stuff in pype-config.

Example

You have a project where you need to disable some validators - let's say overlapping UVs validator in Maya.

Project name is FooProject. Your PYPE_PROJECT_CONFIGS points to /studio/pype/projects.

Create projects settings directory:

mkdir $PYPE_PROJECT_CONFIGS/FooProject

Now you can use plugin overrides to disable validator:

Put:

{
  "ValidateMeshHasOverlappingUVs": {
    "enabled": false
  }
}

into:

$PYPE_PROJECT_CONFIGS/FooPoject/presets/plugins/maya/publish.json

And its done. ValidateMeshHasOverlappingUVs is a class name of validator - you can find that name by looking into python file containing validator code, or in Pyblish GUI.

That way you can make it optional or set whatever properties you want on plugins and those settings will take precedence over the default site-wide settings.