- Collect the rig sets only once (I've ordered it before Collect History so that the instance still contains less node, as an optimization)
- Also fixes a hard error when `out_SET` is not found, instead now only relevant `PublishValidationError` are raised to generate a nice report
This makes it so again that it's up to the studio admin to enable or disable the plug-in per project instead of whether the modeler's workfile just happened to have the current renderer set to arnold or not. Note that the plug-in is currently disabled by default in project settings.
The new publisher always updates and imprints the publish instances with the latest Attribute Definitions of the Creator, thus making this legacy validator redundant. Currently legacy instances do not work through the new publisher (they need to be converted to work correctly) and thus even for legacy workfiles this validator is redundant.
* Use existing subset_name as group_name by default
New publisher already carries real subset name (`renderModelingMain`), it should build group name only if subset_name is weird.
* Let legacy conversion of render instance recreate subset_name
Without it would create subset names like `renderingMain` which are not matching to newly created `renderMain` instances. This would cause issue in version restarts.
* Let Render Creator for Maya create proper subset_name
It was using hardcoded logic not matching other DCCs.
* Hound
* Fix method calls
* Fix typos
* Do not import unnecessary
* Capitalize is wrong function for here
* Overwrite get_subset_name for standardized results
It makes sense to override this method for other parts of code getting same results.
* Force change
It seems that GH doesn't recognize changes with adding()
* Update openpype/hosts/maya/plugins/create/convert_legacy.py
Co-authored-by: Jakub Trllo <43494761+iLLiCiTiT@users.noreply.github.com>
* Hound
---------
Co-authored-by: Jakub Trllo <43494761+iLLiCiTiT@users.noreply.github.com>
* OP-6357 - removed unneeded import
* OP-6357 - extracted logic for getting custom group and namespace from Settings
Mimicing logic in ReferenceLoader, eg. group could be left empty >> no groupping of imported subset.
* OP-6357 - same logic for abc animation as Reference
* OP-6357 - same logic for yeti rig as ReferenceLoder
Allows to not create wrapping group.
* OP-6357 - added separate import_loader to settings
Could be used to not creating wrapping groups when Group kept empty.
* OP-6357 - added product subset conversion for ayon settings
* OP-6357 - fix using correct variable
Artist input comes from `data` not directly from self.options
* OP-6357 - add attach_to_root to options to allow control by same key
* OP-6357 - added docstring
* Added settings for Import loaders in maya
* OP-6357 - refactored formatting
* OP-6358 - allow not creation of group for reference loader in Maya
Creation of wrapping group could be controlled by checkbox in ReferenceLoader options.
* OP-6358 - group name could be empty in Settings
This enables default behavior of not creating wrapping group without need of artists unchecking `Group imported assets`.
* OP-6358 - changed to safer logic
Stripping of | was weird and potentially dangerous (collision of names), this logic should be safer.
* ApplicationManager can have more granular way how applications are launched
* executable is optional to be able create ApplicationLaunchContext
* launch context can run prelaunch hooks without launching application
* 'get_app_environments_for_context' is using launch context to prepare environments
* added 'launch_type' as one of filtering options for LaunchHook
* added 'local' launch type filter to existing launch hooks
* define 'automated' launch type in remote publish function
* modified publish and extract environments cli commands
* launch types are only for local by default
* fix import
* fix launch types of global host data
* change order or kwargs
* change unreal filter attribute
* use set instead of list
* removed '__init__' from celaction hooks
* use 'CELACTION_ROOT_DIR' in pre setup
* use full import from applications
* ApplicationManager can have more granular way how applications are launched
* executable is optional to be able create ApplicationLaunchContext
* launch context can run prelaunch hooks without launching application
* 'get_app_environments_for_context' is using launch context to prepare environments
* added 'launch_type' as one of filtering options for LaunchHook
* added 'local' launch type filter to existing launch hooks
* define 'automated' launch type in remote publish function
* modified publish and extract environments cli commands
* launch types are only for local by default
* fix import
* fix launch types of global host data
* change order or kwargs
* change unreal filter attribute
* implemented helper functions to get ffmpeg and oiio tool arguments
* modified validation functions to be able to handle list of arguments
* path getters can return a path in AYON mode if one argument is returned
* removed test exception
* modified docstrings
* is_oiio_supported is using new functions to get launch arguments
* new functions are in lib public =
* use new functions all over the place
* renamed 'ffmpeg_path' to 'ffmpeg_args'
* raise 'ToolNotFoundError' if tool argument could not be found
* reraise 'KnownPublishError' in publish plugins
* fix comment
* simplify args start
* ffmpeg and oiio function require tool name and support additional arguments
* renamed 'get_oiio_tools_args' to 'get_oiio_tool_args'
* fix variable name
