# SPDX-FileCopyrightText: 2021-2025 Blender Authors # # SPDX-License-Identifier: GPL-2.0-or-later """ Copy Global Transform Simple operators for copying world-space transforms. It's called "global" to avoid confusion with the Blender World data-block. """ import abc from typing import Iterable, Optional, Any, TypeAlias import bpy from bpy.types import ( Context, Object, Operator, PoseBone, Camera, ID, ActionChannelbag, ) from mathutils import Matrix _axis_enum_items = [ ("x", "X", "", 1), ("y", "Y", "", 2), ("z", "Z", "", 3), ] # Mapping from frame number to the dominant (in terms of genetics) key type. # GENERATED is the only recessive key type, others are dominant. KeyInfo: TypeAlias = dict[float, str] def get_matrix(context: Context) -> Matrix: bone = context.active_pose_bone if bone: # Convert matrix to world space arm = context.active_object mat = arm.matrix_world @ bone.matrix else: mat = context.active_object.matrix_world return mat def set_matrix(context: Context, mat: Matrix) -> None: from bpy_extras.anim_utils import AutoKeying bone = context.active_pose_bone if bone: # Convert matrix to local space arm_eval = context.active_object.evaluated_get(context.view_layer.depsgraph) bone.matrix = arm_eval.matrix_world.inverted() @ mat AutoKeying.autokey_transformation(context, bone) else: context.active_object.matrix_world = mat AutoKeying.autokey_transformation(context, context.active_object) def _channelbag_for_id(animated_id: ID) -> ActionChannelbag | None: # This is on purpose limited to the first layer and strip. To support more # than 1 layer, a rewrite of the caller is needed. adt = animated_id.animation_data action = adt and adt.action if action is None: return None slot = adt.action_slot for layer in action.layers: for strip in layer.strips: assert strip.type == 'KEYFRAME' channelbag = strip.channelbag(slot) return channelbag return None def _selected_keyframes(context: Context) -> list[float]: """Return the list of frame numbers that have a selected key. Only keys on the active bone/object are considered. """ bone = context.active_pose_bone if bone: return _selected_keyframes_for_bone(context.active_object, bone) return _selected_keyframes_for_object(context.active_object) def _selected_keyframes_for_bone(object: Object, bone: PoseBone) -> list[float]: """Return the list of frame numbers that have a selected key. Only keys on the given pose bone are considered. """ name = bpy.utils.escape_identifier(bone.name) return _selected_keyframes_for_action_slot(object, "pose.bones[\"{:s}\"].".format(name)) def _selected_keyframes_for_object(object: Object) -> list[float]: """Return the list of frame numbers that have a selected key. Only keys on the given object are considered. """ return _selected_keyframes_for_action_slot(object, "") def _selected_keyframes_for_action_slot(object: Object, rna_path_prefix: str) -> list[float]: """Return the list of frame numbers that have a selected key. Only keys on the given object's Action Slot on FCurves starting with rna_path_prefix are considered. """ cbag = _channelbag_for_id(object) if not cbag: return [] keyframes = set() for fcurve in cbag.fcurves: if not fcurve.data_path.startswith(rna_path_prefix): continue for kp in fcurve.keyframe_points: if not kp.select_control_point: continue keyframes.add(kp.co.x) return sorted(keyframes) def _copy_matrix_to_clipboard(window_manager: bpy.types.WindowManager, matrix: Matrix) -> None: rows = [" {!r},".format(tuple(row)) for row in matrix] as_string = "\n".join(rows) window_manager.clipboard = "Matrix((\n{:s}\n))".format(as_string) class OBJECT_OT_copy_global_transform(Operator): bl_idname = "object.copy_global_transform" bl_label = "Copy Global Transform" bl_description = ( "Copies the matrix of the currently active object or pose bone to the clipboard. Uses world-space matrices" ) # This operator cannot be un-done because it manipulates data outside Blender. bl_options = {'REGISTER'} @classmethod def poll(cls, context: Context) -> bool: return bool(context.active_pose_bone) or bool(context.active_object) def execute(self, context: Context) -> set[str]: mat = get_matrix(context) _copy_matrix_to_clipboard(context.window_manager, mat) return {'FINISHED'} def get_relative_ob(context: Context) -> Optional[Object]: """Get the 'relative' object. This is the object that's configured, or if that's empty, the active scene camera. """ rel_ob = context.scene.tool_settings.anim_relative_object return rel_ob or context.scene.camera class OBJECT_OT_copy_relative_transform(Operator): bl_idname = "object.copy_relative_transform" bl_label = "Copy Relative Transform" bl_description = "Copies the matrix of the currently active object or pose bone to the clipboard. " \ "Uses matrices relative to a specific object or the active scene camera" # This operator cannot be un-done because it manipulates data outside Blender. bl_options = {'REGISTER'} @classmethod def poll(cls, context: Context) -> bool: rel_ob = get_relative_ob(context) if not rel_ob: return False return bool(context.active_pose_bone) or bool(context.active_object) def execute(self, context: Context) -> set[str]: rel_ob = get_relative_ob(context) if not rel_ob: self.report( {'ERROR'}, "No 'Relative To' object found, set one explicitly or make sure there is an active object") return {'CANCELLED'} mat = rel_ob.matrix_world.inverted() @ get_matrix(context) _copy_matrix_to_clipboard(context.window_manager, mat) return {'FINISHED'} class UnableToMirrorError(Exception): """Raised when mirroring is enabled but no mirror object/bone is set.""" class OBJECT_OT_paste_transform(Operator): bl_idname = "object.paste_transform" bl_label = "Paste Global Transform" bl_description = ( "Pastes the matrix from the clipboard to the currently active pose bone or object. Uses world-space matrices" ) bl_options = {'REGISTER', 'UNDO'} _method_items = [ ( 'CURRENT', "Current Transform", "Paste onto the current values only, only manipulating the animation data if auto-keying is enabled", ), ( 'EXISTING_KEYS', "Selected Keys", "Paste onto frames that have a selected key, potentially creating new keys on those frames", ), ( 'BAKE', "Bake on Key Range", "Paste onto all frames between the first and last selected key, creating new keyframes if necessary", ), ] method: bpy.props.EnumProperty( # type: ignore items=_method_items, name="Paste Method", description="Update the current transform, selected keyframes, or even create new keys", options={'SKIP_SAVE'}, ) bake_step: bpy.props.IntProperty( # type: ignore name="Frame Step", description="Only used for baking. Step=1 creates a key on every frame, step=2 bakes on 2s, etc", min=1, soft_min=1, soft_max=5, options={'SKIP_SAVE'}, ) use_mirror: bpy.props.BoolProperty( # type: ignore name="Mirror Transform", description="When pasting, mirror the transform relative to a specific object or bone", default=False, options={'SKIP_SAVE'}, ) mirror_axis_loc: bpy.props.EnumProperty( # type: ignore items=_axis_enum_items, name="Location Axis", description="Coordinate axis used to mirror the location part of the transform", default='x', options={'SKIP_SAVE'}, ) mirror_axis_rot: bpy.props.EnumProperty( # type: ignore items=_axis_enum_items, name="Rotation Axis", description="Coordinate axis used to mirror the rotation part of the transform", default='z', options={'SKIP_SAVE'}, ) use_relative: bpy.props.BoolProperty( # type: ignore name="Use Relative Paste", description="When pasting, assume the pasted matrix is relative to another object (set in the user interface)", default=False, options={'SKIP_SAVE'}, ) @classmethod def poll(cls, context: Context) -> bool: if not context.active_pose_bone and not context.active_object: cls.poll_message_set("Select an object or pose bone") return False return True @classmethod def string_to_matrix(cls, value: str) -> Matrix | None: if value.startswith("Matrix"): return cls.parse_matrix(value) if value.startswith(" Matrix | None: import ast try: return Matrix(ast.literal_eval(value[6:])) except Exception: # ast.literal_eval() can raise a slew of exceptions, all of # which means that it's not a matrix on the clipboard. return None @staticmethod def parse_print_m4(value: str) -> Optional[Matrix]: """Parse output from Blender's print_m4() function. Expects four lines of space-separated floats. """ lines = value.strip().splitlines() if len(lines) != 4: return None try: floats = tuple(tuple(float(item) for item in line.split()) for line in lines) except ValueError: # Apparently not the expected format. return None return Matrix(floats) @staticmethod def parse_repr_m4(value: str) -> Optional[Matrix]: """Four lines of (a, b, c, d) floats.""" lines = value.strip().splitlines() if len(lines) != 4: return None try: floats = tuple(tuple(float(item.strip()) for item in line.strip()[1:-1].split(',')) for line in lines) except ValueError: # Apparently not the expected format. return None return Matrix(floats) def execute(self, context: Context) -> set[str]: clipboard = context.window_manager.clipboard.strip() mat = self.string_to_matrix(clipboard) if mat is None: self.report({'ERROR'}, "Clipboard does not contain a matrix") return {'CANCELLED'} try: mat = self._preprocess_matrix(context, mat) except UnableToMirrorError: self.report({'ERROR'}, "Unable to mirror, no mirror object/bone configured") return {'CANCELLED'} applicator = { 'CURRENT': self._paste_current, 'EXISTING_KEYS': self._paste_existing_keys, 'BAKE': self._paste_bake, }[self.method] return applicator(context, mat) def _preprocess_matrix(self, context: Context, matrix: Matrix) -> Matrix: if self.use_relative: matrix = self._relative_to_world(context, matrix) if self.use_mirror: matrix = self._mirror_matrix(context, matrix) return matrix def _relative_to_world(self, context: Context, matrix: Matrix) -> Matrix: rel_ob = get_relative_ob(context) if not rel_ob: return matrix rel_ob_eval = rel_ob.evaluated_get(context.view_layer.depsgraph) return rel_ob_eval.matrix_world @ matrix def _mirror_matrix(self, context: Context, matrix: Matrix) -> Matrix: mirror_ob = context.scene.tool_settings.anim_mirror_object mirror_bone = context.scene.tool_settings.anim_mirror_bone # No mirror object means "current armature object". ctx_ob = context.object if not mirror_ob and mirror_bone and ctx_ob and ctx_ob.type == 'ARMATURE': mirror_ob = ctx_ob if not mirror_ob: raise UnableToMirrorError() if mirror_ob.type == 'ARMATURE' and mirror_bone: return self._mirror_over_bone(matrix, mirror_ob, mirror_bone) return self._mirror_over_ob(matrix, mirror_ob) def _mirror_over_ob(self, matrix: Matrix, mirror_ob: bpy.types.Object) -> Matrix: mirror_matrix = mirror_ob.matrix_world return self._mirror_over_matrix(matrix, mirror_matrix) def _mirror_over_bone(self, matrix: Matrix, mirror_ob: bpy.types.Object, mirror_bone_name: str) -> Matrix: bone = mirror_ob.pose.bones[mirror_bone_name] mirror_matrix = mirror_ob.matrix_world @ bone.matrix return self._mirror_over_matrix(matrix, mirror_matrix) def _mirror_over_matrix(self, matrix: Matrix, mirror_matrix: Matrix) -> Matrix: # Compute the matrix in the space of the mirror matrix: mat_local = mirror_matrix.inverted() @ matrix # Decompose the matrix, as we don't want to touch the scale. This # operator should only mirror the translation and rotation components. trans, rot_q, scale = mat_local.decompose() # Mirror the translation component: axis_index = ord(self.mirror_axis_loc) - ord('x') trans[axis_index] *= -1 # Flip the rotation, and use a rotation order that applies the to-be-flipped axes first. match self.mirror_axis_rot: case 'x': rot_e = rot_q.to_euler('XYZ') rot_e.x *= -1 # Flip the requested rotation axis. rot_e.y *= -1 # Also flip the bone roll. case 'y': rot_e = rot_q.to_euler('YZX') rot_e.y *= -1 # Flip the requested rotation axis. rot_e.z *= -1 # Also flip another axis? Not sure how to handle this one. case 'z': rot_e = rot_q.to_euler('ZYX') rot_e.z *= -1 # Flip the requested rotation axis. rot_e.y *= -1 # Also flip the bone roll. # Recompose the local matrix: mat_local = Matrix.LocRotScale(trans, rot_e, scale) # Go back to world space: mirrored_world = mirror_matrix @ mat_local return mirrored_world @staticmethod def _paste_current(context: Context, matrix: Matrix) -> set[str]: set_matrix(context, matrix) return {'FINISHED'} def _paste_existing_keys(self, context: Context, matrix: Matrix) -> set[str]: if not context.scene.tool_settings.use_keyframe_insert_auto: self.report({'ERROR'}, "This mode requires auto-keying to work properly") return {'CANCELLED'} frame_numbers = _selected_keyframes(context) if not frame_numbers: self.report({'WARNING'}, "No selected frames found") return {'CANCELLED'} self._paste_on_frames(context, frame_numbers, matrix) return {'FINISHED'} def _paste_bake(self, context: Context, matrix: Matrix) -> set[str]: if not context.scene.tool_settings.use_keyframe_insert_auto: self.report({'ERROR'}, "This mode requires auto-keying to work properly") return {'CANCELLED'} bake_step = max(1, self.bake_step) # Put the clamped bake step back into RNA for the redo panel. self.bake_step = bake_step frame_start, frame_end = self._determine_bake_range(context) frame_range = range(round(frame_start), round(frame_end) + bake_step, bake_step) self._paste_on_frames(context, frame_range, matrix) return {'FINISHED'} def _determine_bake_range(self, context: Context) -> tuple[float, float]: frame_numbers = _selected_keyframes(context) if frame_numbers: # Note that these could be the same frame, if len(frame_numbers) == 1: return frame_numbers[0], frame_numbers[-1] if context.scene.use_preview_range: self.report({'INFO'}, "No selected keys, pasting over preview range") return context.scene.frame_preview_start, context.scene.frame_preview_end self.report({'INFO'}, "No selected keys, pasting over scene range") return context.scene.frame_start, context.scene.frame_end def _paste_on_frames(self, context: Context, frame_numbers: Iterable[float], matrix: Matrix) -> None: current_frame = context.scene.frame_current_final try: for frame in frame_numbers: context.scene.frame_set(int(frame), subframe=frame % 1.0) set_matrix(context, matrix) finally: context.scene.frame_set(int(current_frame), subframe=current_frame % 1.0) class Transformable(metaclass=abc.ABCMeta): """Interface for a bone or an object.""" def __init__(self) -> None: self._key_info_cache: Optional[KeyInfo] = None @abc.abstractmethod def matrix_world(self) -> Matrix: pass def set_matrix_world(self, context: Context, matrix: Matrix) -> None: """Set the world matrix, without auto-keying.""" self._set_matrix_world(context, matrix) def set_matrix_world_autokey(self, context: Context, matrix: Matrix) -> None: """Set the world matrix, and auto-key the resulting transform.""" self._set_matrix_world(context, matrix) self._autokey_matrix_world(context) @abc.abstractmethod def _set_matrix_world(self, context: Context, matrix: Matrix) -> None: pass @abc.abstractmethod def _autokey_matrix_world(self, context: Context) -> None: pass @abc.abstractmethod def _my_fcurves(self) -> Iterable[bpy.types.FCurve]: pass def key_info(self) -> KeyInfo: if self._key_info_cache is not None: return self._key_info_cache keyinfo: KeyInfo = {} for fcurve in self._my_fcurves(): for kp in fcurve.keyframe_points: frame = kp.co.x if kp.type == 'GENERATED' and frame in keyinfo: # Don't bother overwriting other key types. continue keyinfo[frame] = kp.type self._key_info_cache = keyinfo return keyinfo def remove_keys_of_type( self, key_type: str, *, frame_start: float | int = float("-inf"), frame_end: float | int = float("inf")) -> None: self._key_info_cache = None for fcurve in self._my_fcurves(): to_remove = [ kp for kp in fcurve.keyframe_points if kp.type == key_type and (frame_start <= kp.co.x <= frame_end) ] for kp in reversed(to_remove): fcurve.keyframe_points.remove(kp, fast=True) fcurve.keyframe_points.handles_recalc() class TransformableObject(Transformable): object: Object def __init__(self, object: Object) -> None: super().__init__() self.object = object def __str__(self) -> str: return "TransformableObject({:s})".format(self.object.name) def matrix_world(self) -> Matrix: return self.object.matrix_world def _set_matrix_world(self, _context: Context, matrix: Matrix) -> None: self.object.matrix_world = matrix def _autokey_matrix_world(self, context: Context) -> None: from bpy_extras.anim_utils import AutoKeying AutoKeying.autokey_transformation(context, self.object) def __hash__(self) -> int: return hash(self.object.as_pointer()) def _my_fcurves(self) -> Iterable[bpy.types.FCurve]: cbag = _channelbag_for_id(self.object) if not cbag: return yield from cbag.fcurves class TransformableBone(Transformable): arm_object: Object pose_bone: PoseBone def __init__(self, pose_bone: PoseBone) -> None: super().__init__() self.arm_object = pose_bone.id_data self.pose_bone = pose_bone def __str__(self) -> str: return "TransformableBone({:s}, bone={:s})".format(self.arm_object.name, self.pose_bone.name) def matrix_world(self) -> Matrix: mat = self.arm_object.matrix_world @ self.pose_bone.matrix return mat def _set_matrix_world(self, context: Context, matrix: Matrix) -> None: # Convert matrix to armature-local space arm_eval = self.arm_object.evaluated_get(context.view_layer.depsgraph) self.pose_bone.matrix = arm_eval.matrix_world.inverted() @ matrix def _autokey_matrix_world(self, context: Context) -> None: from bpy_extras.anim_utils import AutoKeying AutoKeying.autokey_transformation(context, self.pose_bone) def __hash__(self) -> int: return hash(self.pose_bone.as_pointer()) def _my_fcurves(self) -> Iterable[bpy.types.FCurve]: cbag = _channelbag_for_id(self.arm_object) if not cbag: return rna_prefix = self.pose_bone.path_from_id() + "." for fcurve in cbag.fcurves: if fcurve.data_path.startswith(rna_prefix): yield fcurve class FixToCameraCommon: """Common functionality for the Fix To Scene Camera operator + its 'delete' button.""" keytype = 'GENERATED' # Operator method stubs to avoid PyLance/MyPy errors: @classmethod def poll_message_set(cls, message: str) -> None: super().poll_message_set(message) def report(self, level: set[str], message: str) -> None: super().report(level, message) # Implement in subclass: def _execute(self, context: Context, transformables: list[Transformable]) -> None: raise NotImplementedError() @classmethod def poll(cls, context: Context) -> bool: if not context.active_pose_bone and not context.active_object: cls.poll_message_set("Select an object or pose bone") return False if context.mode not in {'POSE', 'OBJECT'}: cls.poll_message_set("Switch to Pose or Object mode") return False if not context.scene.camera: cls.poll_message_set("The Scene needs a camera") return False return True def execute(self, context: Context) -> set[str]: match context.mode: case 'OBJECT': transformables = self._transformable_objects(context) case 'POSE': transformables = self._transformable_pbones(context) case mode: self.report({'ERROR'}, "Unsupported mode: {!r}".format(mode)) return {'CANCELLED'} restore_frame = context.scene.frame_current restore_matrices = [(transformable, transformable.matrix_world().copy()) for transformable in transformables] try: self._execute(context, transformables) finally: # Restore the state of the scene & the transformables. This is necessary # as not all properties may have been auto-keyed (for example 'only # available' enabled, and rotation is not actually keyed yet), so we can't # assume that going to the original frame restores the entire matrix. context.scene.frame_set(restore_frame) for transformable, matrix in restore_matrices: transformable.set_matrix_world(context, matrix) return {'FINISHED'} def _transformable_objects(self, context: Context) -> list[Transformable]: return [TransformableObject(object=ob) for ob in context.selected_editable_objects] def _transformable_pbones(self, context: Context) -> list[Transformable]: return [TransformableBone(pose_bone=bone) for bone in context.selected_pose_bones] class OBJECT_OT_fix_to_camera(FixToCameraCommon, Operator): bl_idname = "object.fix_to_camera" bl_label = "Fix to Scene Camera" bl_description = "Generate new keys to fix the selected object/bone to the camera on unkeyed frames" bl_options = {'REGISTER', 'UNDO'} use_location: bpy.props.BoolProperty( # type: ignore name="Location", description="Create Location keys when fixing to the scene camera", default=True, ) use_rotation: bpy.props.BoolProperty( # type: ignore name="Rotation", description="Create Rotation keys when fixing to the scene camera", default=True, ) use_scale: bpy.props.BoolProperty( # type: ignore name="Scale", description="Create Scale keys when fixing to the scene camera", default=True, ) def _get_matrices(self, camera: Camera, transformables: list[Transformable]) -> dict[Transformable, Matrix]: camera_mat_inv = camera.matrix_world.inverted() return {t: camera_mat_inv @ t.matrix_world() for t in transformables} def _execute(self, context: Context, transformables: list[Transformable]) -> None: from bpy_extras.anim_utils import AutoKeying from bpy_extras.wm_utils import progress_report depsgraph = context.view_layer.depsgraph scene = context.scene scene.frame_set(scene.frame_start) camera_eval = scene.camera.evaluated_get(depsgraph) last_camera_name = scene.camera.name matrices = self._get_matrices(camera_eval, transformables) if scene.use_preview_range: frame_start = scene.frame_preview_start frame_end = scene.frame_preview_end else: frame_start = scene.frame_start frame_end = scene.frame_end with ( AutoKeying.options( keytype=self.keytype, use_loc=self.use_location, use_rot=self.use_rotation, use_scale=self.use_scale, force_autokey=True, ), progress_report.ProgressReport(context.window_manager) as progress, ): frames_to_visit = range(frame_start, frame_end + scene.frame_step, scene.frame_step) progress.enter_substeps(len(frames_to_visit)) for frame in frames_to_visit: scene.frame_set(frame) progress.step() camera_eval = scene.camera.evaluated_get(depsgraph) cam_matrix_world = camera_eval.matrix_world camera_mat_inv = cam_matrix_world.inverted() if scene.camera.name != last_camera_name: # The scene camera changed, so the previous # relative-to-camera matrices can no longer be used. matrices = self._get_matrices(camera_eval, transformables) last_camera_name = scene.camera.name for t, camera_rel_matrix in matrices.items(): key_info = t.key_info() key_type = key_info.get(frame, "") if key_type not in {self.keytype, ""}: # Manually set key, remember the current camera-relative matrix. matrices[t] = camera_mat_inv @ t.matrix_world() continue # No key, or a generated one. Overwrite it with a new transform. t.set_matrix_world_autokey(context, cam_matrix_world @ camera_rel_matrix) class OBJECT_OT_delete_fix_to_camera_keys(Operator, FixToCameraCommon): bl_idname = "object.delete_fix_to_camera_keys" bl_label = "Delete Generated Keys" bl_description = "Delete all keys that were generated by the 'Fix to Scene Camera' operator" bl_options = {'REGISTER', 'UNDO'} def _execute(self, context: Context, transformables: list[Transformable]) -> None: scene = context.scene if scene.use_preview_range: frame_start = scene.frame_preview_start frame_end = scene.frame_preview_end else: frame_start = scene.frame_start frame_end = scene.frame_end for t in transformables: t.remove_keys_of_type(self.keytype, frame_start=frame_start, frame_end=frame_end) # MessageBus subscription to monitor changes & refresh panels. _msgbus_owner = object() def _refresh_3d_panels(): refresh_area_types = {'VIEW_3D'} for win in bpy.context.window_manager.windows: for area in win.screen.areas: if area.type not in refresh_area_types: continue area.tag_redraw() classes = ( OBJECT_OT_copy_global_transform, OBJECT_OT_copy_relative_transform, OBJECT_OT_paste_transform, OBJECT_OT_fix_to_camera, OBJECT_OT_delete_fix_to_camera_keys, ) def _register_message_bus() -> None: bpy.msgbus.subscribe_rna( key=(bpy.types.ToolSettings, "use_keyframe_insert_auto"), owner=_msgbus_owner, args=(), notify=_refresh_3d_panels, options={'PERSISTENT'}, ) def _unregister_message_bus() -> None: bpy.msgbus.clear_by_owner(_msgbus_owner) @bpy.app.handlers.persistent # type: ignore def _on_blendfile_load_post(_none: Any, _other_none: Any) -> None: # The parameters are required, but both are None. _register_message_bus() def register(): bpy.app.handlers.load_post.append(_on_blendfile_load_post) def unregister(): _unregister_message_bus() bpy.app.handlers.load_post.remove(_on_blendfile_load_post)