tornavis/tests/python/bl_alembic_io_test.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

413 lines
16 KiB
Python
Raw Normal View History

# SPDX-FileCopyrightText: 2017-2022 Blender Authors
#
# SPDX-License-Identifier: GPL-2.0-or-later
"""
./blender.bin --background -noaudio --factory-startup --python tests/python/bl_alembic_io_test.py -- --testdir /path/to/lib/tests/alembic
"""
import math
import pathlib
import sys
import tempfile
import unittest
import bpy
args = None
class AbstractAlembicTest(unittest.TestCase):
@classmethod
def setUpClass(cls):
cls.testdir = args.testdir
def setUp(self):
self.assertTrue(self.testdir.exists(),
'Test dir %s should exist' % self.testdir)
# Make sure we always start with a known-empty file.
bpy.ops.wm.open_mainfile(filepath=str(self.testdir / "empty.blend"))
def assertAlmostEqualFloatArray(self, actual, expect, places=6, delta=None):
"""Asserts that the arrays of floats are almost equal."""
self.assertEqual(len(actual), len(expect),
'Actual array has %d items, expected %d' % (len(actual), len(expect)))
for idx, (act, exp) in enumerate(zip(actual, expect)):
self.assertAlmostEqual(act, exp, places=places, delta=delta,
msg='%f != %f at index %d' % (act, exp, idx))
class SimpleImportTest(AbstractAlembicTest):
def test_import_cube_hierarchy(self):
res = bpy.ops.wm.alembic_import(
filepath=str(self.testdir / "cubes-hierarchy.abc"),
as_background_job=False)
self.assertEqual({'FINISHED'}, res)
# The objects should be linked to scene.collection in Blender 2.8,
# and to scene in Blender 2.7x.
objects = bpy.context.scene.collection.objects
self.assertEqual(13, len(objects))
# Test the hierarchy.
self.assertIsNone(objects['Cube'].parent)
self.assertEqual(objects['Cube'], objects['Cube_001'].parent)
self.assertEqual(objects['Cube'], objects['Cube_002'].parent)
self.assertEqual(objects['Cube'], objects['Cube_003'].parent)
self.assertEqual(objects['Cube_003'], objects['Cube_004'].parent)
self.assertEqual(objects['Cube_003'], objects['Cube_005'].parent)
self.assertEqual(objects['Cube_003'], objects['Cube_006'].parent)
def test_inherit_or_not(self):
res = bpy.ops.wm.alembic_import(
filepath=str(self.testdir / "T52022-inheritance.abc"),
as_background_job=False)
self.assertEqual({'FINISHED'}, res)
# The objects should be linked to scene.collection in Blender 2.8,
# and to scene in Blender 2.7x.
objects = bpy.context.scene.collection.objects
# ABC parent is top-level object, which translates to nothing in Blender
self.assertIsNone(objects['locator1'].parent)
# ABC parent is locator1, but locator2 has "inherits Xforms" = false, which
# translates to "no parent" in Blender.
self.assertIsNone(objects['locator2'].parent)
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
depsgraph = bpy.context.evaluated_depsgraph_get()
# Shouldn't have inherited the ABC parent's transform.
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
loc2 = depsgraph.id_eval_get(objects['locator2'])
x, y, z = objects['locator2'].matrix_world.to_translation()
self.assertAlmostEqual(0, x)
self.assertAlmostEqual(0, y)
self.assertAlmostEqual(2, z)
# ABC parent is inherited and translates to normal parent in Blender.
self.assertEqual(objects['locator2'], objects['locatorShape2'].parent)
# Should have inherited its ABC parent's transform.
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
locshp2 = depsgraph.id_eval_get(objects['locatorShape2'])
x, y, z = locshp2.matrix_world.to_translation()
self.assertAlmostEqual(0, x)
self.assertAlmostEqual(0, y)
self.assertAlmostEqual(2, z)
def test_select_after_import(self):
# Add a sphere, so that there is something in the scene, selected, and active,
# before we do the Alembic import.
bpy.ops.mesh.primitive_uv_sphere_add()
sphere = bpy.context.active_object
self.assertEqual('Sphere', sphere.name)
self.assertEqual([sphere], bpy.context.selected_objects)
bpy.ops.wm.alembic_import(
filepath=str(self.testdir / "cubes-hierarchy.abc"),
as_background_job=False)
# The active object is probably the first one that was imported, but this
2023-08-25 00:56:58 +02:00
# behavior is not defined. At least it should be one of the cubes, and
# not the sphere.
self.assertNotEqual(sphere, bpy.context.active_object)
self.assertTrue('Cube' in bpy.context.active_object.name)
# All cubes should be selected, but the sphere shouldn't be.
for ob in bpy.data.objects:
self.assertEqual('Cube' in ob.name, ob.select_get())
def test_change_path_constraint(self):
fname = 'cube-rotating1.abc'
abc = self.testdir / fname
relpath = bpy.path.relpath(str(abc))
res = bpy.ops.wm.alembic_import(filepath=str(abc), as_background_job=False)
self.assertEqual({'FINISHED'}, res)
cube = bpy.context.active_object
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
depsgraph = bpy.context.evaluated_depsgraph_get()
# Check that the file loaded ok.
bpy.context.scene.frame_set(10)
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
cube = depsgraph.id_eval_get(cube)
x, y, z = cube.matrix_world.to_euler('XYZ')
self.assertAlmostEqual(x, 0)
self.assertAlmostEqual(y, 0)
self.assertAlmostEqual(z, math.pi / 2, places=5)
# Change path from absolute to relative. This should not break the animation.
bpy.context.scene.frame_set(1)
bpy.data.cache_files[fname].filepath = relpath
bpy.context.scene.frame_set(10)
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
cube = depsgraph.id_eval_get(cube)
x, y, z = cube.matrix_world.to_euler('XYZ')
self.assertAlmostEqual(x, 0)
self.assertAlmostEqual(y, 0)
self.assertAlmostEqual(z, math.pi / 2, places=5)
# Replace the Alembic file; this should apply new animation.
bpy.data.cache_files[fname].filepath = relpath.replace('1.abc', '2.abc')
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
depsgraph.update()
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
cube = depsgraph.id_eval_get(cube)
x, y, z = cube.matrix_world.to_euler('XYZ')
self.assertAlmostEqual(x, math.pi / 2, places=5)
self.assertAlmostEqual(y, 0)
self.assertAlmostEqual(z, 0)
def test_change_path_modifier(self):
fname = 'animated-mesh.abc'
abc = self.testdir / fname
relpath = bpy.path.relpath(str(abc))
res = bpy.ops.wm.alembic_import(filepath=str(abc), as_background_job=False)
self.assertEqual({'FINISHED'}, res)
plane = bpy.context.active_object
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
depsgraph = bpy.context.evaluated_depsgraph_get()
# Check that the file loaded ok.
bpy.context.scene.frame_set(6)
scene = bpy.context.scene
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
plane_eval = plane.evaluated_get(depsgraph)
mesh = plane_eval.to_mesh()
self.assertAlmostEqual(-1, mesh.vertices[0].co.x)
self.assertAlmostEqual(-1, mesh.vertices[0].co.y)
self.assertAlmostEqual(0.5905638933181763, mesh.vertices[0].co.z)
plane_eval.to_mesh_clear()
# Change path from absolute to relative. This should not break the animation.
scene.frame_set(1)
bpy.data.cache_files[fname].filepath = relpath
scene.frame_set(6)
Dependency graph API changes Main goal here is to make it obvious and predictable about what is going on. Summary of changes. - Access to dependency graph is now only possible to a fully evaluated graph. This is now done via context.evaluated_depsgraph_get(). The call will ensure both relations and datablocks are updated. This way we don't allow access to some known bad state of the graph, and also making explicit that getting update dependency graph is not cheap. - Access to evaluated ID is now possible via id.evaluated_get(). It was already possible to get evaluated ID via dependency graph, but that was a bit confusing why access to original is done via ID and to evaluated via depsgraph. If datablock is not covered by dependency graph it will be returned as-is. - Similarly, request for original from an ID which is not evaluated will return ID as-is. - Removed scene.update(). This is very expensive to update all the view layers. - Added depsgraph.update(). Now when temporary changes to objects are to be done, this is to happen on original object and then dependency graph is to be updated. - Changed object.to_mesh() to behave the following way: * When is used for original object modifiers are ignored. For meshes this acts similar to mesh-copy, not very useful but allows to keep code paths similar (i.e. for exporter which has Apply Modifiers option it's only matter choosing between original and evaluated object, the to_mesh() part can stay the same). For curves this gives a mesh which is constructed from displist without taking own modifiers and modifiers of bevel/taper objects into account. For metaballs this gives empty mesh. Polygonization of metaball is not possible from a single object. * When is used for evaluated object modifiers are always applied. In fact, no evaluation is happening, the mesh is either copied as-is, or constructed from current state of curve cache. Arguments to apply modifiers and calculate original coordinates (ORCO, aka undeformed coordinates) are removed. The ORCO is to be calculated as part of dependency graph evaluation. File used to regression-test (a packed Python script into .blend): {F7033464} Patch to make addons tests to pass: {F7033466} NOTE: I've included changes to FBX exporter, and those are addressing report T63689. NOTE: All the enabled-by-default addons are to be ported still, but first want to have agreement on this part of changes. NOTE: Also need to work on documentation for Python API, but, again, better be done after having agreement on this work. Reviewers: brecht, campbellbarton, mont29 Differential Revision: https://developer.blender.org/D4834
2019-05-09 11:26:49 +02:00
plane_eval = plane.evaluated_get(depsgraph)
mesh = plane_eval.to_mesh()
self.assertAlmostEqual(1, mesh.vertices[3].co.x)
self.assertAlmostEqual(1, mesh.vertices[3].co.y)
self.assertAlmostEqual(0.5905638933181763, mesh.vertices[3].co.z)
plane_eval.to_mesh_clear()
def test_import_long_names(self):
# This file contains very long names. The longest name is 4047 chars.
bpy.ops.wm.alembic_import(
filepath=str(self.testdir / "long-names.abc"),
as_background_job=False)
self.assertIn('Cube', bpy.data.objects)
self.assertEqual('CubeShape', bpy.data.objects['Cube'].data.name)
class VertexColourImportTest(AbstractAlembicTest):
def test_import_from_houdini(self):
# Houdini saved "face-varying", and as RGB.
res = bpy.ops.wm.alembic_import(
filepath=str(self.testdir / "vertex-colours-houdini.abc"),
as_background_job=False)
self.assertEqual({'FINISHED'}, res)
ob = bpy.context.active_object
layer = ob.data.vertex_colors['Cf'] # MeshLoopColorLayer
# Test some known-good values.
self.assertAlmostEqualFloatArray(layer.data[0].color, (0, 0, 0, 1.0))
self.assertAlmostEqualFloatArray(layer.data[98].color, (0.9019607, 0.4745098, 0.2666666, 1.0))
self.assertAlmostEqualFloatArray(layer.data[99].color, (0.8941176, 0.4705882, 0.2627451, 1.0))
def test_import_from_blender(self):
# Blender saved per-vertex, and as RGBA.
res = bpy.ops.wm.alembic_import(
filepath=str(self.testdir / "vertex-colours-blender.abc"),
as_background_job=False)
self.assertEqual({'FINISHED'}, res)
ob = bpy.context.active_object
layer = ob.data.vertex_colors['Cf'] # MeshLoopColorLayer
# Test some known-good values.
self.assertAlmostEqualFloatArray(layer.data[0].color, (1.0, 0.0156862, 0.3607843, 1.0))
self.assertAlmostEqualFloatArray(layer.data[98].color, (0.0941176, 0.1215686, 0.9137254, 1.0))
self.assertAlmostEqualFloatArray(layer.data[99].color, (0.1294117, 0.3529411, 0.7529411, 1.0))
class CameraExportImportTest(unittest.TestCase):
names = [
'CAM_Unit_Transform',
'CAM_Look_+Y',
'CAM_Static_Child_Left',
'CAM_Static_Child_Right',
'Static_Child',
'CAM_Animated',
'CAM_Animated_Child_Left',
'CAM_Animated_Child_Right',
'Animated_Child',
]
def setUp(self):
self._tempdir = tempfile.TemporaryDirectory()
self.tempdir = pathlib.Path(self._tempdir.name)
def tearDown(self):
# Unload the current blend file to release the imported Alembic file.
# This is necessary on Windows in order to be able to delete the
# temporary ABC file.
bpy.ops.wm.read_homefile(use_empty=True, use_factory_startup=True)
self._tempdir.cleanup()
def test_export_hierarchy(self):
self.do_export_import_test(flatten=False)
# Double-check that the export was hierarchical.
objects = bpy.context.scene.collection.objects
for name in self.names:
if 'Child' in name:
self.assertIsNotNone(objects[name].parent)
else:
self.assertIsNone(objects[name].parent)
def test_export_flattened(self):
self.do_export_import_test(flatten=True)
# Double-check that the export was flat.
objects = bpy.context.scene.collection.objects
for name in self.names:
self.assertIsNone(objects[name].parent)
def do_export_import_test(self, *, flatten: bool):
bpy.ops.wm.open_mainfile(filepath=str(args.testdir / "camera_transforms.blend"))
abc_path = self.tempdir / "camera_transforms.abc"
self.assertIn('FINISHED', bpy.ops.wm.alembic_export(
filepath=str(abc_path),
flatten=flatten,
))
# Re-import what we just exported into an empty file.
bpy.ops.wm.open_mainfile(filepath=str(args.testdir / "empty.blend"))
self.assertIn('FINISHED', bpy.ops.wm.alembic_import(filepath=str(abc_path)))
# Test that the import was ok.
bpy.context.scene.frame_set(1)
self.loc_rot_scale('CAM_Unit_Transform', (0, 0, 0), (0, 0, 0))
self.loc_rot_scale('CAM_Look_+Y', (2, 0, 0), (90, 0, 0))
self.loc_rot_scale('CAM_Static_Child_Left', (2 - 0.15, 0, 0), (90, 0, 0))
self.loc_rot_scale('CAM_Static_Child_Right', (2 + 0.15, 0, 0), (90, 0, 0))
self.loc_rot_scale('Static_Child', (2, 0, 1), (90, 0, 0))
self.loc_rot_scale('CAM_Animated', (4, 0, 0), (90, 0, 0))
self.loc_rot_scale('CAM_Animated_Child_Left', (4 - 0.15, 0, 0), (90, 0, 0))
self.loc_rot_scale('CAM_Animated_Child_Right', (4 + 0.15, 0, 0), (90, 0, 0))
self.loc_rot_scale('Animated_Child', (4, 0, 1), (90, 0, 0))
bpy.context.scene.frame_set(10)
self.loc_rot_scale('CAM_Animated', (4, 1, 2), (90, 0, 25))
self.loc_rot_scale('CAM_Animated_Child_Left', (3.864053, 0.936607, 2), (90, 0, 25))
self.loc_rot_scale('CAM_Animated_Child_Right', (4.135946, 1.063392, 2), (90, 0, 25))
self.loc_rot_scale('Animated_Child', (4, 1, 3), (90, -45, 25))
def loc_rot_scale(self, name: str, expect_loc, expect_rot_deg):
"""Assert world loc/rot/scale is OK."""
objects = bpy.context.scene.collection.objects
depsgraph = bpy.context.evaluated_depsgraph_get()
ob_eval = objects[name].evaluated_get(depsgraph)
actual_loc = ob_eval.matrix_world.to_translation()
actual_rot = ob_eval.matrix_world.to_euler('XYZ')
actual_scale = ob_eval.matrix_world.to_scale()
# Precision of the 'almost equal' comparisons.
delta_loc = delta_scale = 1e-6
delta_rot = math.degrees(1e-6)
self.assertAlmostEqual(expect_loc[0], actual_loc.x, delta=delta_loc)
self.assertAlmostEqual(expect_loc[1], actual_loc.y, delta=delta_loc)
self.assertAlmostEqual(expect_loc[2], actual_loc.z, delta=delta_loc)
self.assertAlmostEqual(expect_rot_deg[0], math.degrees(actual_rot.x), delta=delta_rot)
self.assertAlmostEqual(expect_rot_deg[1], math.degrees(actual_rot.y), delta=delta_rot)
self.assertAlmostEqual(expect_rot_deg[2], math.degrees(actual_rot.z), delta=delta_rot)
# This test doesn't use scale.
self.assertAlmostEqual(1, actual_scale.x, delta=delta_scale)
self.assertAlmostEqual(1, actual_scale.y, delta=delta_scale)
self.assertAlmostEqual(1, actual_scale.z, delta=delta_scale)
class OverrideLayersTest(AbstractAlembicTest):
def test_import_layer(self):
fname = 'cube-base-file.abc'
fname_layer = 'cube-hi-res.abc'
abc = self.testdir / fname
abc_layer = self.testdir / fname_layer
# We need a cache reader to ensure that the data will be updated after adding a layer.
res = bpy.ops.wm.alembic_import(filepath=str(abc), as_background_job=False, always_add_cache_reader=True)
self.assertEqual({'FINISHED'}, res)
# Check that the file loaded ok.
cube = bpy.context.active_object
depsgraph = bpy.context.evaluated_depsgraph_get()
scene = bpy.context.scene
cube_eval = cube.evaluated_get(depsgraph)
mesh = cube_eval.to_mesh()
# The base file should be a default cube.
self.assertEqual(len(mesh.vertices), 8)
self.assertEqual(len(mesh.edges), 12)
self.assertEqual(len(mesh.polygons), 6)
# Add a layer.
cache_file = bpy.data.cache_files[fname]
self.assertEqual(len(cache_file.layers), 0)
layer = cache_file.layers.new(filepath=str(abc_layer))
self.assertEqual(len(cache_file.layers), 1)
self.assertIsNotNone(layer)
# The layer added a higher res version of the mesh.
depsgraph = bpy.context.evaluated_depsgraph_get()
cube_eval = cube.evaluated_get(depsgraph)
mesh = cube_eval.to_mesh()
self.assertEqual(len(mesh.vertices), 26)
self.assertEqual(len(mesh.edges), 48)
self.assertEqual(len(mesh.polygons), 24)
# Remove the layer.
cache_file.layers.remove(layer)
self.assertEqual(len(cache_file.layers), 0)
# We should have reverted to the default cube.
depsgraph = bpy.context.evaluated_depsgraph_get()
cube_eval = cube.evaluated_get(depsgraph)
mesh = cube_eval.to_mesh()
self.assertEqual(len(mesh.vertices), 8)
self.assertEqual(len(mesh.edges), 12)
self.assertEqual(len(mesh.polygons), 6)
def main():
global args
import argparse
if '--' in sys.argv:
2017-06-12 05:35:00 +02:00
argv = [sys.argv[0]] + sys.argv[sys.argv.index('--') + 1:]
else:
argv = sys.argv
parser = argparse.ArgumentParser()
parser.add_argument('--testdir', required=True, type=pathlib.Path)
args, remaining = parser.parse_known_args(argv)
unittest.main(argv=remaining)
if __name__ == "__main__":
main()