fix: integer cast issue for colored projector patterns

[beekama]

When calling pattern.ravel(), there is an issue with integer casting, which results in the array values being cast from floats to integers. As a result, the end values are either 0 or 1, leading to the loss of color information and, consequently, the creation of an incorrect result image with only the color values cyan, yellow, and magenta.

[cla-bot]

For contributing to BlenderProc you need to sign our Contributor License Agreement. As an individual please sign CLA_individuals.pdf, as a company please sign CLA_entities.pdf and send it to blenderproc@dlr.de

[cornerfarmer]

BlenderProcHelper result for run 541:

Done with: [amass_human_poses_python 43s, auto_shading_python 10s, basic_python 11s, blenderkit_python 19s, bop_object_on_surface_sampling_python 79s, bop_object_physics_positioning_python 105s, bop_object_pose_sampling_python 42s, bop_scene_replication_python 166s, camera_depth_of_field_python 18s, camera_object_pose_python 16s, camera_random_trajectories_python 50s, camera_sampling_python 20s, coco_annotations_python 14s, diffuse_color_image_python 18s, dust_python 29s, entity_displacement_modifier_python 14s, entity_manipulation_python 11s, front_3d_object_sampling_python 83s, front_3d_python 98s, front_3d_with_improved_mat_python 102s, git_log 1s, haven_python 23s, ikea_python 18s, kinect_azure_noise_python 14s, lens_distortion_python 15s, lens_distortion_python 13s, light_sampling_python 12s, material_manipulation_python 10s, material_randomizer_python 13s, matterport3d_python 69s, motion_blur_rolling_shutter_python 31s, motion_blur_rolling_shutter_python 21s, multi_render_python 245s, nocs_python 18s, object_pose_sampling_python 11s, on_surface_object_sampling_python 11s, optical_flow_python 25s, physics_convex_decomposition_python 133s, physics_positioning_python 24s, pix3d_python 17s, random_backgrounds_python 17s, random_room_constructor_python 75s, replica_python 337s, scenenet_python 51s, scenenet_with_cctextures_python 53s, semantic_segmentation_python 14s, shapenet_python 19s, shapenet_with_scenenet_python 52s, shapenet_with_suncg_python 68s, spotlight_python 46s, stereo_matching_python 42s, stereo_matching_with_projector_python 136s, suncg_basic_python 32s, suncg_with_cam_sampling_python 56s, suncg_with_improved_mat_python 74s, suncg_with_object_replacer_python 60s, tests_python 15s, urdf_loading_and_manipulation_python 50s]
Done with 58 of 58 tests.
There were errors: ❌
Unit tests output:

${INTERNAL_NODES_DISC_SPACE}/Dominik_Winkelbauer/blender/blender-3.5.1-linux-x64/3.5/python/lib/python3.10/site-packages/numpy/core/getlimits.py:500: UserWarning: The value of the smallest subnormal for <class 'numpy.float64'> type is zero.
  setattr(self, word, getattr(machar, word).flat[0])
${INTERNAL_NODES_DISC_SPACE}/Dominik_Winkelbauer/blender/blender-3.5.1-linux-x64/3.5/python/lib/python3.10/site-packages/numpy/core/getlimits.py:89: UserWarning: The value of the smallest subnormal for <class 'numpy.float64'> type is zero.
  return self._float_to_str(self.smallest_subnormal)
${INTERNAL_NODES_DISC_SPACE}/Dominik_Winkelbauer/blender/blender-3.5.1-linux-x64/3.5/python/lib/python3.10/site-packages/numpy/core/getlimits.py:500: UserWarning: The value of the smallest subnormal for <class 'numpy.float32'> type is zero.
  setattr(self, word, getattr(machar, word).flat[0])
${INTERNAL_NODES_DISC_SPACE}/Dominik_Winkelbauer/blender/blender-3.5.1-linux-x64/3.5/python/lib/python3.10/site-packages/numpy/core/getlimits.py:89: UserWarning: The value of the smallest subnormal for <class 'numpy.float32'> type is zero.
  return self._float_to_str(self.smallest_subnormal)
...........
----------------------------------------------------------------------
Ran 11 tests in 9.293s

OK

urdf_loading_and_manipulation (main_python)

The output of urdf_loading_and_manipulation:main_python:000000_scene_gt_info_json has changed:

Expand

+ root['2'][3]['px_count_visib']: 2025
- root['2'][3]['px_count_visib']: 2038
+ root['3'][3]['px_count_visib']: 1898
- root['3'][3]['px_count_visib']: 1909
+ root['4'][6]['bbox_visib'][3]: 17
- root['4'][6]['bbox_visib'][3]: 16

The output of urdf_loading_and_manipulation:main_python:000000_scene_gt_coco_json has changed:

Expand

+ root['annotations'][19]['area']: 2025
- root['annotations'][19]['area']: 2038
+ root['annotations'][27]['area']: 1898
- root['annotations'][27]['area']: 1909

stereo_matching (main_python)

The output of stereo_matching:main_python:disparity has changed: Diff: 0.0320, Pixel diff: 0.0047

Compare image	Newly generated image

The output of stereo_matching:main_python:disparity has changed: Diff: 0.0224, Pixel diff: 0.0097

Compare image	Newly generated image

Pylint output

Expand

Messages

Module blenderproc.python.loader.CCMaterialLoader (blenderproc/python/loader/CCMaterialLoader.py)

Line	Col.	Type	Symbol	ID	Obj	Message
72	0	convention	line-too-long	C0301		Line too long (152/120)

Module blenderproc.python.types.LightUtility (blenderproc/python/types/LightUtility.py)

Line	Col.	Type	Symbol	ID	Obj	Message
125	12	refactor	consider-using-augmented-assign	R6104	Light.setup_as_projector	Use '/=' to do an augmented assign directly
216	0	convention	trailing-whitespace	C0303		Trailing whitespace

Module blenderproc.python.writer.BopWriterUtility (blenderproc/python/writer/BopWriterUtility.py)

Line	Col.	Type	Symbol	ID	Obj	Message
15	0	convention	wrong-import-order	C0411		standard import "import sys" should be placed before "import numpy as np"

Metrics

Count per types

Name	Count
convention	3
refactor	1

Count per messages

Name	Count
line-too-long	1
trailing-whitespace	1
consider-using-augmented-assign	1
wrong-import-order	1

Count per modules

Name	Count
blenderproc.python.loader.CCMaterialLoader	1
blenderproc.python.types.LightUtility	2
blenderproc.python.writer.BopWriterUtility	1

Count per path

Name	Count
blenderproc/python/loader/CCMaterialLoader.py	1
blenderproc/python/types/LightUtility.py	2
blenderproc/python/writer/BopWriterUtility.py	1

***The tests are done.*** Check the log:

Last update: 10/24/2023, 17:33:28

[cornerfarmer]

Thank you!