cos method ready : version 3

Author: maximeraafat

.gitignore

@@ -1,2 +1,2 @@
 .DS_Store
-testing.py
+__pycache__

CITATION.cff

@@ -8,4 +8,4 @@ title: "Blender x NeRF"
 version: 2.0.0
 doi: 10.5281/zenodo.6862485
 date-released: 2022-07-19
-url: "https://github.com/maximeraafat/BlenderNeRF"
+url: "https://github.com/maximeraafat/BlenderNeRF"

LICENSE

@@ -18,4 +18,4 @@ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+SOFTWARE.

__init__.py

@@ -1,99 +1,80 @@
-import os
-import shutil
 import bpy
-from bpy.app.handlers import persistent
-from . import blender_nerf_ui, sof_ui, ttc_ui, sof_operator, ttc_operator
+from . import helper, blender_nerf_ui, sof_ui, ttc_ui, cos_ui, sof_operator, ttc_operator, cos_operator
 
 
+# blender info
 bl_info = {
-    'name': 'Blender x NeRF',
-    'description': 'Simple and quick NeRF dataset creation tool',
+    'name': 'BlenderNeRF',
+    'description': 'Easy NeRF synthetic dataset creation within Blender',
     'author': 'Maxime Raafat',
-    'version': (2, 0, 0),
+    'version': (3, 0, 0),
     'blender': (3, 0, 0),
     'location': '3D View > N panel > BlenderNeRF',
     'doc_url': 'https://github.com/maximeraafat/BlenderNeRF',
     'category': 'Object',
 }
 
-
-# camera pointer property poll function
-def poll_is_camera(self, obj):
-    return obj.type == 'CAMERA'
-
 # global addon script variables
 TRAIN_CAM = 'Train Cam'
 TEST_CAM = 'Test Cam'
 
 # addon blender properties
 PROPS = [
-    # global manually defined properties
-    ('train_data', bpy.props.BoolProperty(name='Train', description='Construct training data', default=True) ),
-    ('test_data', bpy.props.BoolProperty(name='Test', description='Construct testing data', default=True) ),
-    ('aabb', bpy.props.IntProperty(name='AABB', description='AABB scale as defined in Instant NGPs NeRF version', default=4, soft_min=1, soft_max=128) ),
-    ('render_frames', bpy.props.BoolProperty(name='Render Frames', description='Whether the training frames for NeRF should be rendered or not, in which case only the transforms.json files will be generated', default=True) ),
-    ('save_path', bpy.props.StringProperty(name='Save Path', description='Path to the output directory in which the dataset will be stored', subtype='DIR_PATH') ),
-
-    # global automatically defined properties
+    # global controllable properties
+    ('train_data', bpy.props.BoolProperty(name='Train', description='Construct the training data', default=True) ),
+    ('test_data', bpy.props.BoolProperty(name='Test', description='Construct the testing data', default=True) ),
+    ('aabb', bpy.props.IntProperty(name='AABB', description='AABB scale as defined in Instant NGP', default=4, soft_min=1, soft_max=128) ),
+    ('render_frames', bpy.props.BoolProperty(name='Render Frames', description='Whether training frames should be rendered. If not selected, only the transforms.json files will be generated', default=True) ),
+    ('save_path', bpy.props.StringProperty(name='Save Path', description='Path to the output directory in which the synthetic dataset will be stored', subtype='DIR_PATH') ),
+
+    # global automatic properties
     ('init_frame_step', bpy.props.IntProperty(name='Initial Frame Step') ),
     ('init_output_path', bpy.props.StringProperty(name='Initial Output Path', subtype='DIR_PATH') ),
-    ('is_rendering', bpy.props.BoolVectorProperty(name='Method Rendering?', description='Whether SOF, TTC or COS is rendering', default=(False, False, False), size=3) ),
+    ('rendering', bpy.props.BoolVectorProperty(name='Rendering', description='Whether one of the SOF, TTC or COS methods is rendering', default=(False, False, False), size=3) ),
 
     # sof properties
     ('sof_dataset_name', bpy.props.StringProperty(name='Name', description='Name of the SOF dataset : the data will be stored under <save path>/<name>', default='dataset') ),
-    ('train_frame_steps', bpy.props.IntProperty(name='Frame Step', description='Number of frames to skip forward for the NeRF training dataset created while rendering/playing back each frame', default=3, soft_min=1) ),
+    ('train_frame_steps', bpy.props.IntProperty(name='Frame Step', description='Frame step N for the captured training frames. Every N-th frame will be used for training NeRF', default=3, soft_min=1) ),
 
     # ttc properties
     ('ttc_dataset_name', bpy.props.StringProperty(name='Name', description='Name of the TTC dataset : the data will be stored under <save path>/<name>', default='dataset') ),
-    ('camera_train_target', bpy.props.PointerProperty(type=bpy.types.Object, name=TRAIN_CAM, description='Pointer to training camera', poll=poll_is_camera) ),
-    ('camera_test_target', bpy.props.PointerProperty(type=bpy.types.Object, name=TEST_CAM, description='Pointer to testing camera', poll=poll_is_camera) ),
+    ('camera_train_target', bpy.props.PointerProperty(type=bpy.types.Object, name=TRAIN_CAM, description='Pointer to the training camera', poll=helper.poll_is_camera) ),
+    ('camera_test_target', bpy.props.PointerProperty(type=bpy.types.Object, name=TEST_CAM, description='Pointer to the testing camera', poll=helper.poll_is_camera) ),
+
+    # cos controllable properties
+    ('cos_dataset_name', bpy.props.StringProperty(name='Name', description='Name of the COS dataset : the data will be stored under <save path>/<name>', default='dataset') ),
+    ('sphere_location', bpy.props.FloatVectorProperty(name='Location', description='Center position of the training sphere', unit='LENGTH', update=helper.properties_ui_upd) ),
+    ('sphere_rotation', bpy.props.FloatVectorProperty(name='Rotation', description='Rotation of the training sphere', unit='ROTATION', update=helper.properties_ui_upd) ),
+    ('sphere_scale', bpy.props.FloatVectorProperty(name='Scale', description='Scale of the training sphere in xyz axes', default=(1.0, 1.0, 1.0), update=helper.properties_ui_upd) ),
+    ('sphere_radius', bpy.props.FloatProperty(name='Radius', description='Radius scale of the training sphere', default=4.0, soft_min=0.01, unit='LENGTH', update=helper.properties_ui_upd) ),
+    ('focal', bpy.props.FloatProperty(name='Lens', description='Focal length of the training camera', default=50, soft_min=1, soft_max=5000, unit='CAMERA', update=helper.properties_ui_upd) ),
+    ('seed', bpy.props.IntProperty(name='Seed', description='Random seed for sampling views on the training sphere', default=0) ),
+    ('nb_frames', bpy.props.IntProperty(name='Frames', description='Number of training frames randomly sampled from the training sphere', default=100, soft_min=1) ),
+    ('show_sphere', bpy.props.BoolProperty(name='Sphere', description='Whether to show the training sphere from which random views will be sampled', default=False, update=helper.visualize_sphere) ),
+    ('show_camera', bpy.props.BoolProperty(name='Camera', description='Whether to show the training camera', default=False, update=helper.visualize_camera) ),
+    ('upper_views', bpy.props.BoolProperty(name='Upper Views', description='Whether to sample views from the upper hemisphere of the training sphere only', default=False) ),
+
+    # cos automatic properties
+    ('sphere_exists', bpy.props.BoolProperty(name='Sphere Exists', description='Whether the sphere exists', default=False) ),
+    ('init_sphere_exists', bpy.props.BoolProperty(name='Init sphere exists', description='Whether the sphere initially exists', default=False) ),
+    ('camera_exists', bpy.props.BoolProperty(name='Camera Exists', description='Whether the camera exists', default=False) ),
+    ('init_camera_exists', bpy.props.BoolProperty(name='Init camera exists', description='Whether the camera initially exists', default=False) ),
+    ('init_active_camera', bpy.props.PointerProperty(type=bpy.types.Object, name='Init active camera', description='Pointer to initial active camera', poll=helper.poll_is_camera) ),
+    ('init_frame_start', bpy.props.IntProperty(name='Initial Frame Start') ),
+    ('init_frame_end', bpy.props.IntProperty(name='Initial Frame End') ),
 ]
 
 # classes to register / unregister
 CLASSES = [
     blender_nerf_ui.BlenderNeRF_UI,
     sof_ui.SOF_UI,
     ttc_ui.TTC_UI,
+    cos_ui.COS_UI,
     sof_operator.SubsetOfFrames,
-    ttc_operator.TrainTestCameras
+    ttc_operator.TrainTestCameras,
+    cos_operator.CameraOnSphere
 ]
 
-
-## blender handler functions
-
-# set frame step and filepath back to initial
-@persistent
-def post_render(scene):
-    if any(scene.is_rendering): # execute this function only when rendering with addon
-        scene.is_rendering = (False, False, False)
-        scene.frame_step = scene.init_frame_step
-        scene.render.filepath = scene.init_output_path
-
-        method_dataset_name = scene.sof_dataset_name if scene.is_rendering[0] else scene.ttc_dataset_name
-
-        # clean directory name (unsupported characters replaced) and output path
-        output_dir = bpy.path.clean_name(method_dataset_name)
-        output_path = os.path.join(scene.save_path, output_dir)
-
-        # compress dataset and remove folder (only keep zip)
-        shutil.make_archive(output_path, 'zip', output_path) # output filename = output_path
-        shutil.rmtree(output_path)
-
-# set initial property values (bpy.data and bpy.context require a loaded scene)
-@persistent
-def set_init_props(scene):
-    scene.init_frame_step = scene.frame_step
-    scene.init_output_path = scene.render.filepath
-
-    # set save_path to path in which blender file is stored
-    # filepath = bpy.data.filepath
-    # filename = bpy.path.basename(filepath)
-    # default_save_path = filepath[:-len(filename)] # remove file name from blender file path = directoy path
-    # scene.save_path = default_save_path
-
-    bpy.app.handlers.depsgraph_update_post.remove(set_init_props)
-
-
 # load addon
 def register():
     for (prop_name, prop_value) in PROPS:
@@ -102,22 +83,26 @@ def register():
     for cls in CLASSES:
         bpy.utils.register_class(cls)
 
-    bpy.app.handlers.render_complete.append(post_render)
-    bpy.app.handlers.render_cancel.append(post_render)
-    bpy.app.handlers.depsgraph_update_post.append(set_init_props)
+    bpy.app.handlers.render_complete.append(helper.post_render)
+    bpy.app.handlers.render_cancel.append(helper.post_render)
+    bpy.app.handlers.frame_change_post.append(helper.cos_camera_update)
+    bpy.app.handlers.depsgraph_update_post.append(helper.properties_desgraph_upd)
+    bpy.app.handlers.depsgraph_update_post.append(helper.set_init_props)
 
 # deregister addon
 def unregister():
     for (prop_name, _) in PROPS:
         delattr(bpy.types.Scene, prop_name)
 
-    bpy.app.handlers.render_complete.remove(post_render)
-    bpy.app.handlers.render_cancel.remove(post_render)
-    # bpy.app.handlers.depsgraph_update_post.remove(set_init_props)
+    bpy.app.handlers.render_complete.remove(helper.post_render)
+    bpy.app.handlers.render_cancel.remove(helper.post_render)
+    bpy.app.handlers.frame_change_post.remove(helper.cos_camera_update)
+    bpy.app.handlers.depsgraph_update_post.remove(helper.properties_desgraph_upd)
+    # bpy.app.handlers.depsgraph_update_post.remove(helper.set_init_props)
 
     for cls in CLASSES:
         bpy.utils.unregister_class(cls)
 
 
 if __name__ == '__main__':
-    register()
+    register()

blender_nerf_operator.py

@@ -9,20 +9,11 @@
 OUTPUT_TEST = 'images_test'
 
 
-# helper function
-def listify_matrix(matrix):
-    matrix_list = []
-    for row in matrix:
-        matrix_list.append(list(row))
-    return matrix_list
-
-
 # blender nerf operator parent class
 class BlenderNeRF_Operator(bpy.types.Operator):
 
-    # camera intrinsics and other relevant camera data
-    # https://blender.stackexchange.com/questions/38009/3x4-camera-matrix-from-blender-camera
-    def get_camera_intrinsics(self, scene, camera):
+    # camera intrinsics : https://blender.stackexchange.com/questions/38009/3x4-camera-matrix-from-blender-camera
+    def get_camera_intrinsics(self, scene,camera):
         camera_angle_x = camera.data.angle_x
         camera_angle_y = camera.data.angle_y
 
@@ -76,16 +67,18 @@ def get_camera_extrinsics(self, scene, camera, mode='TRAIN', method='SOF'):
 
         initFrame = scene.frame_current
         step = scene.train_frame_steps if (mode == 'TRAIN' and method == 'SOF') else scene.frame_step
+        start = 1 if (mode == 'TRAIN' and method == 'COS') else scene.frame_start
+        end = scene.nb_frames if (mode == 'TRAIN' and method == 'COS') else scene.frame_end
 
         camera_extr_dict = []
-        for frame in range(scene.frame_start, scene.frame_end + 1, step):
+        for frame in range(start, end + 1, step):
             scene.frame_set(frame)
             filename = os.path.basename( scene.render.frame_path(frame=frame) )
             filedir = OUTPUT_TRAIN * (mode == 'TRAIN') + OUTPUT_TEST * (mode == 'TEST')
 
             frame_data = {
                 'file_path': os.path.join(filedir, filename),
-                'transform_matrix': listify_matrix( camera.matrix_world )
+                'transform_matrix': self.listify_matrix( camera.matrix_world )
             }
 
             camera_extr_dict.append(frame_data)
@@ -102,30 +95,43 @@ def save_json(self, directory, filename, data, indent=4):
     def is_power_of_two(self, x):
         return math.log2(x).is_integer()
 
+    # function from original nerf 360_view.py code for blender
+    def listify_matrix(self, matrix):
+        matrix_list = []
+        for row in matrix:
+            matrix_list.append(list(row))
+        return matrix_list
+
+    # assert messages
     def asserts(self, scene, method='SOF'):
         assert method == 'SOF' or method == 'TTC' or method == 'COS'
 
         camera = scene.camera
         train_camera = scene.camera_train_target
         test_camera = scene.camera_test_target
 
+        sof_name = scene.sof_dataset_name
+        ttc_name = scene.ttc_dataset_name
+        cos_name = scene.cos_dataset_name
+
         error_messages = []
 
-        if method == 'SOF':
-            if not camera.data.type == 'PERSP':
-                error_messages.append('Only perspective cameras are supported!')
-            if scene.sof_dataset_name == '':
-                error_messages.append('Dataset name cannot be empty!')
-        elif method == 'TTC':
-            if not (train_camera.data.type == 'PERSP' and test_camera.data.type == 'PERSP'):
-                error_messages.append('Only perspective cameras are supported!')
-            if scene.ttc_dataset_name == '':
-                error_messages.append('Dataset name cannot be empty!')
+        if (method == 'SOF' or method == 'COS') and not camera.data.type == 'PERSP':
+            error_messages.append('Only perspective cameras are supported!')
+
+        if method == 'TTC' and not (train_camera.data.type == 'PERSP' and test_camera.data.type == 'PERSP'):
+           error_messages.append('Only perspective cameras are supported!')
+
+        if (method == 'SOF' and sof_name == '') or (method == 'TTC' and ttc_name == '') or (method == 'COS' and cos_name == ''):
+            error_messages.append('Dataset name cannot be empty!')
+
+        if method == 'COS' and any(x == 0 for x in scene.sphere_scale):
+            error_messages.append('The BlenderNeRF Sphere cannot be flat! Change its scale to be non zero in all axes.')
 
         if not self.is_power_of_two(scene.aabb):
             error_messages.append('AABB scale needs to be a power of two!')
 
         if scene.save_path == '':
             error_messages.append('Save path cannot be empty!')
 
-        return error_messages
+        return error_messages

blender_nerf_ui.py

@@ -3,9 +3,9 @@
 
 # blender nerf shared ui properties class
 class BlenderNeRF_UI(bpy.types.Panel):
-    '''Blender x NeRF UI'''
+    '''BlenderNeRF UI'''
     bl_idname = 'panel.blender_nerf_ui'
-    bl_label = 'Blender x NeRF shared UI'
+    bl_label = 'BlenderNeRF shared UI'
     bl_space_type = 'VIEW_3D'
     bl_region_type = 'UI'
     bl_category = 'BlenderNeRF'
@@ -15,7 +15,7 @@ def draw(self, context):
         scene = context.scene
 
         layout.alignment = 'CENTER'
-        row = layout.row()
+        row = layout.row(align=True)
 
         row.prop(scene, 'train_data', toggle=True)
         row.prop(scene, 'test_data', toggle=True)
@@ -32,4 +32,4 @@ def draw(self, context):
 
             layout.separator()
             layout.use_property_split = True
-            layout.prop(scene, 'save_path')
+            layout.prop(scene, 'save_path')