mirror of
https://github.com/alicevision/Meshroom.git
synced 2025-05-01 19:27:10 +02:00
f8141796f3
Added the possibility of rendering the output of the meshing node into a edge detection render of the obj. Added the activation and deactivation of the background images as an option. Improving the way the arguments are shown with a conditionnal display of some arguments.
405 lines
No EOL
21 KiB
Python
405 lines
No EOL
21 KiB
Python
import bpy
|
|
import bmesh
|
|
import os
|
|
import re
|
|
import mathutils
|
|
import sys # to get command line args
|
|
import argparse # to parse options for us and print a nice help message
|
|
from distutils.util import strtobool
|
|
|
|
def main():
|
|
argv = sys.argv
|
|
|
|
if "--" not in argv:
|
|
argv = [] # as if no args are passed
|
|
else:
|
|
argv = argv[argv.index("--") + 1:] # get all args after "--"
|
|
|
|
# When --help or no args are given, print this help
|
|
usage_text = (
|
|
"Run blender in background mode with this script:"
|
|
" blender --background --python " + __file__ + " -- [options]"
|
|
)
|
|
|
|
parser = argparse.ArgumentParser(description=usage_text)
|
|
|
|
|
|
parser.add_argument(
|
|
"--sfMCameraPath", dest="SFM_cam_path", metavar='FILE', required=True,
|
|
help="This text will be used to render an image",
|
|
)
|
|
|
|
parser.add_argument(
|
|
"--useBackground", dest="Use_Background", type=strtobool, required=True,
|
|
help="Diplay the background image or not.",
|
|
)
|
|
|
|
parser.add_argument(
|
|
"--undistortedImages", dest="undisto_images", metavar='FILE', required=False,
|
|
help="Save the generated file to the specified path",
|
|
)
|
|
|
|
parser.add_argument(
|
|
"--sfMData", dest="SFM_Data", metavar='FILE', required=True,
|
|
help="These info carry the Point Cloud or mesh we need.",
|
|
)
|
|
|
|
#Point Cloud Arguments (When SFM Data is .abc)
|
|
|
|
parser.add_argument(
|
|
"--pointCloudDensity", dest="Point_Cloud_Density", type=float, required=False,
|
|
help="Number of point from the cloud rendered",
|
|
)
|
|
|
|
parser.add_argument(
|
|
"--particleSize", dest="Particle_Size", type=float, required=False,
|
|
help="Scale of every particle used to show the point cloud",
|
|
)
|
|
|
|
parser.add_argument(
|
|
"--particleColor", dest="Particle_Color", type=str, required=False,
|
|
help="Color of every particle used to show the point cloud (SFM Data is .abc)",
|
|
)
|
|
|
|
#Mesh Arguments (When SFM Data is .obj)
|
|
|
|
parser.add_argument(
|
|
"--edgeColor", dest="Edge_Color", type=str, required=False,
|
|
help="Color of the edges of the rendered object (SFM Data is .obj)",
|
|
)
|
|
|
|
|
|
#Output Arguments
|
|
|
|
parser.add_argument(
|
|
"--outputFormat", dest="Output_Format", type=str, required=True,
|
|
help="Format of the video output",
|
|
)
|
|
|
|
parser.add_argument(
|
|
"--outputPath", dest="output_path", metavar='FILE', required=True,
|
|
help="Render an image to the specified path",
|
|
)
|
|
|
|
|
|
args = parser.parse_args(argv)
|
|
|
|
if not argv:
|
|
parser.print_help()
|
|
return
|
|
|
|
if not args.undisto_images and args.Use_Background :
|
|
print("Error: --undisto_images argument not given, aborting.")
|
|
parser.print_help()
|
|
return
|
|
|
|
if not args.Point_Cloud_Density and args.SFM_Data.endswith('.abc'):
|
|
print("Error: --Point_Cloud_Density argument not given, aborting.")
|
|
parser.print_help()
|
|
return
|
|
|
|
if not args.Particle_Size and args.SFM_Data.endswith('.abc'):
|
|
print("Error: --Particle_Size argument not given, aborting.")
|
|
parser.print_help()
|
|
return
|
|
|
|
if not args.Particle_Color and args.SFM_Data.endswith('.abc'):
|
|
print("Error: --Particle_Color argument not given, aborting.")
|
|
parser.print_help()
|
|
return
|
|
|
|
if not args.Edge_Color and args.SFM_Data.endswith('.obj'):
|
|
print("Error: --Edge_Color argument not given, aborting.")
|
|
parser.print_help()
|
|
return
|
|
|
|
#Clear Current Scene
|
|
try:
|
|
for objects in bpy.data.objects:
|
|
bpy.data.objects.remove(objects)
|
|
except RuntimeError:
|
|
print("Error: While clearing current scene")
|
|
|
|
#The Switcher is the setting for most of the colors (if you want to add some, do it here and in the arguments of the node)
|
|
# Keep in mind that we use the same switcher for both the Edge Color and the Particle Color settings.
|
|
# So if you add a color to one of them in the node, might has well add it to the other.
|
|
|
|
switcher={
|
|
'Grey':(0.2, 0.2, 0.2, 1),
|
|
'White':(1, 1, 1, 1),
|
|
'Red':(0.5, 0, 0, 1),
|
|
'Green':(0, 0.5, 0, 1),
|
|
'Magenta':(1.0, 0, 0.75, 1)
|
|
}
|
|
|
|
# import Undistorted Images
|
|
|
|
undis_imgs = []
|
|
#Some of these variable will be very useful in the next steps keep them in mind
|
|
number_of_frame = 0
|
|
offset = 0
|
|
first_image_name = ""
|
|
try:
|
|
# In this part of the code we take the undistorted images and we process some info about them
|
|
# undis_imgs is the list of the images' names
|
|
# first_image_name says it all in the name
|
|
# The offset is important, it corresponds to the last part of the name of the first frame
|
|
# In most case it will hopefully be 0 but the sequence may start from a more advanced frame
|
|
if args.Use_Background :
|
|
files = os.listdir(args.undisto_images)
|
|
for f in files :
|
|
if f.endswith(".exr") and not f.__contains__("UVMap"):
|
|
undis_imgs.append({"name":f})
|
|
number_of_frame = len(undis_imgs)
|
|
first_image_name = undis_imgs[0]['name']
|
|
offset = int(re.findall(r'\d+', first_image_name)[-1]) - 1
|
|
|
|
except RuntimeError:
|
|
print("Error: while importing the undistorted images.")
|
|
|
|
#import abc (Animated Camera)
|
|
|
|
try:
|
|
|
|
# In this part of the code we import the alembic in the cam_path to get the animated camera
|
|
# We use cam_location and cam_obj to store information about this camera
|
|
# We look for a camera (of type 'Persp') with the name 'anim' (not to confuse them with previously imported cams)
|
|
|
|
# Once the cam has been found we select the main camera of the scene.
|
|
# The rest of the code is setting up the display of the background image,
|
|
# Since it's not a simple image but an image Sequence, we have to use the offset and the number of frame
|
|
# Information taken from the previous block of code.
|
|
# The frame method is the one that align with the Point Cloud althought this may change,
|
|
# so feel free to try out the two other settings if something changes on previous nodes.
|
|
# We also have to make the scene render film transparent because we want to be able to display
|
|
# our background afterward in the next block of code
|
|
|
|
bpy.ops.wm.alembic_import(filepath=args.SFM_cam_path)
|
|
animated_cams = bpy.context.selected_editable_objects[:]
|
|
cam_location = mathutils.Vector((0, 0, 0))
|
|
cam_obj = None
|
|
for obj in animated_cams:
|
|
if obj.data and obj.data.type == 'PERSP' and "anim" in obj.data.name:
|
|
bpy.context.scene.collection.objects.link(obj)
|
|
bpy.context.view_layer.objects.active = obj
|
|
bpy.context.scene.camera = obj
|
|
cam_location = obj.location
|
|
cam_obj = obj
|
|
if args.Use_Background :
|
|
bpy.ops.image.open(filepath=args.undisto_images + "/" + first_image_name, directory=args.undisto_images, files=undis_imgs, relative_path=True, show_multiview=False)
|
|
bpy.data.cameras[obj.data.name].background_images.new()
|
|
bpy.data.cameras[obj.data.name].show_background_images = True
|
|
bpy.data.cameras[obj.data.name].background_images[0].image = bpy.data.images[first_image_name]
|
|
bpy.data.cameras[obj.data.name].background_images[0].frame_method = 'CROP'
|
|
bpy.data.cameras[obj.data.name].background_images[0].image_user.frame_offset = offset
|
|
bpy.data.cameras[obj.data.name].background_images[0].image_user.frame_duration = number_of_frame
|
|
bpy.data.cameras[obj.data.name].background_images[0].image_user.frame_start = 1
|
|
bpy.context.scene.render.film_transparent = True
|
|
except RuntimeError:
|
|
print("Error: while importing the alembic file (Animated Camera).")
|
|
|
|
#Place the particle plane
|
|
try:
|
|
|
|
# This is a key step if you are displaying a Point Cloud.
|
|
# We are using a particle system later in the code to display the Point Cloud.
|
|
# To make it so, we need a model for the particle, a object that will be repeated a lot to make a shape.
|
|
# In order to do that we need a plane (one face only for optimisation purpose) that always face the camera.
|
|
# So we made a plane and made it a child (in the parenting system) of the camera. That way whenever the cam
|
|
# moves, the plane moves and turn accordingly.
|
|
|
|
# Bmesh creates the plane and put it into the mesh. We change the size of the plane according to
|
|
# the scale given in arguments. We need to adjust the plane's location because putting it at the
|
|
# exact location of the camera blocks the view. Then, the switcher gives a RGBA color according to
|
|
# the given argument. We have to use a material that uses 'Emission'
|
|
# otherwise the particle is going to react to lights and we don't really need that (the color wouldn't be clear).
|
|
# To do that we have to use the shader 'node_tree' we clear all links between nodes, create the emission node
|
|
# and connect it to the 'Material Output' node (which is what we will see in render).
|
|
# Finally we use the switcher to color the model.
|
|
|
|
plane = bpy.data.meshes.new('Plane')
|
|
objectsPlane = bpy.data.objects.new(name="Plane", object_data=plane)
|
|
bm = bmesh.new()
|
|
bmesh.ops.create_grid(bm, x_segments = 1, y_segments = 1, size = 1.0)
|
|
bm.to_mesh(plane)
|
|
bm.free()
|
|
if (args.SFM_Data.endswith('.abc')):
|
|
objectsPlane.scale = mathutils.Vector((args.Particle_Size, args.Particle_Size, args.Particle_Size))
|
|
cam_location.y += -2.0
|
|
objectsPlane.location = cam_location
|
|
bpy.context.scene.collection.objects.link(objectsPlane)
|
|
bpy.data.objects['Plane'].parent = cam_obj
|
|
bpy.context.view_layer.objects.active = objectsPlane
|
|
|
|
col = bpy.data.materials.new('Color')
|
|
objectsPlane.active_material = col
|
|
objectsPlane.active_material.use_nodes = True
|
|
objectsPlane.active_material.node_tree.links.clear()
|
|
objectsPlane.active_material.node_tree.nodes.new(type='ShaderNodeEmission')
|
|
objectsPlane.active_material.node_tree.links.new(objectsPlane.active_material.node_tree.nodes['Emission'].outputs['Emission'], objectsPlane.active_material.node_tree.nodes['Material Output'].inputs['Surface'])
|
|
if (args.SFM_Data.endswith('.abc')):
|
|
objectsPlane.active_material.node_tree.nodes['Emission'].inputs[0].default_value = switcher.get(args.Particle_Color, 'Invalid Color')
|
|
|
|
except RuntimeError:
|
|
print("Error: while setting up the particle model.")
|
|
|
|
if (args.SFM_Data.endswith('.abc')):
|
|
# This part is all about importing the Point Cloud and setting up the Particle System.
|
|
# After importing the alembic, we look for a specific mesh in the file. Again the hardcoded name would be a
|
|
# problem if the previous nodes hadn't name it specificaly that (.001 because a mesh with the same name has
|
|
# been imported with the animated camera).
|
|
|
|
# Once the Point Cloud has been found. We make it the active object (important for the node_tree later).
|
|
# Then, we create a particle system on it. Render_type set to object and the said object is the plane,
|
|
# thanks to that the particle format is set to repeat the plane. Emit_from 'vert' so the points of the
|
|
# point cloud are the one rendering the particle.
|
|
# The count is the number of particle repeated on the point cloud. We use the rate given as arguments
|
|
# to give a number. Most of the following settings are just formalities but use_rotation and use_rotation_instance,
|
|
# those two make sure to use the same rotaion than the model (which is needed to have the particle always facing the camera).
|
|
|
|
#import abc (Point Cloud)
|
|
try:
|
|
bpy.ops.wm.alembic_import(filepath=args.SFM_Data)
|
|
all_abc_info = bpy.context.selected_editable_objects[:]
|
|
for obj in all_abc_info:
|
|
if obj.name == 'mvgPointCloud.001': #May have a problem with such hard code
|
|
bpy.context.scene.collection.objects.link(obj)
|
|
bpy.context.view_layer.objects.active = obj
|
|
obj.modifiers.new("ParticleSystem", "PARTICLE_SYSTEM")
|
|
particle_system = bpy.data.particles["ParticleSystem"]
|
|
particle_system.render_type = 'OBJECT'
|
|
|
|
particle_system.instance_object = bpy.data.objects["Plane"]
|
|
particle_system.emit_from = 'VERT'
|
|
|
|
if (args.SFM_Data.endswith('.abc')):
|
|
particle_system.count = int(args.Point_Cloud_Density * len(obj.data.vertices.values()))
|
|
particle_system.frame_end = 1.0
|
|
particle_system.use_emit_random = False
|
|
particle_system.particle_size = 0.02
|
|
particle_system.physics_type = 'NO'
|
|
particle_system.use_rotations = True
|
|
particle_system.use_rotation_instance = True
|
|
particle_system.rotation_mode = 'GLOB_X'
|
|
|
|
except RuntimeError:
|
|
print("Error: while importing the alembic file (Point Cloud).")
|
|
#Or import obj directly
|
|
|
|
# The import via obj needs a bit of work too. For showing an outline of the object, we need to add two materials to the mesh :
|
|
# Center and Edge, we are using a method that consists in having a "bold" effect on the Edge Material so we can see it
|
|
# around the Center material. We do that by using a Solidify Modifier on which we flip normals and reduce Thickness to bellow zero.
|
|
# The more the thickness get bellow zero, the more the egde will be largely revealed.
|
|
elif (args.SFM_Data.endswith('.obj')):
|
|
bpy.ops.import_scene.obj(filepath=args.SFM_Data)
|
|
|
|
center = bpy.data.materials.new('Center')
|
|
center.use_nodes = True
|
|
center.node_tree.links.clear()
|
|
center.node_tree.nodes.new(type='ShaderNodeEmission')
|
|
center.node_tree.links.new(center.node_tree.nodes['Emission'].outputs['Emission'], center.node_tree.nodes['Material Output'].inputs['Surface'])
|
|
center.node_tree.nodes['Emission'].inputs[0].default_value = (0,0,0,0)
|
|
|
|
if not args.Use_Background and args.SFM_Data.endswith('.obj'):
|
|
center.node_tree.nodes['Emission'].inputs[0].default_value = (0.05, 0.05, 0.05, 1) #Same Color as the no background color in blender
|
|
|
|
|
|
edge = bpy.data.materials.new('Edge')
|
|
|
|
edge.use_nodes = True
|
|
edge.node_tree.links.clear()
|
|
edge.node_tree.nodes.new(type='ShaderNodeEmission')
|
|
edge.use_backface_culling = True
|
|
edge.node_tree.links.new(edge.node_tree.nodes['Emission'].outputs['Emission'], edge.node_tree.nodes['Material Output'].inputs['Surface'])
|
|
edge.node_tree.nodes['Emission'].inputs[0].default_value = switcher.get(args.Edge_Color, 'Invalid Color')
|
|
|
|
bpy.data.meshes['mesh'].materials.clear()
|
|
bpy.data.meshes['mesh'].materials.append(bpy.data.materials['Center'])
|
|
bpy.data.meshes['mesh'].materials.append(bpy.data.materials['Edge'])
|
|
|
|
print(bpy.data.meshes['mesh'].materials.values())
|
|
|
|
bpy.data.objects['mesh'].modifiers.new('New', type='SOLIDIFY')
|
|
bpy.data.objects['mesh'].modifiers["New"].thickness = -0.01
|
|
bpy.data.objects['mesh'].modifiers["New"].use_rim = False
|
|
bpy.data.objects['mesh'].modifiers["New"].use_flip_normals = True
|
|
bpy.data.objects['mesh'].modifiers["New"].material_offset = 1
|
|
else:
|
|
print("SFM_Data isn't in the right format, alembics(.abc) and object(.obj) only are supported")
|
|
|
|
#WE HAVE TO USE THE COMPOSITING GRAPH TO MAKE THE BACKGROUND IMAGE VISIBLE
|
|
# We setup all the nodes in the first place, even if we don't need them in our configuration. We put the setting in all of them.
|
|
# Only after having done that we can control which of the node we link in the graph according to the option we were given.
|
|
# If the SFM Data is a Mesh, its extension is .obj so we have to build the graph accordingly. If the Background image setting was activated,
|
|
# we need to include it in our node tree through the "Image" and Scale node.
|
|
try:
|
|
bpy.context.scene.use_nodes = True
|
|
|
|
#CREATE ALL NODES WE NEED (regardless of the options)
|
|
bpy.context.scene.node_tree.nodes.new(type="CompositorNodeAlphaOver")
|
|
bpy.context.scene.node_tree.nodes.new(type="CompositorNodeScale")
|
|
bpy.context.scene.node_tree.nodes.new(type="CompositorNodeImage")
|
|
|
|
bpy.context.scene.node_tree.nodes.new(type="CompositorNodePremulKey")
|
|
bpy.context.scene.node_tree.nodes.new(type="CompositorNodeMixRGB")
|
|
|
|
#SET THEM UP CORRECTLY (still regardless of the option)
|
|
bpy.data.scenes["Scene"].node_tree.nodes["Mix"].blend_type = 'LIGHTEN'
|
|
bpy.data.scenes["Scene"].node_tree.nodes["Image"].frame_duration = number_of_frame
|
|
bpy.data.scenes["Scene"].node_tree.nodes["Image"].frame_offset = offset
|
|
bpy.data.scenes["Scene"].node_tree.nodes["Scale"].space = 'RENDER_SIZE'
|
|
bpy.data.scenes["Scene"].node_tree.nodes["Scale"].frame_method = 'CROP'
|
|
|
|
#LINKS THE NODES THAT NEEDS TO BE LINKED
|
|
if args.Use_Background :
|
|
if args.SFM_Data.endswith('.obj'):
|
|
bpy.context.scene.node_tree.nodes["Image"].image = bpy.data.images[first_image_name]
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Mix'].outputs['Image'], bpy.context.scene.node_tree.nodes['Composite'].inputs['Image'])
|
|
#Two Inputs of AlphaOver are named "Image" so we'll use index instead
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Render Layers'].outputs['Image'], bpy.context.scene.node_tree.nodes['Alpha Convert'].inputs['Image'])
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Alpha Convert'].outputs['Image'], bpy.context.scene.node_tree.nodes['Mix'].inputs[2])
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Scale'].outputs['Image'], bpy.context.scene.node_tree.nodes['Mix'].inputs[1])
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Image'].outputs['Image'], bpy.context.scene.node_tree.nodes['Scale'].inputs['Image'])
|
|
else:
|
|
bpy.context.scene.node_tree.nodes["Image"].image = bpy.data.images[first_image_name]
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Alpha Over'].outputs['Image'], bpy.context.scene.node_tree.nodes['Composite'].inputs['Image'])
|
|
#Two Inputs of AlphaOver are named "Image" so we'll use index instead
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Render Layers'].outputs['Image'], bpy.context.scene.node_tree.nodes['Alpha Over'].inputs[2])
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Scale'].outputs['Image'], bpy.context.scene.node_tree.nodes['Alpha Over'].inputs[1])
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Image'].outputs['Image'], bpy.context.scene.node_tree.nodes['Scale'].inputs['Image'])
|
|
else:
|
|
if args.SFM_Data.endswith('.obj'):
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Mix'].outputs['Image'], bpy.context.scene.node_tree.nodes['Composite'].inputs['Image'])
|
|
#Two Inputs of AlphaOver are named "Image" so we'll use index instead
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Render Layers'].outputs['Image'], bpy.context.scene.node_tree.nodes['Alpha Convert'].inputs['Image'])
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Alpha Convert'].outputs['Image'], bpy.context.scene.node_tree.nodes['Mix'].inputs[2])
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Scale'].outputs['Image'], bpy.context.scene.node_tree.nodes['Mix'].inputs[1])
|
|
bpy.context.scene.node_tree.links.new(bpy.context.scene.node_tree.nodes['Image'].outputs['Image'], bpy.context.scene.node_tree.nodes['Scale'].inputs['Image'])
|
|
except RuntimeError:
|
|
print("Error: while composing the compositing graph.")
|
|
|
|
## Starts the rendering and launchs it with a blender animator player
|
|
|
|
try:
|
|
# Setup the render format and filepath
|
|
bpy.context.scene.render.image_settings.file_format = 'FFMPEG'
|
|
if args.Output_Format == 'mkv':
|
|
bpy.context.scene.render.ffmpeg.format = 'MKV'
|
|
elif args.Output_Format == 'avi':
|
|
bpy.context.scene.render.ffmpeg.format = 'AVI'
|
|
elif args.Output_Format == 'mov':
|
|
bpy.context.scene.render.ffmpeg.format = 'QUICKTIME'
|
|
else:
|
|
bpy.context.scene.render.ffmpeg.format = 'MPEG4'
|
|
bpy.context.scene.render.filepath = args.output_path + '/render.' + args.Output_Format
|
|
# Render everything on to the filepath
|
|
bpy.ops.render.render(animation=True)
|
|
# Starts a player automatically to play the output (Usefull for developpers to see what they do but it doesn't really have its place in a software)
|
|
# bpy.ops.render.play_rendered_anim()
|
|
except RuntimeError:
|
|
print("Error: while rendering the scene.")
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main() |