Spaces:
Sleeping
Download external_libs/render_toolbox/README.md from FedericoGiulio/LoomNet: direct link, hf CLI and curl.
- Browser
- Download file 3.93 kB
-
https://huggingface.co/spaces/FedericoGiulio/LoomNet/resolve/main/external_libs/render_toolbox/README.md
- Command line
-
hf download hf://spaces/FedericoGiulio/LoomNet/external_libs/render_toolbox/README.md
-
curl -L -o README.md https://huggingface.co/spaces/FedericoGiulio/LoomNet/resolve/main/external_libs/render_toolbox/README.md
A newer version of the Gradio SDK is available: 6.32.0
Render Toolbox
Blender toolbox for rendering 3D scenes.
Tested in Blender 3.5.1 and Python 3.9.
π½οΈ Example
https://github.com/huanngzh/render-toolbox/assets/78398294/b0e9ae6b-bfa4-4863-8194-3cbce523cd5b
π¨ Quick Start
First, get your blender python env path to install necessary packages into blender python correctly. Run the following command in your Blender Python Console.
>>> import sys
>>> print(sys.executable)
/opt/blender-3.5.1/3.5/python/bin/python3.10
Then install packages:
# Update the actual output you get to the following command
/opt/blender-3.5.1/3.5/python/bin/python3.10 -m pip install -r requirements.txt
Run the example script to render:
blender -b -P example.py
π Tutorial
Create a scene handler to handle most rendering configuration operations.
scener = SceneHandler()
Initialize the render engine. You can select BLENDER_EEVEE or BLENDER_EEVEE as your rendering engine. The former is faster.
scener.init_render_engine("BLENDER_EEVEE")
# Or
scener.init_render_engine("CYCLES")
Initialize the intrinsics of camera. Modify the camera lens and sensor width as you need.
cam_default = scener.init_camera(camera_lens, camera_sensor_width)
Import object.
Now formats including object with color on vertex(exported by trimesh), object with mtl, glb, vrm are supported. You can use type to select one of it.
scener.import_object(type="vertex_colored", filepath="**.obj")
# Or import object with mtl. You may need to specify the forward-axis and up-axis, with "Y" and "Z" are default values.
scener.import_object(type="obj", filepath="**.obj")
# Or import glb
scener.import_object(type="glb", filepath="**.glb")
# Or import vrm (VRM is an extension of glTF 2.0, so supported by the same way as glb)
scener.import_object(type="glb", filepath="**.vrm")
Other functions to process on the imported object.
# Modify the color into a single one
scener.modify_vertex_color(((0.52941, 0.80784, 0.92157))) # Sky blue
# Preprocess the objects, like using auto smooth and clear the animation
scener.preprocess_objs()
# Normalize the scene into a sphere
scener.normalize_scene(range=1.0)
Set the environment light. You can set global light with a specified light intensity, or use an environment map.
scener.set_env_light(env_light=1.0)
# Or use an environment map
scener.set_env_light(env_path="**.exr")
Prepare camera layout and add to the scene. Now sphere layout is supported. Specify the center (lookat), radius (camera distance from the center), elevations, num_per_layer.
# The following code will create a circle of cameras at elevation 30 above the object.
distance = 1.5
cam_positions, cam_mats, eles, azis = get_camera_positions_on_sphere(
center=(0, 0, 0), radius=distance, elevations=[30], num_per_layer=120
)
for camera_matrix in cam_mats:
scener.add_camera(camera_matrix)
Rendering output setting. Render color, normal, depth, albedo or pbr maps. Note that the export works well or bad may depend on whether the object has corresponding maps.
scener.set_output(
output_dir="outputs",
width=512,
height=512,
output_types=["color"], # "color", "normal", "depth", "albedo", "pbr"
)
Perform rendering.
scener.render()
You can also export the meta info of the cameras, or save the rendered images to a video. Please refer to example.py for details.
Other Statements
Source of model assets:
- vertex_colored.obj generated by EpiDiff.
- Mario.glb generated by Tripo.
- Android_Figure_Panda comes from Google Scanned Objects.