shapenet_renderer

Rendering script for shapenet object dataset using blender2.8

Some preparation

Firstly, use obj2gltf to convert all your models in your shapenet dataset into .glb format. You can refer to the github page for obj2gltf.
You can install obj2gltf by:

npm install -g obj2gltf

You can convert your model using the "convert_glb.py", in which you need to modify the variable 'shapenet_dataset', which contains a lot of obj models, and 'dst_dir', where you save your converted glb file.

Render the model

You should install blender 2.8a. This script is tested on MacOS and Linux, and it should work on Windows. Run the script by:

blender --background get_pcd.blend -P render_rgb_depth.py

This script will render both RGB and Depth of the model.
Also remember to change 'shapenet_dir' and 'dst_dir' to where you save your glb model and where you want to save your results.
The results will be saved in OPEN_EXR format, you can also save it to png or jpg by modifying line 62 and line 63.

Some issues during the rendering

Some models in shapenet dataset are not able to load, therefore, you will have to delete the model when you find your script stoping.