search

autonomousvision / differentiable_volumetric_rendering

This repository contains the code for the CVPR 2020 paper "Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervision"
http://www.cvlibs.net/publications/Niemeyer2020CVPR.pdf
MIT License
798 stars 90 forks source link

Generate own dataset #27

Closed bell-one closed 4 years ago

bell-one commented 4 years ago

hello,

I'd like to test your great project with my 3D models like shapenet data you provide I try to make same structured data with blender but maybe there's something wrong with my codes

how did you generate your training dataset with shapenet models? can you provide some functions or project to do this?

thank you

m-niemeyer commented 4 years ago

Hi @bell-one , thanks for your message!

We prepared the dataset with some internal rendering scripts which are not released. However, as you correctly mentioned, we use the open source software blender for this. If you search a little bit for python blender scripts, there already great projects on this, for example the stanford-shapenet-renderer.

Good luck with your research!