Skip to content

A lightweight NeRF running purely on Google Colab.

Notifications You must be signed in to change notification settings

ChunHsuehLee/Mini-Nerf

Repository files navigation

Mini-NeRF

This repository presents a lightweight version of NeRF, which is trainable within 5 hours on Google Colab. Pre-trained checkpoints and Colab scripts are provided. It is useful for people new to NeRF to get a quick sense of what NeRF is.

Three features are included in this project.

1. Accessible to Everyone

The original model is hard to train on laptops or computers without strong GPUs. Mini-NeRF eliminate the inaccessibility of NeRF, anyone with a google account will have access to Colab and Mini-NeRF can be trained on free GPUs provided by Colab.

2. Visualization Cells

The training result can also be easily viewed on Colab, the visualization code cells are also included in the scripts. allAngles

blender_paper_lego_spiral_100000_rgb.mp4

3. Interactive Interface

An interactive visualization interface is also provided for user to view customized angles of the synthesized object.

Installation

  1. Clone this repo git clone https://github.com/ChunHsuehLee/Mini-Nerf.git
  2. Upload the whole folder to your google drive.
  3. (Optional) Use the pre-trained checkpoints. Download the checkpoints here, upload the whole folder (include logs) to the Mini-Nerf folder you just uploaded.
  4. Run the Colab script (.ipynb file) cell by cell.
  5. Have fun!

References

Thank yenchenlin and bmild for their super useful repos!

About

A lightweight NeRF running purely on Google Colab.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published