Instant-NGP Simple Install Directions

This is a guide for building and running instant-ngp with custom datasets. Code based on instant-ngp-helper with updated directions, protocol, and dependencies.

Instructions

Follow these steps in order -- it matters.

Setup

1. Install Visual Studio Community 2019. Be sure to include Desktop Development with C++.
2. Install CUDA Toolkit 11.6
3. Install Anaconda Prompt (Python 3.9)
4. Install Git
5. Install OptiX 7.4.0 SDK
6. Install Vulkan SDK
7. Add OptiX_INSTALL_DIR to System variables with the value path C:\ProgramData\NVIDIA Corporation\OptiX SDK 7.4.0

Installation

8. Open Anaconda Prompt and cd to your desired install directory.
9. Type:

git clone --recursive https://github.com/ethanproia/instant-ngp-simple

10. cd into ...\instant-ngp-simple and run:

python install.py

11. When install and build are finished, run:

pip install opencv-python

12. Add COLMAP folder to PATH: C:\Users\[YOUR USERNAME]\...\instant-ngp-simple\COLMAP

How to Use

instant-ngp-helper can process both video and image still sequences into NeRFs.

Images

13. Navigate to your ...\instant-ngp-simple\instant-ngp\tmp directory, make a new folder called images, and load your image stills dataset into the images folder.
14. In a fresh Anaconda Prompt, cd into your instant-ngp-simple directory and run:

python run.py

15. Click the instant-ngp (images) button in the UI, navigate back to the Anaconda Prompt window and follow the COLMAP instructions. (Don't forget to add the COLMAP folder to PATH as shown in Step 12 or you'll get a COLMAP error.)
16. When COLMAP finishes processing and says done, go back to the UI and click instant-ngp NeRF.

Videos

17. Navigate to your ...\instant-ngp-simple\instant-ngp\tmp directory and just drop your video file in the folder.
18. In a fresh Anaconda Prompt, cd into your instant-ngp-simple directory and run:

python run.py

19. Click the instant-ngp (videos) button in the UI, navigate back to the Anaconda Prompt window and follow the ffmpeg and COLMAP instructions. (Don't forget to add the COLMAP folder to PATH as shown in Step 12 or you'll get a COLMAP error.)
20. When COLMAP finishes processing and says done, go back to the UI and click instant-ngp NeRF.

Video Tutorial

Follow along with my guided video tutorial. (Coming soon)

Loading Sample Datasets

1. Copy all files from inside any of the sample folders into ...\instant-ngp-simple\instant-ngp\tmp
2. Open an Anaconda Prompt and cd into ...\instant-ngp-simple\ and run:

python run.py

3. In the UI that opens, click instant-ngp (photos), which will load the dataset, open the testbed.exe, and being training the NeRF.