mitsuba2-docker

A ready to run mitsuba2 docker image with installed:

• Optix 7
• CUDA 10.2
• Conda
• Jupyter Lab
• Mitsuba2

Enabled variants:

• scalar_rgb
• scalar_spectral
• gpu_autodiff_spectral
• gpu_autodiff_rgb
• gpu_autodiff_mono

Usage

Run it on any vast.ai, paperspace or other gpu instance.

1. Visit vast.ai create an account and upload your ssh keys
2. Choose a sufficient machine (CUDA >10.2, more memory the better)
3. Create with the image bonsaielectric/mitsuba2-docker:pathreparam-optix7
4. Connect ssh -p <PORT> <USER>@<HOST> -L 8080:localhost:8080
5. Lunch jupyter lab --ip 0.0.0.0 --allow-root --port 8080

Please bear in mind bonsaielectric/mitsuba2-docker:pathreparam-optix7 might be out of date, but you should be able to use this dockerfile to build one for your use.