Skip to content

Latest commit

 

History

History
38 lines (28 loc) · 1.46 KB

README.md

File metadata and controls

38 lines (28 loc) · 1.46 KB
Raw

Preprocess Data

Setup

  • Ensure that necessary dependencies have been installed (repo-level requirements).
  • Download our provided example datasets.

Overview:

  1. Preprocess multi-exposure HDR image captures.
  2. Estimate geometry.
  3. Estimate light sources.

Process Multi-Exposure HDR Image Captures

  • Follow the Jupyter Notebook provided in: preprocess_multiexposure_hdr/multi_exposure_hdr_preprocess.ipynb

Estimate Geometry

ns-train monosdf --pipeline.model.sdf-field.inside-outside True sdfstudio-data --data data/<your-dataset> --include-mono-prior True

Note that argument inside-outside should be set to True for an indoor scene.

Estimate Light Sources

  • PREREQUISITE: [Estimate Geometry](#Estimate Geometry)

  • Configure the lighting optimizer for your scene. We suggest starting with one of our provided example scenes:

    • lighting_estimation/scenes/lamp_scene.py
    • lighting_estimation/scenes/conference_scene.py

  • Modify lighting_estimation/optimize.py to set your desired scene as current_scene in L1.

  • Estimate light sources using the following command:

python lighting_estimation/optimize.py