I'm working on reconstructing a 3D model of a plant using Neural Radiance Fields (NeRF). For camera pose estimation, I'm using the COLMAP GUI and exporting the camera positions and poses as .bin files. My goal is to use these COLMAP-generated poses to train a NeRF model using Nerfstudio. However, the Nerfstudio documentation doesn’t explain how to use COLMAP output directly for training, as it typically relies on the ns-process-data command: ns-process-data {images, video} --data {DATA_PATH} --output-dir {PROCESSED_DATA_DIR} How can I integrate the .bin files from COLMAP into the NeRF training pipeline with Nerfstudio? Continue reading...
