shreyesss / HDRSplat

Official Implementation of BMVC 2024 paper titled "HDRSplat: Gaussian Splatting for High Dynamic Range 3D Scene Reconstruction from Raw Images"
https://aryan-garg.github.io/hdrsplat/
29 stars 0 forks source link

training error about colmap #2

Open serizawa-04013958 opened 4 weeks ago

serizawa-04013958 commented 4 weeks ago

Hello, thank you for your great work. I'd like to train your code by candlefiat, I faced below errors. do you know how to use training dataset?

`Output folder: /cig/common06nb/deserizk/HDRSplat/output/candlefiat Optimizing /cig/common06nb/deserizk/HDRSplat/output/candlefiat /cig/common06nb/deserizk/archive/rawnerf/scenes/candlefiat/sparse [16/10 15:35:51] demosaic [16/10 15:35:51]

Reading camera 1/52Traceback (most recent call last): File "/cig/common06nb/deserizk/HDRSplat/train.py", line 352, in training(lp.extract(args), op.extract(args), pp.extract(args), args.render_iterations, args.test_iterations, args.save_iterations, args.checkpoint_iterations, args.start_checkpoint, args.debug_from) File "/cig/common06nb/deserizk/HDRSplat/train.py", line 62, in training scene = Scene(dataset, gaussians) File "/cig/common06nb/deserizk/HDRSplat/scene/init.py", line 52, in init scene_info = self.callback["Colmap"](args.source_path, args.images, args.eval, "demosaic", args.denoise_method) File "/cig/common06nb/deserizk/HDRSplat/scene/dataset_readers_raw.py", line 202, in readColmapSceneInfo cam_infos_unsorted = readColmapCameras(cam_extrinsics=cam_extrinsics, cam_intrinsics=cam_intrinsics, images_folder=os.path.join(path, reading_img_dir), raw_folder=os.path.join(path, reading_raw_dir),scene_path=path, denoise_method=denoise_method) File "/cig/common06nb/deserizk/HDRSplat/scene/dataset_readers_raw.py", line 125, in readColmapCameras assert False, "Colmap camera model not handled: only undistorted datasets (PINHOLE or SIMPLE_PINHOLE cameras) supported!" AssertionError: Colmap camera model not handled: only undistorted datasets (PINHOLE or SIMPLE_PINHOLE cameras) supported!`

shreyesss commented 4 weeks ago

3DGS implementation only supports simple pinhole camera model for now, to convert the dataset to simple pinhole camera model, install colmap CLI and follow these steps:

$ DATASET_PATH=/path/to/dataset

$ colmap feature_extractor \ --database_path $DATASET_PATH/database.db \ --image_path $DATASET_PATH/images

$ colmap exhaustive_matcher \ --database_path $DATASET_PATH/database.db

$ mkdir $DATASET_PATH/sparse

$ colmap mapper \ --database_path $DATASET_PATH/database.db \ --image_path $DATASET_PATH/images \ --output_path $DATASET_PATH/sparse

$ mkdir $DATASET_PATH/dense

$ colmap image_undistorter \ --image_path $DATASET_PATH/images \ --input_path $DATASET_PATH/sparse/0 \ --output_path $DATASET_PATH/dense \ --output_type COLMAP \ --max_image_size 2000

For more details refer to: 
https://colmap.github.io/cli.html