ndming 29f03a303d Fix inconsistent training results with RGBA/PNG images ... The training relies on PIL to resize the input images and extracts the resized alpha to mask the rendered image during training. Since PIL pre-multiplies the resized RGB with the resized alpha, the training produces different Gaussian points depending on whether the input get resized or not. Moreover, the extracted alpha channel from PIL is not perfectly binarized, causing floaters around the edges.		2025-03-19 13:46:13 +01:00
..
camera_utils.py	added NeRF synthetic datasets training using depth regularization	2024-09-17 10:15:46 +02:00
general_utils.py	Fix inconsistent training results with RGBA/PNG images	2025-03-19 13:46:13 +01:00
graphics_utils.py	Added licenses, bumped submodule versions	2023-07-05 12:22:16 +02:00
image_utils.py	Added licenses, bumped submodule versions	2023-07-05 12:22:16 +02:00
loss_utils.py	New release: issues fixed and integrated training speed acceleration.	2024-10-24 10:52:57 +02:00
make_depth_scale.py	new release	2024-08-21 14:30:43 +02:00
read_write_model.py	added missing file, missing package	2024-09-02 15:36:17 +02:00
sh_utils.py	Initial commit	2023-07-04 10:00:48 +02:00
system_utils.py	Added licenses, bumped submodule versions	2023-07-05 12:22:16 +02:00