Low-quality mesh output on the TnT Meetingroom dataset.

[zhuixunforever]

Thank you for your interesting work. I followed your suggestion and trained Meetingroom with the following parameters, but I got a low-quality mesh, as follows:

1. Commands for training and extracting the mesh:

python train.py --imp_metric indoor --rasterizer radegs --dense_gaussians --mesh_config default --decoupled_appearance --log_interval 200 --data_device cpu -s <dataset_path> -m <output_path>

python mesh_extract_sdf.py --rasterizer radegs -s <dataset_path> -m <output_path>

2. Mesh reconstruction results:

Could you give me some suggestions?