About Params

[201722111112027]

Hello, while reading the source code, I noticed that during the model training process, many key parameters such as args.rune_dead_points, args.mercy-points, and args.lambda-alpha_regul are invalid (False or 0.). Setting these parameters to invalid makes the model appear equivalent to the original 3DGS. I am quite puzzled about this issue and hope to receive your reply.

[PanagiotisP]

Yeah. Running the code with no changes to the default arguments will run the baseline. The set of parameters used for the paper are described in the full_eval.py file.

In the future I should probably change that and have the full_final model as the default one...