Danpeng Chen
Danpeng Chen
Hello, to reconstruct more details, we adopted the split strategy of absGS(https://ty424.github.io/AbsGS.github.io/) and reduced the point cloud splitting threshold in certain scenes.
For regions with poor initial accuracy or weak textures, the PGSR optimization may fall into local optima, resulting in geometric errors. The suggestion is: if the error in such regions...
Hello, the rendered normal is in the camera coordinate system.
Reconstructing with weak textures without any prior information can be quite challenging. There are a few suggestions that might alleviate these issues: 1) Disable the abs splitting strategy by setting...
Our training commands： python train.py -s data_path -m out_path -r2 --densify_abs_grad_threshold 0.002 --single_view_weight_from_iter 0 --multi_view_weight_from_iter 0 We have identified weak textures, complex textures, and non-global illumination variations in this scene....
你好，需要先按照这个方案进行训练的。
Hi, the multi-view nearest camera selection strategy may require choosing appropriate thresholds based on the actual scenario, as the default parameters may not always be suitable. You can check whether...
Yes, if the entire scene size can be considered as the basis for adjustment, the parameters will be more universal.
你好， 1、一开始添加多视图损失也是可以的，可能训练耗时长一些，表面更平滑一些 2、计算ncc loss时候，只在部分不需要梯度回传的运算上使用了torch.no_grad()
Hello, if Mask optimization is applied, the existing loss functions need to be modified accordingly. Therefore, Mask-based training will not be considered.