Strange random color when training

[Sylvia6]

Hi，I have some doubt of my train result on other dataset. It looks like below:( left picture is the train result, right is ground trurh)

I can not find out why it happens, is it caused by the rasterization method?

Below show some of the rasterization result:( left: train render result, medium: rgb ground truth, right: rasterization mask)

[darglein]

Can it be that these are spots that are not seen during training? So the color is uninitialized?

[Sylvia6]

@darglein But it appears during the training time. I use the dense point cloud map collected from lidar. So some of the views have relatively sparse points, which means there might be more holes in a picture and might be filled with environment map sample texture. I wonder what causes the above abnormal phenomena.

[Gatsby23]

@darglein But it appears during the training time. I use the dense point cloud map collected from lidar. So some of the views have relatively sparse points, which means there might be more holes in a picture and might be filled with environment map sample texture. I wonder what causes the above abnormal phenomena.

您好，我也尝试在自己的数据集上用ADOP进行训练，但有问题出来，可以加下您微信，跟您交流一下吗？

[Gatsby23]

@darglein But it appears during the training time. I use the dense point cloud map collected from lidar. So some of the views have relatively sparse points, which means there might be more holes in a picture and might be filled with environment map sample texture. I wonder what causes the above abnormal phenomena.

Sorry for troubing you. I also have some problem when test the algorithm on lidar point cloud. I build the lidar point cloud densly. However, when I train the algorithm. There are a lot of black holes on the viewer GUI. I debug the point_cloud_initial, it still densely enough. I really don't know why