Run my own dataset

[luoshuiyue]

Hello! Thank you for your excellent work. I would like to ask how to run the data collected by myself on human performers and extract those human surfaces. Specifically, I want to know how to prepare my own data, such as whether raw data, mask and pose data are required, and then run which command to get the surface reconstruction results.

[YuxueYang1204]

Hi @luoshuiyue ! Our TrimGS follows the 3D Gaussian Splatting and use the same COLMAP loader as 3DGS, you can prepare your data following here.

[luoshuiyue]

Thank you for your reply. I successfully got the Gaussian model, how to get the mesh model?

[YuxueYang1204]

The warning indicates that mesh has 0 vertices. The mesh extraction is based on the rendered depth, so it won't fail as long as the depth map can be rendered. You can check the rendered depth maps first as mentioned in #12. Or you can upload an output sample if possible, I'm willing to analyze the specific issue based on it when I am available.

[luoshuiyue]

Hello~ I first set up the required environment according to the readme. Then put my dataset lj_0705_new under the directory data, which contains the masked raw image and the pose in sparse format (the image contains the shooting data of three cameras (top, middle and bottom) around the human body), as shown in the Figure 1, Figure 2: Next I run the train tune render extract_mesh command, as shown in Figure 3, Figure 4, Figure 5: The result directory is shown in Figure 6 below:

Can you give me some guidance to get the mesh result?

[GuoPingPan]

Is there any new progress on this issue?

[luoshuiyue]

No, I gave up for the moment then continue to solve.