facebookresearch
How to remove view positions from the generated .glb file?
I have successfully generated the .glb file for the set of images. When i visualise this file it shows me the correct reconstruction of the point clouds. My question is how can i remove all of the marked camera view points(in red) and others before generating .glb file so that it wont show in the generated file?
Have you tried in function "predictions_to_glb" set show_cam=False ? That should work i guess. By default is set to True.
@vgutierrez2404 Hey thanks for your quick reply the solution works. Although I have one more question for you can i convert the .glb file into .pcd fle so that I can do surface reconstruction on the point clouds. Do you have any suggestion on this?
Hi, the author created a file called "vggt_to_colmap", which parses the predictions of the model into the colmap style file https://colmap.github.io/format.html. This didn't work for me since I needed a point cloud, as you. What I've done is creating a parser to export a .ply file that looks like: ["ply", "format ascii 1.0", "element vertex {}".format(num_vertices), "property float32 x", "property float32 y", "property float32 z", "property uint8 red", "property uint8 green", "property uint8 blue", "end_header"]
And works with the open3d function "open3d.io.read_point_cloud()" (also it should work with other libraries such as pyvista). If the author allows me to commit the solution ( I added it in the issue https://github.com/facebookresearch/vggt/issues/76 ) I will do it, and it should work for you.
If you need further help tell me, i can provide you the code i used
@vgutierrez2404 It would be better if you can provide the code to convert into point cloud. Thanks
Yeah I could work on that. I can`t at the moment, but probably in a couple of days.
@vgutierrez2404 Hey i think this is the function I have been asking for write_points3d_ply() ?