Closed NiluK closed 1 year ago
You can actually check at the shapeformer.data.paper_datasets.list_dataset.ListDataset
dataset.
The simplest way to test new unseen point cloud is to add a folder (for example, unseen
) in the demo/dataset
that contains a Xct.npy
file. The Xct.npy file should contain the Nx3
point cloud in numpy array format.
Then you will need to add the name of the folder to the demo.lst
file.
After this, just run the demo script and you will see the results.
Also notice that these are all customizable in the config files (demo_vqdif.yaml).
Hi there, thank you for this.
This paper is very promising and I was able to generate some good meshes with your above instructions. VQIF is pretty impressive..
ShapeFormer takes a substantial amount of time on a v100 so hopefully new advances can accelerate this.
All the best!.
Hello, this paper seems amazing! I am curious as to how to generate a mesh from an unseen point cloud.
I have trained the shapes on both the demo steps and got some great results with the demo examples but I am confused as to how to input a .ply file into this.
Is there a simple way to do it?
Thanks