Screened Poisson surface reconstruction stopping halfway or printing error for wrong normals about pymeshlab

[alemuntoni]

Hi, sorry for the late reply.
I don't understand what exactly is not working. Are you running something like

ms.save_current_mesh('file.obj')
ms.load_new_mesh('file.obj')

and it does not find the file in the load? It is not clear in your post.
Could you please share the script that you are running?

[Schlomoh]

Hey, thanks so much for trying to help and excuse my late response..

But yeah basically its just loading a mesh and saving it after the screened Poisson reconstruction. Since opening this issue the problem of the reconstruction returning 0 nodes after the depth of 5 just fixed itself.
But still it only computes the reconstruction if the preclean argument is set to true. No matter what the input obj file is (in this case various photo scanned models). If not set applying the filter always fails because of 'incorrect vertex normals'. Can you think of a consistent way of having the normals 'fixed' before applying the reconstruction filter?

Also when having a textured model, and using the 'transfer_vertex_attributes_to_texture_1_or_2_meshes' filter to bake the high poly geometry onto the low poly mesh as a normal map, is there any way of having the filter generate a tangent space normal map instead of an object space normal map?

Again thanks a lot for replying and for creating this awesome library!

Best regards
Mo

[alemuntoni]

What is the exact error you get when you talk about "incorrect vertex normals"?
Is it complaining because the point cloud does not have vertex normals? Because screened poisson requires non-null vertex normals on all the vertices. If the mesh you load does not have them, you need to compute them in some way.
Or, with the preclean parameter, all the vertices with non-proper normal are removed.

Also when having a textured model, and using the 'transfer_vertex_attributes_to_texture_1_or_2_meshes' filter to bake the high poly geometry onto the low poly mesh as a normal map, is there any way of having the filter generate a tangent space normal map instead of an object space normal map?

I am not sure, but I think there is not such filter in meshlab or pymeshlab.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.