Just letting people know there is a new Ai repository for those interesting in creating 3D models from images. It is incredible at creating 3D geometry.
Igor Aherne’s Github repository including a one click installer is here:
I tried to use it as a custom Node in ComfyUi, but it killed my Comfy because it (at the time created a conflict with Torch/Cuda versions). I have it running as a stand alone tool with a Gradio interface which I am v. happy with.
Select an image you want to pull a 3d model from. There is an automatic background remover, but to get good results best to be picky about even lighting, oblique angle, plain background.
The Ai first creates a spinning preview and normal map.
If happy with it you can create a .glb file and a .ply file, and then download it.
Using a 1024 x 1024 image the .glb will be about 5MB and the .ply will be about 50MB.
TD cannot understand .glb, but can import .ply as a pointcloud.
To get the geometry and mapping ready for TD check out Factory Settings Tutorial and save yourself a lot of agony by using his TD Script 1.4.2 for exporting content between TD and Blender:
His tox and Py.script is available as a free download from his Patreon.
https://www.patreon.com/posts/td-scripts-1-4-2-112901735
Having dragged/dropped the .glb into Blender it becomes a smooth workflow to export it to TD as a mesh and as a material (texture).
Once inside TD it’s over to you and your imagination . . .
Post-script: MeshLab is a great open source tool for cleaning up meshes and exporting them in alternative formats such as .3ds, .xyz, .stl, .obj, .glb, .ply, and others.
https://www.meshlab.net/
