hack-mans
hack-mans
@Lustgard Currently we're missing the camera pose estimation stage which means you can't use any random input image. What I did for testing was to use Stable Diffusion + ControlNet...
@OverwriteDev I've almost got it working using the EG3D + Deep3DFaceRecon code but it's slightly off
@SizheAn Thank you so much for the updates, this is great!!!
@3dcinetv The output mesh is reconstructed by sampling SDF of a 3D voxel volume using Marching Cubes. It only creates the vertex + faces + normals for the PLY mesh,...
It uses Trimesh which does support = `Export meshes as binary STL, binary PLY, ASCII OFF, OBJ, GLTF/GLB 2.0, COLLADA`
@3dcinetv Polycount for one of my PLY outputs is 1.04million, it's fairly detailed but obviously not optimised at all and needs clean up, but is watertight. It only uses around...
@3dcinetv Sorry I didn't know your experience. I'm not the developer, I'm just trying to integrate it into existing workflows similar to you. I think the issue with the methods...
On my 3090Ti with WSL2 I got 331.4s for image -> video, and 343s for image -> mesh