When M3 Sees a Part — M4 Creates the Whole

The M4 neural network model can extend an image. For example, if we have a photo of a girl in a swimsuit where her body is not fully visible — her legs are missing — the M3 model will generate a 3D object based only on the visible part, resulting in an incomplete model.

The M4 model, on the other hand, can reconstruct and generate the missing parts, such as the legs.

The example below shows the original image, along with two meshes — M3 and M4. M4 includes the legs that are missing in the original photo.

Links to both 3D objects are provided below.

M4 - https://meshzeus.com/model/95f4d020-de42-4179-a02f-4696ae5cac2e

M3 - https://meshzeus.com/model/2af43e21-af32-4950-be3d-4b5de44177ec