Really good. This is just geometric analysis though. Geometric in the sense that the model likely doesn't understand what it's rendering. All it sees is some shape.
The next step is geometry with organized contours that make sense, meaning that the model needs to cohesively understand the picture and not just the geometry. For example, if a person in the picture is wearing armor the model generates two separate models overlayed on one another, the armor and the mesh.
The next step is geometry with organized contours that make sense, meaning that the model needs to cohesively understand the picture and not just the geometry. For example, if a person in the picture is wearing armor the model generates two separate models overlayed on one another, the armor and the mesh.