1.generate a picture with machine learning
2.in VoC use machine learning -> 3D -> depth mapping 
3. displace the machine learning prompt with the depth map
4. do a boolien substract cut