Model trained on pictures of +-150 people , the model can generate images with averaged appearance from the whole sample. Emphasis on dark hair color.