Since only 7 images are used for training, the characters'poses or composition of the generated results may be somehow similar.
