Currently if you use a character from Model trainer and there is more than one person in the image, each person in the image is that character. Is it possible to have the image generator only use the consistent character once in the image, so all others are random or as described in the prompt??