[reedf1]

If you are seriously interested and this isn't just some automated slop experiment you should look into and understand how VLMs and vision encoders work.