I never knew this whole time after all this researching and prompt engineering. I was trying to do prompt framework. I just found a video today:
Give credit to this guy it's amazing!!
Description
When you upload an image to the vlm type "AUTO_BUILD" AND THE VLM WILL CREATE THE TAGS FROM THE IMAGE. THEN YOU COPY THE FINAL OUTPUT AND USE IT AS AN SYSTEM PROMPT THEN WHEN YOU TYPE YOUR IDEA IT WILL WRAP THE FRAMEWORK AROUND YOUR IDEA. ITS AWESOME😎🤯🧠😄✍️😅