Why doesn't ChatGPT use the actual photos I provide?
Asked by
DoggoDreamer99
I have a German Shepherd who missed out on a class photo with a dog group because she can't be off-leash yet. I tried to get ChatGPT to combine two photos I took, asking it specifically to use those images. Instead, it keeps generating its own AI-created version of the dogs that aren't mine. I've repeated my request, but it still doesn't use the actual photos. What am I doing wrong?
Answers
That's just how it operates. It's not copying files; when you provide a photo, it's looking at it and generating something based on its training data. Sometimes it gets a bit far off from what you're hoping for, so maybe try to refine your prompt a bit more.
Reply to this answer
ChatGPT doesn't really use your photos directly; it reimagines them based on what it understands. For combining photos just as they are, you might need to check out other apps designed specifically for that.
Reply to this answer
You could ask ChatGPT directly why it behaves this way. It works off of tokens and general ideas from its training. It identifies what objects are in the picture, like recognizing it's a dog, but then draws from its learned data, which might not represent your specific dog. You could try to prompt it for a more accurate result, but it might still struggle because of how it 'thinks'.
Reply to this answer
Add your answer
Answers are reviewed before publication.
TabbyCatLover17: Right! I got lucky with my cat since she's a gray tabby, similar to a popular brand's bag, so it managed to capture her pretty well. That’s probably because it has loads of images like her in its data.