Hey everyone!
I have a question – is it possible to edit images using an API, for example, with the Gemini 2.0 Flash Experimental model? I’d like to upload a picture of my hand, specify in the API request to generate a tattoo on it, and receive the output as an image.
Has anyone tried something similar or knows if this is possible?
Thanks in advance for any insights! 
Hi @Krzysztof1, It is possible to pass an image to the model and you can add a prompt describing how the image should be modified. while trying to generate the image as output, you have to define the resopnse_modalities to image. For your use case please refer to this gist for code example. Thank You.
I’m working on this now! I’ve never gone this in depth before, so I have all of the AI assistants discussing the best and cheapest ways to accomplish it. I have a bunch of tattoos, and I want my specific tattoos to be placed on my images consistently accurate WITHOUT typing everything into a prompt with a reference image every single time, one tattoo at a time.
I want to use what I learn to create a better workflow when I need a model to retain data and info more accurate and reliably than it is now. Creating projects isn’t working well enough.
Repeating myself is driving me crazy and LoRA training doesn’t work well to get the same tattoo every time. I do have a LoRA almost trained to get all of them correct, but I had to combine 2 Flux models to get this far. LoL
Have you continued with this?