Right now, the image generation feature only works with a text seed prompt. But we need an option to upload a reference image and then generate results based on that reference—similar to how Nano Banana works.

For example:

I upload my photo.

I give instructions like “change the background to an office” or “add a blazer.”

The system generates the output while keeping my face and look accurate.

This makes the tool much more powerful for photo editing, product mockups, and personal model creation, compared to only generating from text.