ComfyUI
Node-based image generation you can save, share and rerun exactly.
Image
Google's image model that edits a photo while keeping the same face, product and lighting.
Nano Banana is Google's image editing model, reached through Gemini and the API. You give it a photo and an instruction — move the product to a marble surface, change the jacket colour, put this shirt on that model — and it edits rather than regenerates, so identity and lighting survive.
That consistency is what makes it commercially useful: catalogue variants, model swaps and background changes that used to need a photographer. It is less good at inventing a scene from nothing, where dedicated text-to-image models still lead on style.
Main job: Edit existing product and lifestyle photos without a studio reshoot.
Catalogue variants, model swaps and background changes at volume.
Start from the highest-resolution original you have, describe one change per request, and keep the previous output as the new input for the next edit. Batch the same instruction across a product set to keep a catalogue coherent.
Watch out: it will happily put a real person's face on a body they never agreed to. For models and endorsements you need consent, and Google's own watermark and provenance metadata should stay in place unless you have a licence that says otherwise.