GenPix AI · Image to Image
Image to Image
Image to image starts from a picture instead of a blank page. You give the model something to hold on to — a composition, a face, a product, a room — and tell it what to change.
The useful mental model: the reference is the anchor, the prompt is the instruction. Say what should stay as much as what should change. "Keep the pose and the face, change the background to a rainy street at night" gets you far closer than a description of the final image alone.
- 2 images free — no account, no card, no watermark
- Private by default — nothing you make is ever public unless you publish it yourself
- Never used for training — not your images, not your prompts
- What you make is yours — full resolution, commercial use, we take no cut
Made here, with the sentence that made it
Every picture here came out of this studio. Hover one to read the exact sentence that made it, press it to open the studio with that sentence already loaded.
-
The outfit, on you, before you buy itFull-length street photograph of a woman in a camel wool coat over a black turtleneck, straight-leg jeans and leather boots, walking down a quiet European street in… -
Your own hero, your own suitFull body photograph of a masked woman in a matte black feline vigilante suit with pointed ears and a clawed glove, crouched on a wet city rooftop at night, neon signs… -
The costume, tried on earlyAtmospheric portrait of a woman in an elaborate dark fairy Halloween costume with black feathered wings and silver face jewels, standing among misty gravestones with…
How to do it
- 1
Upload one or more reference images.
- 2
Write the change as an instruction, naming what must stay untouched.
- 3
Iterate: take the result, feed it back in, refine again.
Prompts that work
Press one to open the studio with it loaded — then change two words rather than face a blank box.
Keep the subject and pose exactly, replace the background with a foggy pine forest at dawnSame room, same furniture, change the season outside the window to deep winterCombine these two images: the jacket from the first, worn by the person in the second
Questions
- How many references can I use?
- Up to ten in one generation.
- Does it keep faces consistent?
- Largely, yes — that is exactly what our engine is chosen for. Say explicitly that the face must be preserved.
- What file types can I upload?
- PNG, JPEG and WebP, up to 12 MB each.