A picture instead of a long prompt
The reference carries the style and composition, so you describe less. Some models take a prompt as well, for what should differ.
Use any picture as a reference for style, colour, framing and scene, and get a new image that closely matches it. This is not an editor — it makes a new picture.
Reference-to-image makes a new picture that follows one you already have — its style, palette, framing and setting — rather than editing the original. Add a photo of your character to place them into the reference scene. Fourteen models are available, and each shows its price before you generate.
The reference carries the style and composition, so you describe less. Some models take a prompt as well, for what should differ.
Add a photo of your character and the result places them into the reference scene.
Including Seedream 5 Pro, GPT Image 2.5 and Nano Banana Pro. Each shows its price before you generate.
Add the picture whose look you want to follow.
Pick a model, and add a character photo if the result should feature them.
Generate and keep the frames that work.
No. The editor changes your picture; reference-to-image makes a new one that follows its look.
Yes. Add a photo of the character and the new image places them into the reference’s scene and style.
From 1 credit (≈ $0.06) an image, depending on the model and size. The price is shown before you generate.
Priced per image by model and size, shown before you generate.