Haddock
    Image Edit
    Instruction-based editing. Pick a model and describe the change.
    Editor
    Disabled
    Model
    Qwen-Image-Edit-2511
    Coming soon
    Inputs
    2 images (recommended) for identity consistency.
    Drift control
    Keep the subject stable while applying the change.
    Prompt
    Describe the edit you want (what to change + what to keep).
    How to use Qwen-Image-Edit-2511
    What it is: instruction-based image editing with strong identity consistency and geometry reasoning. Ideal for character tweaks, wardrobe changes, or multi-person edits.
    Recommended input: provide 1 “base” image, plus an optional second reference image if you need to preserve identity/style (face, outfit, hairstyle).
    Prompt format: be explicit about what to change and what must stay the same.
    Example prompts
    Change the outfit to a black leather jacket. Keep the same face, hairstyle, pose and background.
    Make it sunset lighting with warm tones. Preserve identity, facial features and clothing.
    Replace the background with a cyberpunk street at night. Keep the subject unchanged.
    Good to know: the model can drift if the instruction is vague. If the subject changes too much, add stronger constraints like “same face”, “same person”, “keep clothing”, and include a second reference image.
    This page is ready for UI/UX. API integration will be added once the inference endpoint is available.
    Preview
    Qwen-Image-Edit-2511
    Multi-image edit with strong identity consistency & geometry.
    Planned inputs
    1-2 images
    Edit instruction prompt
    Steps / guidance / seed (planned)