Before
AfterSingle image edit
Change one clear part of an image
Best for changing clothing, background, material, light or a small object. First say what must stay, then describe the replacement.
Wan 2.5 Image-to-Image
A Wanxiang image editing model for single-image edits, multi-reference fusion and controlled visual variations.
Wan 2.5 Image-to-Image
Keep the value, then change the image
Wan 2.5 Image-to-Image is for continuing from existing images: upload one or more references, describe what to keep, what to combine and what to change, then review a controlled visual variation instead of a full redraw.
Single-image editReference fusionSubject consistencyVisual variations
Change requestKeep the product shape, switch to cleaner commercial lighting
- Do not redraw the subject
- Make the light consistent
- Ready for product display
Best-fit tasks
Where Wan 2.5 Image-to-Image works best
Use it when the source image already has value and you want a controlled change, a reference blend or a first set of related directions.
BeforeAfter
Before
AfterMulti-image fusion
Combine two or three references into one scene
Give each reference a job: one for subject, one for scene, one for material or style, so the result has less conflict.
Before
AfterImage set direction
Extend one reference into related directions
Useful for first-round product scenes, portraits, social covers or campaign concepts before polishing the winning option.
Brief structure
The brief should separate what stays from what changes
Wan 2.5 Image-to-Image works best when the prompt reads like an edit note: preserve the valuable parts, name the edit area and explain how the result will be used.
Keep
Subject, angle, logo, face, material, composition and light direction.
Change
Describe only the area that should change and keep unrelated ideas out.
Blend
Ask for natural edges, shadows, perspective and color matching.
Use
State whether it is for ecommerce, poster, social cover, profile or detail page.
Flux Art
How to use Wan 2.5 Image-to-Image in Flux Art
Give each reference a clear role: subject, scene, style or material. Then state what stays, where the edit happens and how the image will be used so the result is easier to judge and refine.
Edit with Wan 2.5- 01Upload the reference
Start with one valuable source image, or add a small set of references when each image has a clear role.
- 02Write the edit note
State what must stay, what should change, how references should blend and where the result will be used.
- 03Review the boundary
Check subject identity, edges, shadows, perspective and whether the result still feels connected to the original image.
Prompt templates
Wan 2.5 Image-to-Image prompt templates
Replace the bracketed parts with your own product, person, scene or style reference. Keep each prompt focused on one edit goal.
Product scene replacement
Use the uploaded [product] image as the main subject. Keep its shape, color, logo position and material unchanged. Replace the background with [new scene], add natural shadows and matching light, keep the product centered and ready for an ecommerce hero image.
Poster copy refresh
Keep the original poster composition, main visual, color tone and empty space. Update the headline area to fit [new campaign message], keep the layout clean and readable, make the result feel like the same campaign asset revised for a new promotion.
Two-reference fusion
Use image 1 for the [subject], image 2 for the [scene or background]. Keep the subject recognizable, place it naturally inside the scene, match perspective, shadows and color temperature, and produce a believable final image for [final use].
Compare related AI image editing models
- Wan 2.6 EditA reference-first Wanxiang image model for controlled edits, multi-image guidance and text-image layouts.
- Wan 2.5 Text-to-ImageA Wanxiang image model for realistic light, design-ready layouts and style-rich visual drafts.
- Wan 2.6 Text-to-ImageA text-first Wanxiang image model for prompt-following, realistic scenes and style exploration.
- Qwen Image Edit MaxA precision image-editing model for text, objects, poses, styles and consistent visual revisions.
Wan 2.5 Image-to-Image FAQ
What is Wan 2.5 Image-to-Image?
Wan 2.5 Image-to-Image is a Wanxiang image editing model in Flux Art for editing existing images, keeping the subject recognizable and combining references into new visual directions.
What should I use it for first?
Start with a simple edit: keep the product, person or layout from the source image, then change one clear element such as background, material, style or placement.
How should I write the prompt?
Write what must stay, what should change, how references should blend and where the final image will be used.