WAN 2.7 Image

Turn a prompt into a picture. Give an existing image a new direction.

Generate from text or upload a reference to create and edit images with WAN 2.7 Image.

0 / 5,000

Up to 9 references · JPG, PNG, WebP · At least 240 × 240 px · 10 MB per image after processing. Large files are compressed; HEIC is converted.

Start with a visual direction

Explore a setting, a product concept, or an illustrated scene. These AI-generated images are creative references, not WAN 2.7 Image test results.

Sunlit living room with large windows, a sage green linen sofa, and a leafy plant — A different setting
Describe the room, materials, and light before adding smaller details.

Clear skincare dropper bottle with a silver collar, white pipette, pale pink liquid, and water droplets — A product concept
Explore a product scene, then review the result against your real item.

Illustrated woman sitting above a city beneath a colorful sunset sky — An illustrated moment
Combine a subject, setting, and visual style in one focused prompt.

Create from words. Edit with a reference.

WAN 2.7 Image brings text-to-image generation and reference-guided image work into one tool. Start with a description when you have a new idea, or add a picture when you already have a visual starting point. Image-to-image and reference editing use the same upload workflow; the scope of the change comes from your instructions.

Text-to-image generation

Describe the picture you want to make: the subject, its surroundings, the lighting, and the style. WAN 2.7 Image works from that prompt without needing a source photograph. Use it to explore an illustration, sketch out a campaign concept, or try a composition before preparing a finished design. You can choose an available aspect ratio to match the intended placement. A square product concept and a tall editorial image may need different framing, even when they share the same subject.

Start with the details that define your idea. Add a camera angle or material only when it changes the picture you want. Review the output before deciding which part of the prompt to refine.

A sunlit living room with a sage linen sofa and oak coffee table — Prompt
A sunlit living room, sage linen sofa, oak coffee table, natural materials.
AI-generated illustration, not a WAN 2.7 Image test result.

Image-to-image generation

Add a source image and describe a new direction for it. You might explore a different setting, change the atmosphere, or ask for an illustrated treatment of a photograph. The reference provides visual context that can be difficult to express in words alone. Your prompt tells the model how to interpret that context, so explain both the desired change and the elements you would like to retain.

A source image guides generation; it does not lock every detail. Compare the output with your original, especially if the shape, proportions, or markings of a real object matter to your project.

Watercolor interpretation of the same living room and furniture arrangement — Prompt
Reimagine this room as a watercolor illustration.
AI-generated illustration, not a WAN 2.7 Image test result.

Reference-image editing

Use a reference when the task is a specific change rather than an entirely new idea. Ask for a different background, a new color for an object, or a change in lighting. State the edit directly and describe what should remain. For example: change the background to a pale green studio wall and retain the bottle shape. This is instruction-guided image generation, so changes can extend beyond the exact area named in your prompt.

The WAN 2.7 image editor can help you explore an edit, but it is not a guarantee of pixel-perfect preservation. Inspect faces, edges, labels, and any detail that must remain accurate before using the result.

Glass dropper bottle on pale stone against a pink background — Prompt
Change the background to pale mint green. Keep the bottle and stone setting.
AI-generated illustration, not a WAN 2.7 Image test result.

The same glass dropper bottle and stone setting against a mint green background — Prompt
Change the background to pale mint green. Keep the bottle and stone setting.
AI-generated illustration, not a WAN 2.7 Image test result.

From your first prompt to your next image

01
Choose your starting point
Enter a prompt for a new image, or select a reference photo for image-to-image work. You can prepare the input before signing in. Choose a clear source image with the subject visible rather than a screenshot full of unrelated elements.
02
Set the direction
Select the available model and image settings, then check the credit cost on the Generate button. You can choose the output aspect ratio with or without a reference image. Keep the first request focused so you can evaluate what changed.
03
Generate when you are ready
Click Generate. If you are signed out, sign in to continue. If your balance is too low, the credit purchase dialog opens. Your draft is kept for your return. When signed in with enough credits, your request is submitted directly from this page.
04
Review and refine
View the completed image below the tool. Check the composition and important details, then adjust your prompt for another attempt. Each new generation uses credits. Download the result when it suits your task, or use it as a reference for a further edit.

Give your prompt a clear job

Useful prompts describe a visual decision, not just a collection of impressive-sounding words. Use the following guidance as a starting point for WAN 2.7 image generation, and adjust it to the image you are actually trying to make.

A red bicycle against a gray brick wall under neutral daylight — Prompt
A red bicycle against a brick wall. Change only the light: soft daylight to golden hour.
AI-generated illustration, not a WAN 2.7 Image test result.

The same red bicycle and brick wall in warm sunset light — Prompt
A red bicycle against a brick wall. Change only the light: soft daylight to golden hour.
AI-generated illustration, not a WAN 2.7 Image test result.

Name the subject and its surroundings

A red bicycle leaning against a brick wall at sunset gives more direction than a bicycle. State what is in the scene, where it is, and what should attract attention. If you need space for a headline, describe an uncluttered area rather than asking the model to guess the layout.

Make editing instructions explicit

When working with a reference, name the change first. Replace the background with a softly lit studio is more focused than rewriting every visible detail. Add the features you want retained, and check the generated image rather than assuming those features will be unchanged.

Change one decision at a time

If the composition works but the mood does not, revise the lighting or palette before replacing the entire prompt. Keeping the rest of the request consistent helps you compare iterations. Avoid combining contradictory styles or several unrelated edits into one instruction.

Review before you publish

Treat generated product scenes, campaign concepts, and illustrations as material to inspect. Look closely at text, object shapes, and fine details. For a real product listing, compare the image with the item you sell. A pleasing scene is not by itself proof that every depicted feature is accurate.

WAN 2.7 Image questions, answered

What is WAN 2.7 Image?

WAN 2.7 Image is an image model for generation and editing. Imya provides an online tool for using it with a text prompt or reference images. This page focuses on still images: creating new pictures, exploring variations, and making instruction-guided edits.

Do I need an account before writing a prompt?

No. You can type a prompt and choose reference images first. Sign-in is required when you click Generate. The tool saves your draft before opening login so you can return to the same image task without starting from a blank input.

How many credits does a generation cost?

The Generate button shows the cost for the selected model and your account plan. Adding a reference does not switch your model. Selecting Pro can change the cost, so check the amount before generating. Each submission on this page requests one image.

Do reference images require Pro?

No. WAN 2.7 Image is selected by default, and both Standard and Pro support text-to-image and reference-image editing. Both offer 1K and 2K output. Pro also offers 4K for text-to-image without references. Adding a reference to a 4K request changes the resolution to 2K with a notice, not the model.

Which reference images can I upload?

Add up to 9 images with both width and height at least 240 px. The API accepts JPG, PNG, and WebP up to 10 MB each. The tool compresses large files for upload and converts HEIC or HEIF to a supported format. Files over 50 MB must be reduced first. Images that are too small or cannot be processed are identified before submission; your originals are unchanged.

What happens if my credits are insufficient?

A credit purchase dialog shows the current balance and the amount needed for the request. Your prompt and reference images remain in the saved draft. After returning, review the current settings and click Generate when you are ready. Purchasing credits does not itself submit a generation.

Are image-to-image and reference editing separate modes?

They share the reference-image upload flow on this page. Use the prompt to ask for a broader reinterpretation or a more specific edit. Neither approach guarantees that every unmentioned detail will remain unchanged, so compare the output with the source image.

Can I generate an image without a reference?

Yes. Leave the reference area empty and describe your picture in the prompt. This uses text-to-image generation. Add a reference when visual context is important, such as an existing composition or an object you want the model to consider.

What should I do if the result is not what I expected?

Identify the most important mismatch and revise that part of your prompt. A clearer source image or a narrower editing request may help. Review the credit cost before generating again. If a task is still processing, check its status instead of submitting the same request as a new generation.

Your next image starts here.

Bring a prompt, a reference, or an idea worth exploring.

Create with WAN 2.7 Image

Explore more image tools Privacy policy

Create from words. Edit with a reference.

From your first prompt to your next image

Choose your starting point

Enter a prompt for a new image, or select a reference photo for image-to-image work. You can prepare the input before signing in. Choose a clear source image with the subject visible rather than a screenshot full of unrelated elements.

Set the direction

Select the available model and image settings, then check the credit cost on the Generate button. You can choose the output aspect ratio with or without a reference image. Keep the first request focused so you can evaluate what changed.

Generate when you are ready

Click Generate. If you are signed out, sign in to continue. If your balance is too low, the credit purchase dialog opens. Your draft is kept for your return. When signed in with enough credits, your request is submitted directly from this page.

Review and refine

View the completed image below the tool. Check the composition and important details, then adjust your prompt for another attempt. Each new generation uses credits. Download the result when it suits your task, or use it as a reference for a further edit.