Wan 2.7 Pro Unified Image Editing

Wan 2.7 Pro unified image editing. One node handles style transfers, image edits, and prompt-driven transformations.

Upload your image, write what you want ("Turn this into a cinematic sunset," "Make it look like a watercolor painting"), and hit Run. The model reads your image, interprets the prompt, and outputs a new version. Resolution matches your input automatically, no manual sizing needed.

Built-in before/after comparison so you can see exactly what changed.

How do you edit images with Wan 2.7 Pro?

Upload an image, write a prompt describing the change you want, and run. The workflow auto-detects your image dimensions and passes them to Wan 2.7 Pro. Two toggles matter: thinking mode makes the model reason about your edit before generating, and prompt extend expands short prompts into richer descriptions.

All defaults work out of the box. You only need to do two things: upload your image and write your prompt.

Prompt This is where you tell the model what to do with your image. Be specific about the result you want. "Turn this into a cinematic sunset" works. "Make the background a snowy mountain range" works. The more concrete your description, the closer the output matches your intent.

Thinking Mode (default: on) When this is on, the model takes extra time to reason about your prompt before generating. You get more accurate edits that follow your instructions closely. Turn it off if you want faster results and don't mind the model taking more creative liberties.

Prompt Extend (default: on) Short prompts get expanded into richer descriptions automatically. If you write "sunset," the model fills in lighting, color temperature, atmosphere. Keep this on for quick prompts. Turn it off when you've written a detailed prompt and don't want the model adding its own interpretation.

Negative Prompt Tell the model what to avoid. "Blurry, low quality, watermark" is a common starting point. Leave it empty if you're happy with defaults.

Image Size Set to "custom" by default. The workflow reads your input image dimensions and passes them straight through, so your output matches your input size. No need to touch this unless you want a different output resolution.

Seed Randomized by default. Lock a specific seed when you want to compare results across different prompts while keeping everything else the same.

Second Reference Image (optional) Feed a second image for style reference or blending. Leave disconnected for standard single-image editing.

What is Wan 2.7 Pro image editing good for?

Wan 2.7 Pro handles style transfers, scene changes, and prompt-driven edits in one unified node. It works best when you want to transform the mood, style, or setting of an existing image while keeping its core composition intact.

Product shots that need a new background or lighting mood. Concept art exploration where you have a base composition and want to see it in different styles. Portrait retouching where you want to shift the overall feel without losing the subject.

The built-in comparison node lets you flip between before and after in the same view, which speeds up iteration when you're dialing in a look.

For pure text-to-image generation without a reference, you can disconnect the input image and use it as a standalone generator. But its strength is image-to-image work where you already have something and want to push it somewhere new.

FAQ

What does thinking mode do in Wan 2.7 Pro? Thinking mode makes the model reason about your prompt before it starts generating. This adds processing time but produces edits that follow your instructions more faithfully. For complex multi-part prompts, keep it on. For quick iterations, try turning it off.

Does Wan 2.7 Pro change my output resolution? No. The workflow reads your input image dimensions and feeds them directly to the model. Your output matches your input size unless you manually override the width and height settings.

Can I use Wan 2.7 Pro without an input image? Yes. Disconnect the image input and it works as a text-to-image generator. But the model performs best with a reference image guiding the output.

What kind of prompts work best with Wan 2.7 Pro? Concrete, visual descriptions. "Turn this into a watercolor painting with warm autumn tones" outperforms vague instructions like "make it better." If prompt extend is on, even short prompts get expanded, but specific prompts still give you more control.

How to run Wan 2.7 Pro online? You can run Wan 2.7 Pro online through Floyo. No installation, no setup. Open the workflow in your browser, upload your inputs, and hit run. Free to try.