Wan 2.7 Image

About the model

Wan 2.7 Image is Alibaba’s model that unites image generation and editing in a single tool. Without sources it creates an image from your description; upload one or more frames and it edits them by text instruction — replacing objects, changing lettering, transferring style, or assembling a scene from several references.

The model stands out for fine portrait control: you can set bone structure, eye shape, contour, makeup, hairstyle, and accessories so faces don’t look generic. It also handles long text in 12 languages with confidence, which suits charts, tables, reports, and infographics.

NanoBananana offers two versions. Wan 2.7 Image is the fast, versatile option for everyday generation and edits at up to 2K. Wan 2.7 Image Pro focuses on higher quality, also at up to 2K.

Strengths

• Text-to-image generation and prompt-driven editing in one model
• Up to 9 source images for edits, style transfer, and reference fusion
• Fine portrait control — facial features, eye shape, makeup, hairstyle, accessories
• Long in-image text in 12 languages — charts, tables, infographics
• 1K or 2K resolution in both versions
• Eight aspect ratios, including ultra-wide 8:1 and 1:8

Best for

• Ad visuals, banners, and social media graphics
• Portraits and avatars with carefully shaped looks
• Product shots and product cards
• Merging several references into one frame
• Infographics and text-heavy materials

Limitations

• Maximum resolution is 2K
• Up to 9 sources, each up to 10 MB and at least 240 pixels per side
• Aspect ratio applies only when generating without source images
• Prompts are limited to 5,000 characters

Prompting tips

• To edit, upload a source and describe what to replace and what to keep
• With several sources, explain each one's role — object, style, background
• Describe looks specifically — face shape, eyes, hairstyle, makeup
• For final-quality output, pick Pro at 2K