8 Hex
Color Palette Control
Wan 2.7 Image — By Alibaba Tongyi Lab
Wan 2.7 Image. Generate and Edit with Precision Control.
Wan 2.7 Image is a unified text-to-image and editing model from Alibaba. Distinct faces, hex-based color palettes, accurate text rendering, multi-reference composition, and region-level editing — all in one workflow.
Built for designers, brands, and content teams.
Face Diversity8-Hex PaletteText Rendering9-Ref InputRegion Editing4K Output
9 Refs
Reference Inputs
4,000
Text Characters
12 Images
Grouped Output
Wan 2.7 Image Gallery
See What Wan 2.7 Image Creates.
Generated with Wan 2.7 Image — text-to-image, multi-reference composition, and region editing.
What Is Wan 2.7 Image
Wan 2.7 Image — A Controllable Visual Creation System.
Wan 2.7 Image is Alibaba Tongyi Lab's unified image generation and editing model, released in April 2026. Unlike generation-only models, it combines text-to-image, multi-reference composition, and region-level editing in one workflow — so you control the output instead of prompting and hoping.
Use Wan 2.7 Image when brand accuracy, facial consistency, or text legibility actually matter. It fits e-commerce product shots, storyboard panels, marketing visuals, educational diagrams, and any workflow where one-shot randomness is too costly.
01
Precise Face Control
Generate distinct, consistent characters across outputs — not the same AI face recycled with minor variation.
02
Hex-Based Color Palettes
Define up to 8 Hex color codes per generation to anchor outputs to exact brand or aesthetic requirements.
03
Accurate Text Rendering
Render up to 4,000 characters — including formulas, tables, and multilingual text — directly inside generated images.
04
Region-Level Editing
Select exactly what to change and edit it in isolation. No full regeneration needed for small adjustments.
How It Works
From Prompt to Polished Image in Three Steps.
Describe, generate or edit, then export — with full control at every stage.
01
Describe Your Image
Write a text prompt or upload up to 9 reference images to guide the output. Add hex color codes to lock your palette. The more specific your inputs, the closer the result to your intent.
Use reference images for characters and hex codes for brand-accurate colors together — they stack.
02
Generate or Edit
Generate a new image from text or references, or load an existing image and use region selection to edit just the area you need. Wan 2.7 Image processes both in the same interface.
For edits, draw a tight selection around only the element you want to change. Smaller selections produce cleaner results.
03
Export and Use
Download your image in 4K resolution or as a transparent-channel PNG to separate subjects from the background. Output is ready for e-commerce, social media, print, or API integration.
Use the transparent PNG export for product shots — it skips the manual background removal step entirely.
Wan 2.7 Image Features
Why Choose Wan 2.7 Image?
A unified generation and editing model built around control, not chance.
01
Face Diversity — No More AI Same-Face
A thousand faces. Not the same one repeated.
Wan 2.7 Image generates genuinely distinct characters across outputs. The model is built to avoid the recycled-face problem that makes AI-generated people look identical. For brands, storytellers, and studios that need a cast instead of a clone, this changes the workflow.
02
8-Hex Color Palette Control
Define the palette. The model follows it.
Input up to 8 Hex color codes to lock the palette before generation. Brand colors stay accurate without post-processing corrections.
03
Text & Formula Rendering
Legible text. Inside the image. First try.
Render up to 4,000 characters of text — including tables, math formulas, and multilingual content in English, Chinese, Japanese, and Korean — directly inside the generated image.
04
Multi-Reference Composition
9 references in. 12 consistent images out.
Upload up to 9 reference images as input and generate up to 12 grouped outputs in one run. Subject consistency holds across the full set — useful for storyboards, campaign series, character sheets, and book illustration spreads.
05
Region-Level Editing
Edit the element. Keep the rest.
Draw a selection around any part of an existing image and edit only that area. No full regeneration. No losing what is already working.
06
Transparent PNG Export
Clean cutouts. Built in.
Export any subject as a full transparent-channel PNG, with the subject cleanly separated from its background. Ready for product listings, composite work, and design layouts without a separate cutout step.
07
Wan 2.7 Image Pro — Stronger Composition
Bigger model. Tighter results.
Wan 2.7 Image Pro is trained on a larger dataset with a larger model size. It produces more stable composition and stronger semantic understanding — better suited for complex layouts, dense text, and multi-subject scenes where the standard model may drift.
08
API Access — Live Today
Build with it. Not just use it.
Both wan2.7 image and wan 2.7 image pro are available via API. Integrate generation and editing into your own tools, pipelines, or platforms at $0.03–$0.075 per image.
Use Cases
Who Uses Wan 2.7 Image?
From product listings to storyboards — precise image generation for teams that can't afford inconsistency.
01
Film & Storyboard
Turn Concept Art Into Consistent Scene Panels
Generate up to 12 storyboard panels in one run with shared character and style references. Lock faces, colors, and settings across the full sequence without manual iteration.
02
Social Media & Content
Build a Visual Identity That Stays On-Brand
Use hex palette control and face reference to produce on-brand content across every post. No color drift, no inconsistent characters — just repeatable, scroll-stopping visuals.
03
E-Commerce & Advertising
Product Shots With Clean Backgrounds — No Retouching
Generate product images with accurate colors and transparent PNG export. Edit specific elements without reshooting. Publish faster with less post-production.
04
Game & Character Design
Character Sheets With Real Facial Diversity
Generate a full cast of visually distinct characters across multiple angles and expressions with consistent identity. No more art-directing AI to not repeat the same face.
05
Design & Brand Studio
Client Mockups With Exact Brand Colors
Define client palettes in hex, upload brand reference images, and generate on-brand visuals in a single pass. Region editing handles revision rounds without full regeneration.
06
Education & Publishing
Illustrated Diagrams With Accurate Text and Formulas
Generate educational visuals with embedded tables, math formulas, and multilingual captions. Text renders correctly inside the image — no separate overlay needed.
What Creators Are Saying
Real Results With Wan 2.7 Image.
“This feels less like a pure image model and more like a controllable visual creation system. If the real-world performance matches the demo, Wan 2.7 Image could be very strong for branding, storytelling, and content production.”
Alisa Q.
AI Content Strategist
“The hex palette control is the feature I didn't know I needed. I can finally stop correcting AI-generated colors in post. For client work, this alone saves an hour per project.”
Marco D.
Brand Designer
“Text rendering inside AI images has always been a deal-breaker for educational content. Wan 2.7 Image is the first model where I can generate a diagram with actual readable formulas and not have to rebuild it in Figma afterward.”
Priya S.
Instructional Designer
“Region editing changes everything for e-commerce. I can generate a product shot, swap the background, fix a label detail, and export a transparent PNG — all without regenerating from scratch. That's a real workflow, not a demo.”
James T.
E-Commerce Creative Lead
“Character consistency across 12 panels with the same face, same style, and my exact color palette? That used to take a full day of iteration. Now it's one run with reference images and hex codes.”
Yuki H.
Manga Artist & Illustrator
Wan 2.7 Image FAQ
Wan 2.7 Image — Frequently Asked Questions.
What is Wan 2.7 Image?
Wan 2.7 Image (wan2.7 image) is Alibaba Tongyi Lab's unified image generation and editing model, released in April 2026. It combines text-to-image generation, multi-reference composition, and region-level editing in one workflow, with built-in controls for face diversity, color palettes, and text rendering.
What is the difference between Wan 2.7 Image and Wan 2.7 Image Pro?
Wan 2.7 Image Pro is trained on a larger dataset with a larger model size. It produces more stable composition and stronger semantic understanding, making it better suited for complex scenes, dense text layouts, and multi-subject images. The standard model covers most generation and editing tasks well at lower cost.
How does face control work in Wan 2.7 Image?
The model is built to generate distinct characters across outputs rather than recycling a single average face. You can provide reference images to anchor a specific character's appearance, and it holds that identity more consistently than standard diffusion models across multiple generations.
Can Wan 2.7 Image render text and formulas inside generated images?
Yes. Wan 2.7 Image supports up to 4,000 English characters, along with tables, mathematical formulas, and multilingual text in Simplified Chinese, Traditional Chinese, English, Japanese, and Korean — rendered directly inside the generated image without a separate text overlay step.
How does region editing work?
You load an existing image, draw a selection around the area you want to change, and describe the edit. The model applies the change only to the selected region while preserving the rest of the image. This is faster than full regeneration for small adjustments like background swaps, object replacements, or color corrections.
What is the API pricing for Wan 2.7 Image?
API access is live for both wan2.7 image and wan 2.7 image pro. Pricing starts at approximately $0.03 per image for the standard model and $0.075 per image for the Pro model, though exact rates may vary by platform. Check wan27.org for current pricing.
Is Wan 2.7 Image open source?
The current image model is available via API and the official Wan web interface. An open-source release for the image model has not been confirmed. The earlier Wan video models were open-sourced, so a future open release is possible but not announced.
Is Wan 2.7 Video available now?
Yes. Wan 2.7 video workflows are now available on wan27.org, including text-to-video, image-to-video, reference-to-video, and video editing. The image model remains available for text-to-image generation, image editing, and multi-reference visual creation.
Still have questions? Contact us
No credit card required. Free generations included.
Start Creating with Wan 2.7 Image
Precise faces, exact colors, legible text, and region editing — all in one model.
No credit card requiredFree generations includedAPI access availableCommercial license included