TongYi AIGC

A powerful Dify plugin providing comprehensive AI-powered image and video generation capabilities using Alibaba Cloud Tongyi's latest Wanxiang, Qwen, and Z-Image models. Supports text-to-image, text-to-video, image-to-image, image-to-video, image translation, and more with professional-grade quality and flexible configuration options.

Version Information

Current Version: v0.0.3
Release Date: 2026-04-26
Compatibility: Dify Plugin Framework
Python Version: 3.12

Version History

v0.0.3 (2026-04-26):
- Added HappyHorse series models for video generation
- Added HappyHorse Text to Video
- Added HappyHorse Image to Video - First Frame
- Added HappyHorse Reference Video
- Added HappyHorse Video Editing
v0.0.2 (2026-04-12):
- Added new Video Continuation tool (wan_video_continue) - Continue video from existing clips using wan2.7-i2v
- Added wan2.7-t2v model support for Text to Video
- Added wan2.7-i2v model support for Image to Video with multi-modal inputs (first-frame, first+last frame, video continuation)
- Added wan2.7-r2v model support for Reference Video with enhanced features
- Added wan2.7-image-pro/wan2.7-image models for Image to Image with sequential grouped output mode
- Added qwen-image2.0 series models: qwen-image-2.0-pro, qwen-image-2.0, qwen-image-edit-max, qwen-image-edit-plus
- Enhanced resolution control with new parameters for wan2.7 models
- Added tags for plugin marketplace (productivity, image, videos)
v0.0.1 (2026-02-16): Initial release with image and video generation capabilities

Quick Start

Install the plugin in your Dify environment
Configure your Alibaba Cloud API credentials (API Key from DashScope)
Start generating images and videos with AI

Key Features

Multiple Generation Modes: Text-to-image, text-to-video, image-to-image, image-to-video, image translation
Latest AI Models: Supports wan2.7, qwen-image2.0, Z-image, wan2.6, wan2.5, wan2.2 for images; wan2.7-t2v, wan2.6-t2v, wan2.5-t2v-preview, wan2.2-t2v-plus for videos
Flexible Image Sizes: Multiple aspect ratios from 1:1 to 21:9 with various resolutions
Video Generation: Create videos with customizable duration (2-15 seconds) and synchronized audio
Image Translation: Translate text in images with AI-powered OCR and translation (14+ languages)
First-Last Frame Video: Create videos from first and last frame images
Reference Video: Generate videos based on reference video style
Video Continuation: Continue existing clips with wan2.7-i2v
Batch Generation: Generate multiple images in a single request (up to 6 images)
Watermark Control: Optional watermark for content authenticity

Core Features

Image Generation

Wan Text to Image (wan_text_2_image)

Generate images from text descriptions using Wanxiang models.

Supported Models: wan2.7-image-pro, wan2.7-image, wan2.6-t2i
Features:
- Multiple aspect ratios (1:1, 3:2, 2:3, 16:9, 9:16)
- High resolution output (up to 4K for wan2.7-image-pro)
- Optional watermark
- Sequential grouped image output mode (wan2.7)
- Thinking mode for enhanced reasoning (wan2.7)
- Prompt intelligent rewriting
- Batch generation (1-4 images; up to 12 in wan2.7 sequential mode)
- Negative prompt support

Wan Image to Image (wan_image_2_image)

Generate images from text and reference images using Wanxiang models.

Supported Models: wan2.7-image-pro, wan2.7-image, wan2.6-image (compatibility)
Features:
- Reference image guided generation (1-9 images)
- Multiple aspect ratios (1:1, 2:3, 3:2, 3:4, 4:3, 9:16, 16:9, 21:9)
- 1K/2K presets for wan2.7 models
- Optional watermark
- Sequential grouped image output mode (wan2.7)
- Interleaved text and image output mode (wan2.6 compatibility)
- Prompt intelligent rewriting
- Batch generation (1-4 images; up to 12 in wan2.7 sequential mode)

Qwen Text to Image (qwen_text_2_image)

Generate images using Qwen image models.

Supported Models: qwen-image-2.0-pro, qwen-image-2.0, qwen-image-2.0-pro-2026-03-03, qwen-image-2.0-2026-03-03, qwen-image-max, qwen-image-plus, qwen-image
Features:
- High quality image generation
- Multiple aspect ratios (16:9, 4:3, 1:1, 3:4, 9:16)
- Optional watermark
- Prompt intelligent rewriting
- Negative prompt support
- Random seed for reproducibility

Qwen Image to Image (qwen_image_2_image)

Generate images from text and reference images using Qwen models.

Supported Models: qwen-image-2.0-pro, qwen-image-2.0, qwen-image-edit-max, qwen-image-edit-max-2026-01-16, qwen-image-edit-plus, qwen-image-edit, qwen-image-edit-plus-2025-10-30
Features:
- Reference image guided generation (1-3 images)
- Multiple aspect ratios (1:1, 2:3, 3:2, 3:4, 4:3, 9:16, 16:9, 21:9)
- Optional watermark
- Prompt intelligent rewriting
- Batch generation (1-6 images)
- High resolution output (up to 2048*872)

Z-Image Text to Image (z_image_text_2_image)

Generate images using Z-Image Turbo model.

Supported Models: z-image-turbo
Features:
- Fast image generation
- Extensive aspect ratio support (1:1, 2:3, 3:2, 3:4, 4:3, 7:9, 9:7, 9:16, 9:21, 16:9, 21:9)
- Multiple resolution options (1024 to 2048 pixels)
- Prompt intelligent rewriting
- Random seed for reproducibility

Image Translation (qwen_image_translate)

Translate text in images with AI-powered OCR and translation.

Supported Models: qwen-mt-image
Features:
- Automatic text detection
- Multi-language translation (14 languages)
- Supported languages: Chinese, English, Japanese, Korean, French, German, Spanish, Russian, Italian, Portuguese, Arabic, Thai, Vietnamese, Indonesian
- Domain hint for improved translation
- Sensitive words filtering
- Terminology support

Video Generation

Text to Video (wan_text_2_video)

Generate videos from text descriptions using Wanxiang models.

Supported Models: wan2.7-t2v, wan2.6-t2v, wan2.5-t2v-preview, wan2.2-t2v-plus, wanx2.1-t2v-turbo, wanx2.1-t2v-plus
Features:
- Duration: 2-15 seconds (model dependent)
- Resolution control: (wan2.7) or (wan2.6 and earlier)
- Synchronized audio generation
- Prompt intelligent rewriting
- Single/Multi shot support (shot_type for wan2.6, prompt-driven for wan2.7)
- Custom audio URL support
- Negative prompt support

Image to Video (wan_first_image_2_video)

Generate video from a single image with text description.

Supported Models: wan2.7-i2v, wan2.6-i2v, wan2.5-i2v-preview, wan2.2-i2v-flash, wan2.2-i2v-plus, wanx2.1-i2v-turbo, wanx2.1-i2v-plus
Features:
- Multi-modal inputs on wan2.7-i2v: first-frame, first+last frame, or video continuation
- Single image input as first frame (legacy models)
- Duration: 2-15 seconds (model dependent)
- Resolution: 480P, 720P, 1080P
- Synchronized audio generation
- Video effect templates
- Prompt intelligent rewriting
- Single/Multi shot support

First-Last Frame Video (wan_first_end_image_2_video)

Generate video from first and last frame images.

Supported Models: wan2.2-kf2v-flash
Features:
- First and last frame input
- Smooth transition generation
- Resolution: 480P, 720P, 1080P
- Video effect templates
- Prompt intelligent rewriting

Reference Video (wan_reference_video)

Generate videos based on reference video style.

Supported Models: wan2.7-r2v, wan2.6-r2v, wan2.6-r2v-flash
Features:
- Multi-modal references (images/videos) input (up to 5)
- Optional first-frame and reference voice support (wan2.7-r2v)
- Duration: 2-10 seconds (or up to 15s for wan2.7 without reference video)
- Resolution: 720P or 1080P (multiple aspect ratios)
- Synchronized audio generation (wan2.6-r2v-flash only)
- Single/Multi shot support
- Prompt intelligent rewriting

Video Continuation (wan_video_continue)

Continue a video from an existing first clip.

Supported Models: wan2.7-i2v
Features:
- Supports and
- Duration 2-15 seconds and 720P/1080P resolution
- Prompt intelligent rewriting and watermark control

Video Query (wan_video_query)

Query the status and results of video generation tasks.

Features:
- Real-time task status
- Video download URL retrieval
- Optional automatic video download

Image Translation Query (qwen_image_translate_query)

Query image translation task status and results.

Features:
- Real-time task status
- Translated image retrieval

Technical Advantages

Latest AI Models: Access to Alibaba Cloud's newest Wanxiang, Qwen, and Z-Image models
High Quality Output: Professional-grade image and video generation
Flexible Configuration: Extensive parameter options for fine-tuning
Async Processing: Efficient video generation with task-based workflow
Multi-Format Support: Support for various image and video formats
Audio Generation: Automatic synchronized audio for videos
Batch Processing: Generate multiple images efficiently
Image Translation: AI-powered text translation in images (14+ languages)

Requirements

Python 3.12
Dify Platform access
Alibaba Cloud API credentials (API Key from DashScope)
Required Python packages (installed via requirements.txt):
- dify_plugin>=0.2.0
- requests>=2.31.0,<3.0.0
- pillow>=10.0.0,<11.0.0

Installation & Configuration

Install the required dependencies:
Configure your Alibaba Cloud API credentials in the plugin settings:
- API Key: Your Alibaba Cloud DashScope API key
Install the plugin in your Dify environment

Usage

Image Generation Tools

1. Wan Text to Image

Generate images from text descriptions.

Parameters:
- : Model version (default: wan2.7-image-pro)
- : Text description of the image (required, wan2.7 <=5000 chars, wan2.6 <=2100 chars)
- : Describe what you don't want (<=500 chars, wan2.6 compatibility)
- : Image size (default: 2K, supports 1K/2K/4K or specific resolutions like 1024*1024)
- : Number of images to generate (1-4, up to 12 with wan2.7 sequential mode)
- : Enable grouped image generation for wan2.7 (default: false)
- : Enable thinking mode for enhanced reasoning on wan2.7 (default: true)
- : Enable prompt intelligent rewriting (default: true)
- : Enable/disable watermark (default: true)
- : Random seed for reproducibility

2. Wan Image to Image

Generate images from text and reference images.

Parameters:
- : Model version (default: wan2.7-image-pro)
- : Text description (required, <=5000 chars)
- : Reference image files (1-9 images, required)
- : Image size (default: 2K)
- : Number of images to generate (1-4, up to 12 when on wan2.7)
- : Enable grouped image generation for wan2.7 (default: false)
- : Enable interleaved output (default: false)
- : Enable prompt intelligent rewriting (default: true)
- : Enable/disable watermark (default: true)

3. Qwen Text to Image

Generate images using Qwen models.

Parameters:
- : Text description (required)
- : Model version (default: qwen-image-2.0-pro)
- : Image size (default: 1664*928)
- : Describe what you don't want
- : Enable prompt intelligent rewriting
- : Enable/disable watermark (default: true)
- : Random seed for reproducibility

4. Qwen Image to Image

Generate images from text and reference images using Qwen models.

Parameters:
- : Text description (required)
- : Reference image files (1-3 images, required)
- : Model version (default: qwen-image-2.0-pro)
- : Image size (default: 1024*1024)
- : Number of images to generate (1-6, default: 3)
- : Enable prompt intelligent rewriting
- : Enable/disable watermark (default: true)

5. Z-Image Text to Image

Generate images using Z-Image Turbo.

Parameters:
- : Text description (required, <=800 chars)
- : Model version (default: z-image-turbo)
- : Image size (default: 1024*1536)
- : Enable prompt intelligent rewriting (default: false)
- : Random seed for reproducibility

6. Image Translation

Translate text in images.

Parameters:
- : Image URL for translation (required)
- : Target language (required)
- : Source language (default: auto)
- : Domain hint for improved translation
- : Sensitive words (JSON array)
- : Terminologies (JSON array)
- : Model version (default: qwen-mt-image)

Video Generation Tools

7. Text to Video

Generate videos from text descriptions.

Parameters:
- : Model version (default: wan2.6-t2v)
- : Text description (required)
- : Describe what you don't want
- : Video resolution (default: 1920*1080)
- : Duration in seconds (model dependent, default: 5)
- : Custom audio file URL
- : Auto generate audio (default: true)
- : Single or multi shot (default: single)
- : Enable prompt intelligent rewriting (default: true)
- : Enable/disable watermark (default: true)
- : Random seed for reproducibility

8. Image to Video

Generate video from an image or continue from a video clip.

Parameters:
- : Model version (default: wan2.6-i2v, supports wan2.7-i2v)
- : Text description
- / : First frame image input
- / : Optional last frame image (wan2.7-i2v)
- : Optional first clip video URL for continuation (wan2.7-i2v)
- : Video resolution (default: 1080P)
- : Duration in seconds (model dependent, default: 5)
- : Video effect template
- : Custom audio file URL (wan2.7-i2v only in first-frame modes)
- : Auto generate audio (default: true)
- : Single or multi shot (default: single)
- : Enable prompt intelligent rewriting (default: true)
- : Enable/disable watermark (default: true)

9. First-Last Frame Video

Generate video from first and last frame images.

Parameters:
- : Model version (default: wan2.2-kf2v-flash)
- : Text description
- : First frame image (required)
- : Last frame image
- : Video resolution (default: 720P)
- : Video effect template
- : Enable prompt intelligent rewriting (default: true)
- : Enable/disable watermark (default: true)

10. Reference Video

Generate videos based on reference video style.

Parameters:
- : Model version (default: wan2.7-r2v)
- : Text description (required, wan2.7 max 5000 / wan2.6 max 1500)
- : Reference URLs (videos or images, semicolon-separated, max 5)
- : Optional first frame image URL (wan2.7-r2v)
- : Optional reference voice URL (wan2.7-r2v)
- : Legacy resolution in (wan2.6; also mapped for wan2.7 compatibility)
- / : Native wan2.7 resolution tier and aspect ratio
- : Duration in seconds (wan2.6: 2-10; wan2.7: 2-10 or up to 15 without reference video)
- : Single or multi shot (wan2.6)
- : Auto generate audio (default: true, wan2.6-r2v-flash only)
- : Enable prompt intelligent rewriting (default: true)
- : Enable/disable watermark (default: true)

11. Video Query

Query video generation task status.

Parameters:
- : Video generation task ID (required)
- : Download video when available (default: false)

12. Image Translation Query

Query image translation task status.

Parameters:
- : Translation task ID (required)

Notes

Video generation is asynchronous; use Video Query to check status and retrieve results
Image translation is asynchronous; use Image Translation Query to check status
Duration limits vary by model (see model documentation in tool descriptions)
Reference images should be under 10MB in size
Watermark is enabled by default for content authenticity
Prompt intelligent rewriting is enabled by default for better results

Developer Information

Author:
Email: [email protected]
License: Apache License 2.0
Source Code:
Support: Through Dify platform and GitHub Issues

License Notice

This project is licensed under Apache License 2.0. See LICENSE [blocked] file for full license text.

Ready to create stunning images and videos with AI?