Alibaba's Qwen Team has released an open-source AI model called Qwen-Image-Edit, which can achieve professional-grade image editing tasks with text inputs alone.
Key Takeaways:
- Qwen-Image-Edit builds on the Qwen-Image foundation model and extends its capabilities to cover a wide spectrum of editing tasks, from subtle appearance changes to broader semantic transformations.
- The model supports both English and Chinese inputs, and its dual-encoding approach allows for edits that remain faithful to both the intent of the prompt and the look of the original image.
- Qwen-Image-Edit is available as an open-source code under an Apache 2.0 license, making it free for enterprises to use and potentially resulting in cost savings from proprietary software.