We are excited to announce that image generation on SCRP-Chat is now powered by Qwen-Image, along with a new, in-house-developed image editor powered by Qwen-Image-Edit. With this update, SCRP-Chat is now offering all users frontier-level image generation capabilities.

A panda in front of CUHK

Image Generation

Image generation can be enabled by clicking on the Image button below the chat input box. Compared with the previous Stable Diffusion model, Qwen‑Image is far better at understanding styling instructions and accurately rendering text.

Image Editing

2026 Note: the image generation and editing features have been consolidated. You can turn on image generation and editing by first clicking the diamond icon at the bottom of the prompt textbox, followed by toggling the ‘Image’ switch.

The follow instruction is no longer valid:

Image editing can be enabled by selecting Qwen Image Editor from the model selection dropdown menu at the top. The editor supports both image upload and image URLs. It also supports multi-round editing.

Tips for Getting the Best Result

To get the best results out of the image generator and editor:

  • For the best quality, include everything you need in a single prompt instead of doing multi-round editing.
  • The Qwen model is not very capable of deducing the style of an existing image, and appears to default to generating photorealistic elements. When you add new elements to an existing image that is not photorealistic, specify the style you want.

What’s Next?

The Qwen team released Qwen-Image-Edit-2509 last week, with the most significant change being the support of using multiple images as input. We are currently working on incorporating this feature.