Qwen 2511 Local Character Replacement Workflow

This workflow is designed for Qwen Image Edit 2511 local character replacement and reference-guided partial image editing. Its main purpose is to replace or modify a selected character region inside an existing image while preserving the original background, composition, lighting direction, camera perspective, and overall visual structure as much as possible.

Unlike a full image-to-image redraw workflow, this setup is focused on localized replacement. The workflow does not try to regenerate the entire picture from scratch. Instead, it separates the main image, mask area, reference information, crop region, and final pasted result into a more controlled editing pipeline. This makes it useful when creators want to change only one character, one body region, one outfit, one foreground subject, or one masked object while keeping the rest of the image stable.

The workflow uses qwen_image_edit_2511_fp8mixed.safetensors as the main image-editing model, qwen_2.5_vl_7b_fp8_scaled.safetensors as the visual-language encoder, and qwen_image_vae.safetensors as the VAE. It also uses the Qwen-Edit-2511-Lightning 4-step LoRA, allowing faster image editing while keeping the process practical for repeated testing. This makes the workflow suitable for creators who need fast iteration instead of waiting for a long full-resolution redraw every time.

A key part of the workflow is the QwenEditConfigPreparer and TextEncodeQwenImageEditPlusCustom route. The input image, mask, reference configuration, prompt, and instruction are organized into a Qwen edit structure. The instruction logic asks the model to understand the key features of the input image, then modify it according to the user’s editing request while maintaining consistency with the original image where appropriate. This is important for character replacement because the edited result must not look like a separate pasted sticker.

The workflow also includes a practical mask and crop system. The source image is resized, the masked area is prepared, and the selected region can be cropped out for more focused editing. After Qwen generates the edited result, the workflow uses crop restoration logic to paste the edited region back into the original frame. This helps reduce unnecessary changes outside the target region and makes the workflow more suitable for controlled local editing.

Another important feature is the reference / comparison structure. The workflow includes image preview, mask preview, image comparer, and image reel composition. This lets users inspect the main image, the reference, the mask, and the final result side by side. For local character replacement, this is useful because the success of the result depends on both the edit quality and how naturally the new character blends into the original scene.

The workflow also uses background removal and mask-growth tools, including rembg-style foreground extraction and mask expansion / blur logic. These tools help prepare cleaner character masks and softer transition edges. When replacing a character, hard mask edges often create visible seams, so mask refinement is one of the most important parts of the workflow.

This workflow is ideal for character replacement, outfit transfer, reference-guided local editing, partial redraw, product / model replacement, AI portrait modification, Civitai preview creation, RunningHub demos, and fast Qwen 2511 editing tests. If you want to see how Qwen 2511, mask control, reference guidance, crop restoration, and final local replacement work together, watch the full tutorial from the YouTube link above.

⚙️ Try the Workflow Online

👉 Workflow: https://www.runninghub.ai/post/2041678315030843394?inviteCode=rh-v1111

Open the link above to run the workflow directly online and view the generation results in real time.

If the results meet your expectations, you can also deploy it locally for further customization.

🎁 Fan Benefits: Register now to get 1000 points, plus 100 daily login points — enjoy 4090-level performance and 48 GB of powerful compute!

📺 Bilibili Updates (Mainland China & Asia-Pacific)

If you are in Mainland China or the Asia-Pacific region, you can watch the video below for workflow demos and a detailed creative breakdown.

📺 Bilibili Video: https://www.bilibili.com/video/BV1QrDzBsELY/

I will continue updating model resources on Quark Drive:

👉 https://pan.quark.cn/s/20c6f6f8d87b

These resources are mainly prepared for local users, making creation and learning more convenient.

⚙️ 在线体验工作流

👉 工作流： https://www.runninghub.ai/post/2041678315030843394?inviteCode=rh-v1111

打开上方链接即可直接运行该工作流，实时查看生成效果。

如果觉得效果理想，你也可以在本地进行自定义部署。

🎁 粉丝福利： 注册即送 1000 积分，每日登录 100 积分，畅玩 4090 体验 48 G 超级性能！

📺 Bilibili 更新（中国大陆及南亚太地区）

如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。

📺 B站视频： https://www.bilibili.com/video/BV1QrDzBsELY/

我会在 夸克网盘 持续更新模型资源：

👉 https://pan.quark.cn/s/20c6f6f8d87b

这些资源主要面向本地用户，方便进行创作与学习。