Main features:
- Qwen-Image-2.1 product/subject compositing workflow for placing a foreground subject into a natural scene.
- Provides a standard 40-step quality route and a faster Viggle acceleration route.
- Uses Qwen Image 2.1 checkpoints, Qwen3-VL CLIP, the Qwen Image 2.1 VAE, and dedicated LoRAs for the standard and accelerated paths.
- Includes a 4x UltraSharp upscale branch and separate image save outputs for the standard and accelerated results.
Suggested workflow:
1. Load the subject image and the target scene/background image.
2. Use the standard 40-step route when maximum control and quality are preferred.
3. Use the Viggle route with its acceleration LoRA for faster output.
4. Decode, optionally upscale, compare the two routes, and save the selected composited image.
Dependencies and limitations:
- Requires the referenced Qwen Image 2.1 checkpoints, Qwen3-VL CLIP, VAE, UltraSharp upscaler, and the two named LoRA files.
- The graph exposes two routes with different step counts; the accelerated route is not equivalent to the standard 40-step route.
- Only nodes connected to the final save outputs are treated as enabled; disconnected previews or side branches are not guaranteed outputs.
RunningHub:
https://www.runninghub.ai/zh-cn/post/2107777471256297473?inviteCode=rh-v1111
Bilibili:
https://www.bilibili.com/video/BV1AsHy6nEPR/
RunningHub registration:
https://www.runninghub.ai/?inviteCode=rh-v1111
Description
Main features:
- Qwen-Image-2.1 product/subject compositing workflow for placing a foreground subject into a natural scene.
- Provides a standard 40-step quality route and a faster Viggle acceleration route.
- Uses Qwen Image 2.1 checkpoints, Qwen3-VL CLIP, the Qwen Image 2.1 VAE, and dedicated LoRAs for the standard and accelerated paths.
- Includes a 4x UltraSharp upscale branch and separate image save outputs for the standard and accelerated results.
Suggested workflow:
1. Load the subject image and the target scene/background image.
2. Use the standard 40-step route when maximum control and quality are preferred.
3. Use the Viggle route with its acceleration LoRA for faster output.
4. Decode, optionally upscale, compare the two routes, and save the selected composited image.
Dependencies and limitations:
- Requires the referenced Qwen Image 2.1 checkpoints, Qwen3-VL CLIP, VAE, UltraSharp upscaler, and the two named LoRA files.
- The graph exposes two routes with different step counts; the accelerated route is not equivalent to the standard 40-step route.
- Only nodes connected to the final save outputs are treated as enabled; disconnected previews or side branches are not guaranteed outputs.
RunningHub:
https://www.runninghub.ai/zh-cn/post/2107777471256297473?inviteCode=rh-v1111
Bilibili:
https://www.bilibili.com/video/BV1AsHy6nEPR/
RunningHub registration:
https://www.runninghub.ai/?inviteCode=rh-v1111
