vllm-omni/docs/user_guide/examples/offline_inference/image_to_image.md

Image-To-Image

Source https://github.com/vllm-project/vllm-omni/tree/main/examples/offline_inference/image_to_image.

This example edits an input image with Qwen/Qwen-Image-Edit and JoyAI-Image-Edit using the image_edit.py CLI.

Local CLI Usage

Single Image Editing

Download the example image:

wget https://vllm-public-assets.s3.us-west-2.amazonaws.com/omni-assets/qwen-bear.png

Then run:

python image_edit.py \
  --image qwen-bear.png \
  --prompt "Let this mascot dance under the moon, surrounded by floating stars and poetic bubbles such as 'Be Kind'" \
  --output output_image_edit.png \
  --num-inference-steps 50 \
  --cfg-scale 4.0

JoyAI-Image-Edit

JoyAI-Image-Edit uses the same image_edit.py CLI and supports one input image per request. Requested dimensions are snapped to the nearest supported Joy bucket; for square outputs, use 1024x1024.

python image_edit.py \
  --model jdopensource/JoyAI-Image-Edit-Diffusers \
  --image input.png \
  --prompt "Change the background to a clean studio while preserving the subject." \
  --height 1024 \
  --width 1024 \
  --num-inference-steps 50 \
  --guidance-scale 4.0 \
  --output output_joyai_edit.png

JoyAI defaults to guidance scale 4.0; pass --guidance-scale explicitly when you need a different value.

Multiple Image Editing (Qwen-Image-Edit-2509)

For multiple image inputs, use Qwen/Qwen-Image-Edit-2509 or Qwen/Qwen-Image-Edit-2511:

python image_edit.py \
  --model Qwen/Qwen-Image-Edit-2509 \
  --image img1.png img2.png \
  --prompt "Combine these images into a single scene" \
  --output output_image_edit.png \
  --num-inference-steps 50 \
  --cfg-scale 4.0 \
  --guidance-scale 1.0

Key arguments:

• --model: model name or path. Use Qwen/Qwen-Image-Edit-2509 or later for multiple image support. jdopensource/JoyAI-Image-Edit-Diffusers supports one input image.
• --image: path(s) to the source image(s) (PNG/JPG, converted to RGB). Can specify multiple images.
• --prompt / --negative-prompt: text description (string).
• --cfg-scale: true classifier-free guidance scale (default: 4.0). Classifier-free guidance is enabled by setting cfg_scale > 1 and providing a negative_prompt. Higher guidance scale encourages images closely linked to the text prompt, usually at the expense of lower image quality.
• --guidance-scale: guidance scale for guidance-distilled models (default: 1.0, disabled). Unlike classifier-free guidance (--cfg-scale), guidance-distilled models take the guidance scale directly as an input parameter. Enabled when guidance_scale > 1. Ignored when not using guidance-distilled models.
• --num-inference-steps: diffusion sampling steps (more steps = higher quality, slower).
• --output: path to save the generated PNG.
• --vae-use-slicing: enable VAE slicing for memory optimization.
• --vae-use-tiling: enable VAE tiling for memory optimization.
• --cfg-parallel-size: set it to 2 to enable CFG Parallel. See more examples in user_guide.
• --enable-cpu-offload: enable CPU offloading for diffusion models.

ℹ️ If you encounter OOM errors, try using --vae-use-slicing and --vae-use-tiling to reduce memory usage.

Example materials

??? abstract "image_edit.py" py --8<-- "examples/offline_inference/image_to_image/image_edit.py" ??? abstract "run_qwen_image_edit_2511.sh" sh --8<-- "examples/offline_inference/image_to_image/run_qwen_image_edit_2511.sh"