The Plus version of the Qwen image Edit image editing model can load 3 reference images and has been tested to be very powerful with good consistency. Changing clothes, changing posture, taking items from the reference picture, taking off clothes, changing styles, taking photos, etc., are basically perfectly competent.
Attention: To have the character in Figure 1 hold the item in Figure 2, you need to describe it as follows: Have the character in Figure 1 change their posture and hug the character in Figure 2. To change the posture of the character in Figure 1, otherwise the generated image will be very strange.

Local deployment requires downloading the latest model:
https://hf-mirror.com/Comfy-Org/Qwen-Image-Edit_ComfyUI/tree/main/split_files/diffusion_models
Place folder: models/diffusionmodels
We also need to update ComfyUI to the latest version!

Reference instruction:
Replace the clothes in Figure 1 with the clothes in Figure 2. change clothes
Change the posture of the character in Figure 1 to face the camera and focus on it. Change positions
Change the posture of the character in Figure 1 and read the newspaper in Figure 2. Get the newspaper
Change Figure 1 to anime style
Make the character in Figure 1 change their posture and embrace the character in Figure 2