Anyone knows how to make it work with Image 2 image?

[ClothingAI]

Hi,
The problem with image 2 image is that in order for it to work, it needs to have a KSampler that receives the same width and height for both:

The latent image from the (input image)
"feature image" entering the node "Add Magic Clothing Attention"
The problem when having images of the same dimensions is that the image for the clothing will take the full image, and the input image (for img2img) has lot going on: "a person, legs, hands etc)

So when you try to pass all that through the sampler, it will "merge" the cloth (t shirt for instance) into the full body of the input image, and it does not work well.
Also, another problem with img2img:

if you do a high denoise, you can get better clothing reproduction, but you lose the original subject
if you have low denoise, you keep the subject, but the clothing is not copied well.
Any solutions please