Using Inpainting Models isn't effictive (comfyui backend)

when using an inpainting model while using comfyui as a back end it doesnt behave the same way as it is supposed to.

what i mean is usually when using an inpainting model if you have the denoise at 1 then a whole new image is generated and 0 doesnt change anything.

here whenever using an inpaint model, doesnt matter what the denoise is or what the inpainting conditioning mask strength is, the only thing it does is keep the same image with very minimal changes and over saturate the image.

even if i choose latent nothing as a mode and set the denoise to 1 and set the inpainting conditioning mask strength to 0 or 1, all i get is yellowish noise.