for me? it's illusion diffusion

>>11153 I'm still new and use comfyui. I have like 3 workflows that I work with. the first one is the "normal" one where I use 1 checkpoint, 1 LoRa and positive and negative prompts to generate an image. Next workflow I have one with controlnet where I can input a certain pose and a workflow for inpainting. What I struggle with the most is that when I want to remove artifacts via inpainting that it just doesn't blend in with the background at all. I feel like the inpainting models that I have only really work well for facial expressions but if I want to remove a torso or an additional leg, then inpainting is kind of useless for me. How would one go about removing the torso in this image https://files.catbox.moe/8wfbmp.png.Just try to change prompts while keeping the seed fixed? Inpainting? Just using photoshop? or some other method entirely?