How to best combine pose, identity, clothing, and background with Krea 2 in ComfyUI?
A ComfyUI user built a workflow using Krea 2 Turbo, identity LoRA, clothing reference, background reference, and DWPose/OpenPose LoRA, hoping to preserve exact face, clothing, and room while following a completely different pose reference, but current pose consistency is unstable, and clothing and background details also vary. The user asks whether to use a two-pass workflow (first establish pose and composition, then refine identity, clothing, and background) or whether there is a more reliable single-pass conditioning method for Krea 2.