For best outputs: Sampler: DPM++ 2M Karras. Steps: 40 ~ 55, cfg 7 ~ 8.5, CLIP skip 2. VAE: clearvae_v2.3_fp16.safetensors (optional). Pos prompt: use the regular quality / realism keywords. Optional: I have included some detail LoRa's in some of my example runs, "add_detail" and "more_details". Neg prompt: include the embeddings "easynegative", "ng_deepnegative_v1_75t" and "EasyNegativeV2" in your neg (found in my example runs). What changes from v2.0 to v2.5: input_image feature Checkpoint Merge.