ComfyUI PixelSmile

PixelSmile ComfyUI Tutorial — Fine-grained facial expression editing on RunPod In this tutorial I walk through how to set up and use PixelSmile inside ComfyUI on RunPod — a Qwen-Image-Edit LoRA that gives you named emotion + intensity control on still portrait photos. 12 expressions, identity preserved, preview build. WORKFLOW Workflow file (Google Drive): https://drive.google.com/file/d/1qkDUxETyJUn5GPBsCatOwvCfFjsmdhZk/view?usp=sharing MODEL LINKS PixelSmile LoRA: https://huggingface.co/PixelSmile/PixelSmile Lightning LoRA: https://huggingface.co/lightx2v/Qwen-Image-Edit-2511-Lightning Diffusion model: https://huggingface.co/Comfy-Org/Qwen-Image-Edit_ComfyUI Text encoder: https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI/blob/main/split_files/text_encoders/qwen_2.5_vl_7b_fp8_scaled.safetensors VAE: https://huggingface.co/Comfy-Org/Qwen-Image-Edit_ComfyUI NODE CREDIT ComfyUI integration by judian17 https://github.com/judian17/ComfyUI-PixelSmile-Conditioning-Interpolation TIMESTAMPS 0:00 Results 0:30 What is PixelSmile 1:15 Setup on RunPod 2:00 Node walkthrough 3:00 Gotchas 3:45 What works well 4:15 Final results + wrap GOTCHAS Lightning LoRA strength: use 0.1 not 1.0 (grid artifacts at full strength) Score 0 ≠ original image (diffusion still runs, image gets sharpened) Subfolder paths matter: qwen/ and qwen-edit/ required RunPod: download outputs before stopping your pod WHAT WORKS BEST Clean color portraits, even lighting, front-facing Grainy B&W surprisingly good Avoid: animals, very soft photos, heavy skin texture at Lightning 1.0 Best expressions for visible results: happy, surprised, angry at score 1.5