Face Restoration and Enhancement Pipeline

Original prompt: {{ORIGINAL_PROMPT}}

Face Enhancement Pipeline:

Step 1 - Initial Generation:
- Generate base image with your prompt
- Note any face issues (blur, distortion, asymmetry)

Step 2 - CodeFormer/GFPGAN:
- CodeFormer Weight: {{CODEFORMER_WEIGHT}} (0 = max restoration, 1 = max fidelity)
- GFPGAN visibility: {{GFPGAN_WEIGHT}}
- Recommendation: Start CodeFormer at 0.5, adjust based on results

Step 3 - ADetailer Refinement:
- Model: face_yolov8n / mediapipe_face_full
- Detection confidence: 0.3-0.5
- Denoising strength: 0.3-0.4
- Inpaint prompt: detailed face, perfect features, symmetrical, sharp focus

Step 4 - Targeted Inpainting (if needed):
- Mask specific problem areas
- Low denoise (0.3-0.4) for subtle fixes
- Match lighting in prompt

Negative prompt: blurry face, asymmetrical, distorted features, bad eyes, bad teeth