Photo Retouching Tool

Problem

Portrait photos often suffer from low exposure, gray white balance, uneven skin tone, and over-smoothed texture. Manual retouching depends on Photoshop layers and operator experience, which is hard to repeat for non-designers. This skill turns the retouching pipeline into a local Python script that runs with Pillow and NumPy.

How It Works

The workflow is preview-first. The script generates a 2x2 style grid: cool fresh cold tone, warm soft warm tone, vintage film look, and natural balanced tone. After the user selects a style by number or name, the full retouch pass runs. Core capabilities include:
- Exposure correction via histogram stretch and gamma adjustment
- White balance using gray world estimation
- Frequency separation smoothing that reduces uneven tone while preserving pores
- Skin tone unification to reduce redness and yellow cast
- Texture sharpening with high-pass detail and soft-light blending
- Optional upscaling with --upscale 2 or --upscale 4 before retouching

Boundaries

It fits wedding photos, portraits, family photos, couple photos, and children's photography. It is not suitable for landscapes, product images, text-only images, or non-portrait content. Do not use it for cropping, compression, or format conversion only. Strong backlighting or heavy color cast can affect skin detection; prefer natural or lower intensity.