looking-glass-cli 1.7.0

Use your photos to design and try on outfits

Looking Glass

Use your photos to design and try on outfits — powered by AI image generation.

Supports OpenAI and Google Gemini as image generation providers.

Installation

Requires uv.

uv tool install looking-glass-cli

Quick Start

# 1. Set at least one API key (OpenAI, Gemini, or both)
glass config set openai_api_key sk-your-key-here
glass config set gemini_api_key your-gemini-key-here

# 2. Add a photo of yourself
glass photo add ~/selfie.jpg --name me

# 3. Add outfit descriptions (plain text files)
echo "Elegant navy suit, white dress shirt, silk tie" > formal.txt
glass outfit add formal.txt

echo "Hawaiian shirt, cargo shorts, flip-flops" > beach.txt
glass outfit add beach.txt --name beach-vibes

# 4a. Run interactive mode — pick a photo, pick outfits, pick a provider, generate!
glass

# 4b. OR: Batch capture outfit descriptions from a folder of photos
glass outfit capturefolder ~/photos --overwrite

Commands

Command	Description
glass	Interactive mode — select a photo, outfits, and provider, then generate images
glass photo add <file> [--name NAME]	Add a photo
glass photo list	List stored photos
glass photo remove <name>	Remove a photo
glass outfit add <file> [--name NAME]	Add an outfit description
glass outfit capture <file> [--name NAME] [--no-instruction-file]	Capture an outfit description from a photo using AI vision
glass outfit capturefolder [FOLDER] [--overwrite]	Batch capture outfit descriptions from all images in a folder (uses filenames as outfit names)
glass outfit create --name NAME	Create an outfit description in your default editor
glass outfit list	List stored outfits
glass outfit remove <name>	Remove an outfit
glass config set <key> <value>	Set a config value
glass config show	Show all available config options and their current values

Options

• --output, -o PATH — Directory for generated images (default: current directory)
• --version — Show version

For outfit capture, place a matching .txt file next to the photo to provide extra AI instructions automatically. For example, linen-look.jpg uses linen-look.txt. Use glass outfit capture --no-instruction-file to opt out for an individual capture, or pass --instructions/-i to override the file.

Configuration

Run glass config show to see all available options with descriptions and defaults.

Key	Description	Default
openai_api_key	OpenAI API key for image generation	—
gemini_api_key	Google Gemini API key for image generation	—
max_workers	Max parallel image generations	4

At least one API key is required. If both are configured, you'll be prompted to choose a provider during interactive mode.

Note: The glass outfit capture and glass outfit capturefolder commands require a vision-capable provider (OpenAI supports this via API key).

Migrating from an older version? The api_key config key has been replaced by openai_api_key. Your existing key will be migrated automatically on first run.

How It Works

1. Photos and outfit files are stored in ~/.looking-glass/
2. Outfits are plain .txt files — the filename is the name, the content is the description
3. In interactive mode, select a photo, one or more outfits, and a provider
4. For each outfit, the app sends your photo + a prompt to the selected provider's API
5. Generated images are saved as {photo}_{outfit}.png