Art Critique GPT

A professional art analysis tool powered by GPT-4 Vision that provides comprehensive critiques of artworks across all media types.

Features

• Structured Analysis: Get detailed critiques with formal analysis, interpretation, context, and scoring
• Professional UI: Clean Streamlit interface with well-organized sections
• Type-Safe: Uses Pydantic models with instructor for structured output
• Flexible Prompting: Jinja2 templates for customizable prompts
• Comprehensive Scoring: Detailed scores across multiple criteria with explanations

Installation

Make sure you have uv installed, then install the dependencies:

uv pip install -r requirements.txt

Usage

Running the Streamlit App

streamlit run app.py

Then:

1. Upload an artwork image in the sidebar
2. Enter your OpenAI API key
3. Click "Analyze Artwork" to get a comprehensive critique

Features of the Analysis

The app provides:

• Overview: Initial impressions and summary
• Formal Analysis: Composition, color, technique, and texture analysis
• Interpretation: Emotional impact, conceptual depth, and artistic intention
• Context: Historical and cultural context with comparisons
• Feedback: Strengths, areas for improvement, and practical suggestions
• Scoring: Detailed scores (1-10) across multiple criteria

API Usage

You can also use the analysis function directly:

from PIL import Image
from image_analyser_gpt.ask_gpt import analyze_artwork

# Load your image
image = Image.open("artwork.jpg")

# Analyze the artwork
critique = analyze_artwork(image, "your-openai-api-key")

# Access structured data
print(f"Overall Score: {critique.scores.overall_score.score}/10")
print(f"Overview: {critique.overview}")

Project Structure

GPT_art_critique/
├── app.py                          # Main Streamlit application
├── image_analyser_gpt/
│   ├── __init__.py
│   ├── ask_gpt.py                 # GPT interaction with instructor
│   ├── models.py                  # Pydantic models for structured output
│   └── utils.py                   # Utility functions
├── image_analyser_prompt.j2       # Jinja2 template for the prompt
├── requirements.txt
└── README.md

Supported Formats

• Images: PNG, JPG, JPEG, GIF, BMP
• Art Types: Paintings, sculptures, digital art, mixed media, installations, performance documentation

Requirements

• Python 3.8+
• OpenAI API key with GPT-4 Vision access
• See requirements.txt for all dependencies

Configuration

The app uses the image_analyser_prompt.j2 template for generating prompts. You can customize this template to adjust the analysis criteria or output format.

License

This project is open source and available under the MIT License.