I need a small-scale AI model or script that reliably turns an input photo into a cartoon-style portrait with moderate detail. The focus is squarely on portraits, not landscapes or abstract work, and I’m after a playful yet recognisable look—clean lines, balanced shading and colour, no hyper-realistic textures. You’re welcome to fine-tune Stable Diffusion, train a lightweight GAN, or build a custom pipeline in Python using PyTorch or TensorFlow, as long as the result runs locally (GPU) and can be triggered by a simple command-line or minimal web interface. I’ll supply a handful of sample headshots for calibration; please keep facial proportions faithful while still achieving that classic cartoon charm. Deliverables: • Trained model files and any checkpoints • Inference script or micro-web app with clear instructions • A brief README covering environment setup, dependencies and usage • Five sample outputs that demonstrate the expected moderate level of detail Acceptance criteria: given a 1024×1024 photo, the system produces a 1024×1024 cartoon portrait within 60 s on an RTX-class GPU, preserving likeness and meeting the stylistic brief above. Feel free to suggest data augmentation tricks or post-processing tweaks you find useful—I’m open to ideas that improve consistency while keeping inference lightweight.