Back to Works

AI Portraits

Turn your selfie into a Renaissance masterpiece

Year

2019

Category

AI Art

Lab

MIT-IBM Watson AI Lab

Website

aiportraits.com

About the Project

AI Portraits was an online, GAN-based generative art application launched in July 2019. The tool allowed users to upload a personal photograph and receive a newly generated portrait in the style of classical Western art. Unlike traditional style transfer filters, AI Portraits employed Generative Adversarial Networks (GANs) to synthesize original images inspired by historical paintings rather than merely overlaying painterly textures.

The application used a dataset of over 45,000 classical artworks to train its generative models, focusing on portraiture from various art historical periods. When a user uploaded an image, the system generated a wholly new portrait that captured stylistic elements such as brush strokes, lighting, and composition from canonical Western artworks. The model was designed so that the generated portrait was not a direct manipulation of the user's original photo but a re-imagining of the subject's facial features and contours in a classical artistic style.

Cultural Impact

Upon its debut in mid-2019, AI Portraits gained rapid popularity online and quickly went viral, reaching 8 million portraits generated per day worldwide. Users, celebrities, and media outlets shared their generated portraits, often comparing the results to famous museum paintings. The platform sparked conversation about the role of artificial intelligence in creative endeavors, artistic authenticity, and privacy concerns related to uploading personal images to third-party AI tools.

AI Portraits is frequently cited as one of the earliest viral GAN-based art tools, predating later text-to-image generators and multimodal creative platforms. Its popularity paved the way for subsequent innovations such as DALL-E, Midjourney, and Stable Diffusion, all of which would further expand public awareness and engagement with generative AI art.