This Person Does Not Exist

This Person Does Not Exist is a tool that does exactly one thing, and does it with unnerving competence: it generates a photorealistic human face, from scratch, every time you click refresh. The face belongs to no one. It never existed. And yet, it looks like it could be your neighbor, a stock photo model, or someone you passed on the street yesterday. For designers, developers, writers, and researchers who need a human face but don't want to use a real person's likeness, this tool offers a frictionless solution: no sign-up, no configuration, no waiting. You land on the page, you see a face, you click to get another. The engine beneath this simplicity is Nvidia's StyleGAN, a generative adversarial network that has become a benchmark for synthetic image quality. The result is a portrait that, at first glance, is indistinguishable from a photograph. But the tool's minimalism is both its strength and its limitation. There is no way to specify age, gender, ethnicity, expression, or pose. You get what the algorithm decides to produce. This makes it ideal for scenarios where variety is more important than control, such as generating placeholder avatars for UI mockups or populating a fictional world with random characters. For a designer working on a wireframe, the ability to drop in a realistic face without worrying about licensing or model releases is a genuine time-saver. The free tier gives you a 512x512 pixel JPEG with a watermark, which is sufficient for screen mockups but not for print or high-resolution displays. The paid tier, at $14.95 per image, removes the watermark and delivers a 1024x1024 pixel PNG and JPEG, with exclusive use rights. That pricing is steep per image, but for a single hero image in a presentation or a key visual in a product, it might be justifiable. The tool's workflow is entirely manual and click-driven. There is no API, no batch generation, no grid of options. You generate one face at a time, and if you like it, you download it. If not, you click again. This is fine for occasional use, but for anyone needing hundreds of unique faces for a game or a large-scale simulation, the process becomes tedious. The quality of the generated faces is generally high, but not flawless. Common artifacts include asymmetrical features, especially around the eyes and mouth, hair that blends unnaturally into the background, and background distortions that can look like smudged paint. These artifacts are more noticeable in the free 512px version because the lower resolution masks some detail but also amplifies compression artifacts. The paid version's higher resolution reveals more of the model's strengths and weaknesses. For researchers studying GAN outputs, the tool provides a convenient way to sample StyleGAN's latent space, though it offers no controls to manipulate that space. For writers illustrating articles about AI or synthetic media, the images serve as perfect examples of what GANs can achieve. The tool's FAQ acknowledges that it is 'almost impossible' to recognize a fake face, but points to symmetrical issues, uneven teeth, and background distortions as telltales. In practice, many generated faces pass the casual inspection test, but careful examination reveals the uncanny valley. The licensing for the free tier is restricted to personal use, and the site asks for a backlink if you use the images. That may be acceptable for a blog post or a school project, but it is not suitable for commercial work without upgrading. The paid tier grants exclusive use, meaning the image is removed from the public gallery and you have the rights to use it commercially. However, the terms are vague about what 'exclusive use' entails in practice, and users should review the fine print. Overall, This Person Does Not Exist is a niche tool that excels at its single function: generating a random, realistic face on demand. It is not a comprehensive avatar generator, nor is it a tool for creating specific characters. It is a quick, legal-safe source of synthetic faces for those who need them. The lack of controls and batch processing limits its utility for power users, but for the casual need, it is remarkably effective. The paid tier offers a clear upgrade in resolution and rights, but the price per image is high enough that it should be reserved for projects where a single face matters significantly. For most users, the free tier will suffice for exploration and low-stakes use, while professionals with specific needs may find the cost worth the convenience.