What Is This X Does Not Exist?

Updated

This X Does Not Exist is a directory page that collects websites generating realistic-looking fake content with generative adversarial networks (GANs). It is not a single tool: each entry links out to a separate site, built by a different creator, that fakes one specific category of thing — people, cats, rental listings, waifu, questions, resumes, emotions, or ceramic vessels. Use it when you want a quick tour of what generative models can imitate, or when you need a specific fake-content generator and want to find the right one without hunting across the web.

What the site actually is

The page describes itself as a collection of sites that "sprung up in the past month," all built on the same underlying idea: a GAN learns from real examples and then produces new samples that look like they belong to the same category. The directory format means the value is in the links and the short notes about each one, not in any generation happening on the page itself.

The entries it lists

Site What it fakes Model / note Creator
This Person Does Not Exist Human faces StyleGAN; described as the site that started the trend Phillip Wang
This Cat Does Not Exist Cat photos GAN; occasionally produces deformed cats Ryan Hoover
This Rental Does Not Exist Rental listings Presented as a way to "create" a home rather than search for one Christopher Schmidt
This Waifu Does Not Exist Anime-style characters Noted as stranger than the others, with a plot summary Gwern Branwen
This Question Does Not Exist Stack Overflow questions Includes an "Other Random Questions" sidebar of equally synthetic questions Brad Dwyer
This Resume Does Not Exist Resumes Uses TextgenRNN (an RNN, not a GAN); headshots come from thispersondoesnotexist.com Enhancv
This Emotion Does Not Exist Facial emotion interpretation Not necessarily deep learning; a simpler model reading expressions Jan Schneider
This Vessel Does Not Exist Ceramic vessels Draws a parallel between GAN generators/discriminators and apprentice/master potters —

What the examples have in common — and where they differ

Every entry follows the same pattern: a model is trained on a large set of real examples, then asked to produce a new one that passes as real. The differences are worth noticing if you are trying to understand how these systems work.

  • Model type varies. Most entries use GANs, but This Resume Does Not Exist uses TextgenRNN, a recurrent neural network specialized in text generation. This Emotion Does Not Exist is described as not necessarily using deep learning at all.
  • Output type varies. Images (faces, cats, waifu, vessels), text (questions, resumes), and interpretation (emotions) are all represented.
  • Quality varies. The page warns that the cats sometimes have visual deformities from model imperfections. That is a useful reminder that these outputs are samples from a model, not curated results.
  • Components get reused. The resumes pull their headshots from This Person Does Not Exist, so one generator's output becomes another project's input.

How to use it

  1. Open the directory and scan the list for the category you need.
  2. Click through to the individual site — generation happens there, not on the directory page.
  3. Refresh or reload the target site to get a new sample each time.
  4. Treat the output as a demonstration of model capability, not as a verified or licensed asset. The page gives no information about usage rights, pricing, or login requirements for any of the linked sites.

Common points of confusion

  • It is a showcase, not a product. There is nothing to sign up for on the directory itself.
  • "Does not exist" is literal. The people, cats, listings, and resumes are synthetic; they do not correspond to real individuals or properties.
  • The name is a template. "This X Does Not Exist" is the naming convention the linked sites share, which is why the directory reads as a family of projects rather than one brand.
  • Not all of it is a GAN. The page is explicit that at least one entry uses an RNN instead, so do not assume every generator here works the same way.