VQGAN

A VQ-VAEVQ-VAE, n.: A network that compresses an image or sound into a short list of entries from a learned 'codebook' — like describing a picture with a fixed set of tiles — which other models can then generate. trained with additional adversarial and perceptual lossesperceptual loss, n.: A way of judging how different two images look to a person by comparing what a trained image-recognition network 'sees' in each, instead of comparing pixels one by one. so that its discrete codes decode into sharp, realistic images even at high compression. A convolutional encoder maps an image to a grid of codebook indices (for example 16 × 16 for a 256 × 256 image); a patch-based discriminator and a feature-space reconstruction loss push the decoder to restore realistic local texture instead of blur. In the original work a transformertransformer, n.: The design behind today's chatbots and much of modern AI. It handles text by letting every word look at every other word and decide which of them matter for understanding it. was then trained autoregressively on the code grids, pairing the convolutional network's efficiency on local structure with the transformer's long-range modeling and making megapixel synthesis practical. The VQGAN recipe — an autoencoder trained with perceptual plus adversarial losses — became the standard way to build first-stage compressors: latent diffusionlatent diffusion model, n.: A diffusion image generator that works on a shrunken, compressed version of each picture instead of the full-size pixels, which makes it far cheaper to train and run. and Stable DiffusionStable Diffusion, prop. n.: A widely used image-generating AI whose model files were released for anyone to download, so people could run it on their own computers and adapt it. use a continuous variant, and many image tokenizersimage tokenizer, n.: A component that chops a picture into a short sequence of 'visual words' that a language-style AI model can read or write, and turns such sequences back into pictures. descend from it.