autoregressive image generation

Image synthesis that factors the joint distribution of an image into a sequence of conditional predictions, generating one element at a time given those already produced, as an autoregressive modelautoregressive model, n.: A model that produces a sequence one piece at a time, each new piece depending on the pieces already produced — like writing a sentence word by word without going back. does for text. Early models predicted raw pixel values in raster order with recurrent or masked convolutional networks (PixelRNN and PixelCNN, 2016), which gave exact likelihoods but were slow. Later systems first convert images into sequences of discrete codes with an image tokenizerimage 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. such as 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. or VQGANVQGAN, n.: An image compressor that turns a picture into a small grid of codebook 'tiles' while keeping it sharp, used as the vocabulary of many image generators and of an early wave of AI art., then train 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. by next-token predictionnext-token prediction, n.: The simple task most AI chatbots were trained on: given some text, guess the next small piece of it. Repeating that guess, piece by piece, produces whole answers. — the approach of the first DALL·EDALL·E, prop. n.: OpenAI's pioneering series of image generators, named after the painter Salvador Dalí and the robot WALL·E, which showed in 2021–2023 that AI could draw almost anything described in words., Google's Parti (2022) and Meta's Chameleon (2024). Variants predict many tokens in parallel by masking (MaskGIT, Muse), go coarse-to-fine across scales (visual autoregressive modeling, 2024) or use continuous tokens. Because images become ordinary tokens, one multimodal large language modellarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code. can read and write both text and images; such 'native' image generation, popularized by GPT-4o in March 2025, may pair an autoregressive backbone with a diffusion decoder.