畫筆化為文字：文字生成圖像 AI 的演進簡史
Original: Painting with words: a history of text-to-image AI
On the occasion of the first anniversary of Stable Diffusion and Replicate's launch of Stable Diffusion XL (SDXL) fine-tuning services…
本文回顧了文字生成圖像（Text-to-Image）技術的演進歷程。從早期的 GAN 限制，到 2021 年 VQGAN+CLIP 掀起的社群藝術熱潮，再到 2022 年 DALL-E 2 與 Stable Diffusion 的爆發。隨著 Stable Diffusion 迎來一週年及 SDXL 的推出，開源社群與微調技術正以前所未有的速度重塑視覺創作。
On the occasion of the first anniversary of Stable Diffusion and Replicate's launch of Stable Diffusion XL (SDXL) fine-tuning services, this article provides an in-depth retrospective of the key milestones in the evolution of text-to-image AI technology over the past several years.
Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.
See Pro plans →Want the original English / full article?
Read on Replicate Blog →Summaries are AI-generated; the original article is authoritative.