OpenAI has developed a method for generating images based on text prompts, using the CLIP model's latent space. This approach allows for more nuanced and detailed image creation than previous methods. The technique involves encoding text into a hierarchical representation, which is then used to condition an image generator. This results in images that are more closely tied to the original text prompt. --- Why it matters: This matters because it could lead to significant improvements in AI-generated imagery, with potential applications in fields like art, design, and advertising. Source: https://openai.com/index/hierarchical-text-conditional-image-generation-with-clip-latents