Ernie Image

ERNIE Image is an open-source text-to-image model developed by Baidu, built on an 8B-parameter Diffusion Transformer (DiT). It is carefully engineered for: • Generating precise text within images • Creating structured and consistent layouts • Handling complex, multi-object prompts • Supporting both realistic and stylized outputs • Running locally on consumer-grade GPUs This model produces sharp, readable text for posters, infographics, and UI-style images, a common challenge for many image generation tools. It also excels at building consistent layouts for multi-panel designs or storyboards, ensuring the visual logic remains intact across the entire composition. Its Prompt Enhancer expands short inputs into richer, structured prompts, improving output quality without extensive manual engineering. ERNIE Image accurately follows detailed instructions, including multiple objects and their spatial relationships, preserving scene structure effectively. It can generate both photorealistic and creative artwork within the same workflow, offering versatility. Operating on a single 24GB consumer GPU (like an RTX 3090 or 4090), it is ideal for local deployment without API limits. Released under the Apache 2.0 License, ERNIE Image offers flexibility for modification and commercial use.