Available now

ERNIE Image: text to image

Baidu's ERNIE Image turns a prompt in English, Chinese or Japanese into one to four images, with prompt expansion on by default.

Model ID

Input

Fields marked * are required. Review the price before generating.

Text prompt for image generation. Supports English, Chinese, and Japanese.

Example prompt. Edit it or write your own.

This model reads prompts in English, Chinese or Japanese.

Additional settings

Negative prompt to guide what should not be in the image.

Number of denoising steps.

Classifier-free guidance scale.

Random seed for reproducibility.

If True, the prompt will be enhanced using an LLM for more detailed results.

The acceleration level to use for the image generation.

Price$0.05 / megapixel

You sign up first. Your settings are kept.

About this model

Overview

ERNIE Image makes one to four images per run in six preset shapes, as JPEG or PNG. It reads prompts in English, Chinese and Japanese.

Prompt expansion is on by default and adds detail to a short prompt; turn it off when you want your exact wording. A negative prompt, steps, guidance scale and acceleration are all adjustable.

What you can build

  • Images from prompts written in Chinese or Japanese
  • Detailed scenes from a few words with expansion on
  • Square, portrait and landscape images for posts
  • Repeatable results with a fixed seed

Prompting tips

  • Turn prompt expansion off when your prompt is already long and specific.
  • Use the negative prompt for things to leave out, such as text, watermarks or extra hands.
  • Keep the default steps for final images; fewer steps or high acceleration are fine for drafts.

What to check

Check whether expansion changed the intended subject, and review any generated lettering.

Inputs

  • Prompt (required): Text prompt for image generation. Supports English, Chinese, and Japanese.
  • Negative Prompt (optional): Negative prompt to guide what should not be in the image.
  • Image Size (optional): The size of the generated image. Options: Square HD, Square, Portrait 4:3, Portrait 16:9, Landscape 4:3, Landscape 16:9.
  • Number of Inference Steps (optional): Number of denoising steps. From 1 to 100.
  • Guidance Scale (optional): Classifier-free guidance scale. From 1 to 20.
  • Seed (optional): Random seed for reproducibility.
  • Number of Images (optional): Number of images to generate. From 1 to 4.
  • Enable Prompt Expansion (optional): If True, the prompt will be enhanced using an LLM for more detailed results.
  • Output Format (optional): Output image format. Options: jpeg, png.
  • Acceleration (optional): The acceleration level to use for the image generation. Options: none, regular, high.

Output and limits

  • Image sizes: Square HD (1024×1024), Square (512×512), Portrait 4:3 (768×1024), Portrait 16:9 (576×1024), Landscape 4:3 (1024×768), Landscape 16:9 (1024×576).
  • File formats: JPEG, PNG.
  • Up to 4 images per run.
  • Small or dense text in the image can come out misspelled.
  • Hands, crowds and complex poses can show artifacts; generate a few variations and pick the best.
  • Some prompts or outputs can be refused by safety checks.

Price guide

  • 1 megapixel: $0.05 / megapixel