Skip to content

ACE: All-round Creator and Editor Following Instructions via Diffusion Transformer

One diffusion transformer for creation and editing

Visit official site

What ACE is

The model is described as an all-round creator and editor based on a diffusion transformer, trained jointly on visual generation tasks through a unified condition format.

What you can do with it

  • Create and edit visual content
  • Join several generation tasks
  • Feed inputs as condition units

Getting started

  • Visit the official site or get the app
  • Register and log in
  • Enter or upload the condition format

Pros & cons

✓ What we like

  • Data collection is stated as efficient
  • A chat system can respond to image creation requests

! What to watch out for

  • Tasks are conditioned through the LCU format
  • Text instructions come from a multimodal language model

FAQ

What is LCU?

The material names Long-context Condition Unit as the unified condition format.

What benchmark is released?

The material states a manually annotated image-pair benchmark for evaluation.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review