Skip to content

InstanceAssemble

Layout-to-image generation with spatial control

Visit official site

What InstanceAssemble is

The framework is a lightweight layout-to-image generation system with precise spatial control, presented at NeurIPS 2025. It introduces DenseLayout and the Layout Grounding Score.

What you can do with it

  • Generate images from sparse or dense layouts
  • Evaluate results with the Layout Grounding Score
  • Pick text-based or visual control variants

How to start

  • Clone the project repository locally
  • Create and activate a Python 3.10 environment
  • Install dependencies and fetch the model weights

Pros & cons

✓ What we like

  • Fast inference is listed
  • HuggingFace compatibility eases model downloads

! What to watch out for

  • Local runs need a Python 3.10 environment and dependencies
  • Model weights must be downloaded to the preset folder

FAQ

Where was it presented?

The material states it was introduced at NeurIPS 2025.

What evaluation measure is named?

The material names the Layout Grounding Score.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review