AI

From Prompting to Composing: A Spatial Canvas Interface for Poster Generation

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 09/10/2026, 03:18:25 UTC.

Briefings

The authors present Compo, a poster generation model that replaces text prompts with a spatial canvas interface. Compo supports four binding types: semantic, identity, text, and pixel, allowing users to compose generation intent directly in space. The model was trained using a scalable pipeline that automatically constructs supervision data for different binding types and their combinations, enabling efficient adaptation without training from scratch. Experiments show Compo achieves stronger compositional controllability than general-purpose image generation models and dedicated poster generation systems while maintaining high visual quality.

Key points

• Compo is a poster generation model that replaces text prompts with a spatial canvas interface.
• Compo supports four binding types: semantic, identity, text, and pixel, allowing direct composition of generation intent in space.

Context

Poster generation traditionally relies on text prompts, which require users to encode two-dimensional composition intent into a one-dimensional sequence of words. This work introduces Compo, a model adapted from a pretrained image editing model to understand Spatial Canvas inputs and Text Specifications, shifting the paradigm from prompting to composing.

Limitations

The evidence is limited to the authors' experimental results on Compo and its training pipeline. Specific performance metrics for other models or the benchmark's detailed results are not provided. The study does not report long-term usability or real-world adoption data.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 14/11/2026, 08:58:26 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeie6ywth4ibgpw6mzvw4awpzaccomi6a4ora37rwqqm2zblkasvs7m