The authors frame positive-negative space generation as a staged design workflow rather than a single-pass text-to-image problem.
Staged AI agent improves generation of positive and negative space compositions
FaV-A decomposes the task into base object creation, negative-space analysis, and final compositional synthesis, outperforming direct single-pass MLLM prompting.
Research Lab
Shiwen Wang · Jian Yang · Xu Wang · Xincan Wang · Weiming Dong
University of Chinese Academy of Sciences · Renmin University of China · Shanghai Theatre Academy · Institute of Automation, Chinese Academy of Sciences
Research Digest··2 min read
The authors present FaV-A, a multimodal agent that generates positive-negative space images through three connected stages.
Why this paper
From University of Chinese Academy of Sciences and 3 others
In one line
A three-stage multimodal agent produces more coherent, semantically aligned positive-negative-space images than direct zero-shot prompting.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§