RESEARCH · RESEARCH · #2710
StruPO: structure-aware prompt editor that improves faithfulness in text-to-image generation
The authors introduce StruPO, a structure-aware prompt editor that conditions on a calibrated scene graph to guide token-level prompt edits (deletions, reordering, insertions) and uses image-grounded supervision and smooth reward shaping to train a reusable editing policy. Evaluated on three disjoint 2K-prompt test sets across four frozen generation backbones and validated by automated metrics plus a 15-annotator study, StruPO reportedly improves the trade-off between aesthetics and relational faithfulness without iterative critique–rewrite loops.
KEY POINTS
- The authors introduce StruPO, a structure-aware prompt editor that conditions on a calibrated scene graph to guide token-level prompt edits (deletions, reordering, insertions) and uses image-grounded supervision and smooth reward shaping to train a reusable editing policy.
- Evaluated on three disjoint 2K-prompt test sets across four frozen generation backbones and validated by automated metrics plus a 15-annotator study, StruPO reportedly improves the trade-off between aesthetics and relational faithfulness without iterative critique–rewrite loops.
- This matters because it offers a reusable, one-pass method to edit prompts that preserves scene structure and relational fidelity—addressing a key gap in making text-to-image outputs truer to user intent without model fine-tuning or iterative human-in-the-loop loops.
WHY IT MATTERS
This matters because it offers a reusable, one-pass method to edit prompts that preserves scene structure and relational fidelity—addressing a key gap in making text-to-image outputs truer to user intent without model fine-tuning or iterative human-in-the-loop loops.