Rethinking Prompt Design for Inference-time Scaling in Text-to-Visual Generation

TL;DR

This paper introduces PRIS, a framework that adaptively revises prompts during inference for text-to-visual generation, significantly improving alignment and quality by addressing limitations of fixed prompts during scaling.

Contribution

PRIS is the first method to dynamically revise prompts during inference, using a new verifier for fine-grained alignment, leading to substantial quality improvements.

Findings

Achieved a 15% gain on VBench 2.0 benchmark.

Effectively identifies recurring failure patterns in generated visuals.

Enhances prompt-visual alignment through adaptive prompt redesign.

Abstract

Achieving precise alignment between user intent and generated visuals remains a central challenge in text-to-visual generation, as a single attempt often fails to produce the desired output. To handle this, prior approaches mainly scale the visual generation process (e.g., increasing sampling steps or seeds), but this quickly leads to a quality plateau. This limitation arises because the prompt, crucial for guiding generation, is kept fixed. To address this, we propose Prompt Redesign for Inference-time Scaling, coined PRIS, a framework that adaptively revises the prompt during inference in response to the scaled visual generations. The core idea of PRIS is to review the generated visuals, identify recurring failure patterns across visuals, and redesign the prompt accordingly before regenerating the visuals with the revised prompt. To provide precise alignment feedback for prompt revision, we introduce a new verifier, element-level factual correction, which evaluates the alignment between prompt attributes and generated visuals at a fine-grained level, achieving more accurate and interpretable assessments than holistic measures. Extensive experiments on both text-to-image and text-to-video benchmarks demonstrate the effectiveness of our approach, including a 15% gain on VBench 2.0. These results highlight that jointly scaling prompts and visuals is key to fully leveraging scaling laws at inference-time. Visualizations are available at the website: https://subin-kim-cv.github.io/PRIS.