Tiled Prompts: Overcoming Prompt Underspecification in Image and Video Super-Resolution
This addresses a specific bottleneck in super-resolution pipelines for high-resolution image and video generation, offering an incremental improvement over existing methods.
The paper tackled the problem of prompt underspecification in image and video super-resolution by proposing Tiled Prompts, which generates tile-specific prompts to improve localized guidance, resulting in consistent gains in perceptual quality and text alignment while reducing artifacts.
Text-conditioned diffusion models have advanced image and video super-resolution by using prompts as semantic priors, but modern super-resolution pipelines typically rely on latent tiling to scale to high resolutions, where a single global caption causes prompt underspecification. A coarse global prompt often misses localized details (prompt sparsity) and provides locally irrelevant guidance (prompt misguidance) that can be amplified by classifier-free guidance. We propose Tiled Prompts, a unified framework for image and video super-resolution that generates a tile-specific prompt for each latent tile and performs super-resolution under locally text-conditioned posteriors, providing high-information guidance that resolves prompt underspecification with minimal overhead. Experiments on high resolution real-world images and videos show consistent gains in perceptual quality and text alignment, while reducing hallucinations and tile-level artifacts relative to global-prompt baselines.