PreciseCache: Precise Feature Caching for Efficient and High-fidelity Video Generation

TL;DR

This work proposes PreciseCache, a plug-and-play framework that precisely detects and skips truly redundant computations, thereby accelerating inference without sacrificing quality, and proposes BlockCache, which precisely detects and skips redundant computations at the block level within the network.

Abstract

High computational costs and slow inference hinder the practical application of video generation models. While prior works accelerate the generation process through feature caching, they often suffer from notable quality degradation. In this work, we reveal that this issue arises from their inability to distinguish truly redundant features, which leads to the unintended skipping of computations on important features. To address this, we propose \textbf{PreciseCache}, a plug-and-play framework that precisely detects and skips truly redundant computations, thereby accelerating inference without sacrificing quality. Specifically, PreciseCache contains two components: LFCache for step-wise caching and BlockCache for block-wise caching. For LFCache, we compute the Low-Frequency Difference (LFD) between the prediction features of the current step and those from the previous cached step. Empirically, we observe that LFD serves as an effective measure of step-wise redundancy, accurately detecting highly redundant steps whose computation can be skipped through reusing cached features. To further accelerate generation within each non-skipped step, we propose BlockCache, which precisely detects and skips redundant computations at the block level within the network. Extensive experiments on various backbones demonstrate the effectiveness of our PreciseCache, such as achieving an average of $2.6\times$ speedup on Wan2.1-14B without noticeable quality loss.

Figures (9)

• Figure 1: Qualitative Results of PreciseCache on Wan2.1-14Bwan2025. Compared with previous methods, our PreciseCache achieves higher acceleration (about 2.6$\times$ speedup) of the base model without sacrificing the quality of generated videos.
• Figure 2: The Illustration of the Denoising Process for Video Generation. At high-noise timesteps, the prediction of the model varies significantly. Reusing the cached features in this stage (the red line) can significantly affect both the content and the quality of generated videos compared to the videos generated without caching (the orange line). In contrast, the feature caching during low-noise timesteps only introduces negligible impacts (the green line).
• Figure 3: The Impact of Reusing the Cached Model Prediction at Each Timestep. Considering a 50-step denoising process from the Gaussian Noise $\boldsymbol{Z}_{50}$ to a clean latent $\boldsymbol{Z}_{0}$, we respectively reuse the model output at timestep $t_{i+1}$ for each $i \in \{49, 48, \cdots, 0\}$, and perform the subsequent denoising steps to generate the final video. We then compare each resulting video with the baseline (i.e., generated without caching and reusing) to evaluate the impact of reusing cached predictions. Different colors indicate different prompts.
• Figure 4: The Difference between Model Predictions at Adjacent Timesteps. Although the difference is relatively large in the low-noise stage, it primarily arises from high-frequency components, which have limited influence on the perceptual quality of the generated results. Different colors indicate different prompts.
• Figure 5: Pipeline of PreciseCache.