---
title: "PACR-Video: Parameter-Efficient Multi-Shot Long Video Extrapolation"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.06481v1"
permalink: "https://feed.astralor.com/items/itm_8468b1d27b83080f4237823b"
published: "2026-07-07T16:41:25.000Z"
score: 78
channel: "tech"
tags: ["PACR-Video","Low-Rank Temporal Adapter","Parameter-Efficient Fine-Tuning","Diffusion Transformers","Prompt Routing","Long Video Generation"]
---

# PACR-Video: Parameter-Efficient Multi-Shot Long Video Extrapolation

## AI commentary

> This paper is worth attention as it addresses the persistent trade-off between fine-tuning cost and coherence in long video generation. For researchers working on video generation, diffusion models, or parameter-efficient fine-tuning, PACR-Video's idea of structuring narrative into routable prompt modules rather than blind memorization or full parameter tuning is highly inspiring. You may be able to try similar designs on your own model without huge computational budgets.

## At a glance

PACR-Video is a parameter-efficient framework for multi-shot long video extrapolation, keeping the text-to-video diffusion transformer frozen and augmenting it with low-rank temporal adapters and prompt routing. It builds a recursive prompt bank storing entity, location, action, and style prompts, then routes them via adapter gates according to narrative dependencies. Across six multi-shot and long-video benchmarks, PACR-Video outperforms baselines on distributional quality, semantic alignment, identity consistency, temporal smoothness, motion stability, transition coherence, and human preference, demonstrating that compact prompt routing and lightweight temporal adaptation provide sufficient controllable capacity for stable long video generation.

## Key points

- Frozen text-to-video diffusion Transformer with only low-rank temporal adapters and prompt routing trained, achieving parameter efficiency.
- Recursive prompt bank stores entity, location, action, and style prompts across shots, routed via gating for long-horizon coherence.
- Shot-Local/Story-Global tuning objective combines next-shot reconstruction, cross-shot identity contrast, and prompt sparsity regularization.
- Outperforms streaming, recursive, and memory-augmented baselines on six benchmarks across multiple quality metrics.

## Full text

We present PACR-Video, a parameter-efficient framework for multi-shot long video extrapolation that preserves recurring entities, scene structure, visual style, and causal progression without full generator fine-tuning. PACR-Video keeps a text-to-video diffusion transformer frozen and augments it with low-rank temporal adapters conditioned by learned shot-role prompt tokens. To maintain long-horizon coherence, it builds a recursive prompt bank that stores compact entity, location, action, and style prompts from previous shots, then routes them through adapter gates according to predicted narrative dependencies. A Shot-Local/Story-Global tuning objective combines next-shot reconstruction, cross-shot identity contrast, and prompt sparsity regularization, while an adapter composition schedule balances early-shot visual consistency with later-shot event progression and viewpoint change. Across six multi-shot and long-video benchmarks, PACR-Video outperforms text-to-video, tuning-based, memory-augmented, streaming, and recursive-context baselines on distributional quality, semantic alignment, identity consistency, temporal smoothness, motion stability, transition coherence, and human preference. These results show that compact prompt routing and lightweight temporal adaptation provide sufficient controllable capacity for stable long video extrapolation.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_8468b1d27b83080f4237823b