REVIEW 16 cited by
Long Context Tuning for Video Generation
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Recent advances in video generation can produce realistic, minute-long single-shot videos with scalable diffusion transformers. However, real-world narrative videos require multi-shot scenes with visual and dynamic consistency across shots. In this work, we introduce Long Context Tuning (LCT), a training paradigm that expands the context window of pre-trained single-shot video diffusion models to learn scene-level consistency directly from data. Our method expands full attention mechanisms from individual shots to encompass all shots within a scene, incorporating interleaved 3D position embedding and an asynchronous noise strategy, enabling both joint and auto-regressive shot generation without additional parameters. Models with bidirectional attention after LCT can further be fine-tuned with context-causal attention, facilitating auto-regressive generation with efficient KV-cache. Experiments demonstrate single-shot models after LCT can produce coherent multi-shot scenes and exhibit emerging capabilities, including compositional generation and interactive shot extension, paving the way for more practical visual content creation. See https://guoyww.github.io/projects/long-context-video/ for more details.
Forward citations
Cited by 16 Pith papers
-
SlotMem: Character-Addressable Internal Memory for Narrative Long Video Generation
SlotMem keeps a compact, updateable memory slot for each recurring character and injects it only into that character's tokens, reporting improved long-range identity consistency in narrative video generation.
-
Rethinking Position Embedding as a Context Controller for Multi-Reference and Multi-Shot Video Generation
SideInfo-RoPE encodes reference-identity agreement as an extra rotary axis, disambiguating similar characters in multi-reference multi-shot video generation while keeping full semantic attention.
-
End-to-End Training for Autoregressive Video Diffusion via Self-Resampling
Resampling Forcing trains autoregressive video diffusion models on self-resampled degraded histories with a causal mask, achieving stable long-horizon generation without a teacher or discriminator.
-
VideoCanvas: Unified Video Completion from Arbitrary Spatiotemporal Patches via In-Context Conditioning
A single diffusion model with in-context conditioning and fractional RoPE positions completes videos from arbitrary spatio-temporal image patches.
-
LoViC: Efficient Long Video Generation with Context Compression
LoViC uses FlexFormer, a single-query-token Q-Former with interpolated rotary positional encoding, to compress long video-text context for efficient long-video generation.
-
FreeLong++: Training-Free Long Video Generation via Multi-band SpectralFusion
FreeLong++ extends short-video diffusion models to 4x to 8x longer clips, without retraining, by fusing multiple windowed attention branches through frequency-domain filters and a spectral noise initialization.
-
Cross-Frame Representation Alignment for Fine-Tuning Video Diffusion Models
CREPA fine-tunes video diffusion models with a loss that pulls each frame's hidden representation toward pretrained features of adjacent frames, improving semantic consistency in generated videos.
-
Context as Memory: Scene-Consistent Interactive Long Video Generation with Memory Retrieval
Context-as-Memory conditions video generation on selected historical frames chosen by camera FOV overlap, improving scene consistency in long generated videos.
-
AnimeShooter: A Multi-Shot Animation Dataset for Reference-Guided Video Generation
AnimeShooter provides hierarchical story and shot annotations plus reference images for 148K one-minute animation stories, and AnimeShooterGen trained on it shows improved cross-shot consistency.
-
Towards Sparse Video Understanding and Reasoning
A video-QA agent that carries only a structured text summary between rounds beats dense-frame baselines on accuracy while using a handful of frames per video.
-
BIFE: Better Interaction, Fewer Errors for Minute-Long Video Generation
BlockVid generates minute-long videos with a semantic sparse KV cache, Block Forcing training, and chunk-level noise scheduling, reporting large gains on its own LV-Bench and on VBench.
-
From Black Box to Transparency: Enhancing Automated Interpreting Assessment with Explainable AI in College Classrooms
An explainable model using BLEURT, CometKiwi, pause features, and Chinese phraseological diversity predicts human-rated quality dimensions in English-Chinese consecutive interpreting, with SHAP identifying the stronge...
-
Captain Cinema: Towards Short Movie Generation
A two-stage text-to-movie system that plans keyframes for the story and then synthesizes video between them, using a compressed memory bank to keep long narratives consistent.
-
FullDiT2: Efficient In-Context Conditioning for Video Diffusion Transformers
FullDiT2 accelerates FullDiT-style in-context conditioning for video by dynamic token selection and selective context caching, cutting per-step time by 2-3x with minimal quality loss.
-
Frame-Level Captions for Long Video Generation with Complex Multi Scenes
Frame-level captions with per-frame cross-attention and parallel multi-window denoising reduce semantic confusion in long multi-scene video generation in the authors' internal evaluation.
-
Hierarchical Fine-grained Preference Optimization for Physically Plausible Video Generation
A hierarchical direct preference optimization with four alignment levels plus automated data selection improves physical plausibility of text-to-video models.
Discussion (0). Continue with ORCID to comment.