Few-Step Streaming Video Generation via Condition Shifting
cs.CV, cs.AI
Submitted: 2026-11-10
Updated: 2026-11-10
Code: https://github.com/LAION-AI/aesthetic-predictor
Project page: https://attention-survey.github.io/files/Attention_Survey.pdf
Terminology
Sources
- LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes
- SkyReels-V2: Infinite-length Film Generative Model
- Grounded Forcing: Bridging Time-Independent Semantics and Proximal Dynamics in Autoregressive Video Synthesis
- AvatarForcing: One-Step Streaming Talking Avatars via Local-Future Sliding-Window Denoising
- StreamDiffusionV2: A Streaming System for Dynamic and Interactive Video Generation
- OmniAvatar: Efficient Audio-Driven Avatar Video Generation with Adaptive Body Animation
- Ca2-VDM: Efficient Autoregressive Video Diffusion Model with Causal Generation and Cache Sharing
- Wan-S2V: Audio-Driven Cinematic Video Generation
- Mean Flows for One-step Generative Modeling
- ImageBind: One Embedding Space To Bind Them All
- FlowCPO: A Unified Divergence View of Preference Alignment for Flow Models
- ChronoReplay: One Answer Interface Is Not Enough for Streaming Spatial Intelligence
- Supervising the Visible State: Separating Representation from View Selection in Streaming Video
- Speculative Decoding for Autoregressive Video Generation
- VTR-Bench: A Systematic Benchmark for Evaluating Visual Text Rendering in Video Generation
- Live Avatar: Streaming Real-time Audio-Driven Avatar Generation with Infinite Length
- Synchformer: Efficient Synchronization from Sparse Cues
- DeepSpeed Ulysses: System Optimizations for Enabling Training of Extreme Long Sequence Transformer Models
- TurboServe: Serving Streaming Video Generation Efficiently and Economically
- Distribution Matching Distillation without Fake Score Network
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs