Research map / Video / Video generation
StreamDiT: Real-Time Streaming Text-to-Video Generation
Akio Kodaira, Tingbo Hou, Ji Hou, Markos Georgopoulos, Felix Juefei-Xu, Masayoshi Tomizuka, Yue Zhao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train StreamDiT with all chunk sizes [1,2,4,8,16] mixed instead of chunk size 1 only (ablation) | StreamDiT-4B · VBench mixed-training ablation | Quality Score: 0.8129 -> 0.8144 | Improved |
| mix chunk sizes [1,2] with chunk size 1 (ablation) | StreamDiT-4B · VBench mixed-training ablation | Quality Score: 0.8129 -> 0.8100 | Hurt |
| mix chunk sizes [1,2,4] (ablation) | StreamDiT-4B · VBench mixed-training ablation | Quality Score: 0.8129 -> 0.8080 | Hurt |
| mix chunk sizes [1,2,4,8] (ablation) | StreamDiT-4B · VBench mixed-training ablation | Quality Score: 0.8129 -> 0.8076 | Hurt |
| StreamDiT training instead of FIFO-Diffusion inference-only streaming | StreamDiT-4B · VBench | Quality Score: 0.7981 -> 0.8185 | Improved |
| StreamDiT training instead of ReuseDiffuse | StreamDiT-4B · VBench | Quality Score: 0.8019 -> 0.8185 | Improved |
| StreamDiT instead of FIFO-Diffusion | StreamDiT-4B · VBench | Dynamic Degree: 0.3094 -> 0.5240 | Improved |
| StreamDiT instead of FIFO-Diffusion | StreamDiT-4B · VBench | Temporal Flickering: 0.9796 -> 0.9671 | Hurt |
| StreamDiT instead of FIFO-Diffusion | StreamDiT-4B · VBench | Motion Smoothness: 0.9889 -> 0.9861 | Hurt |
| StreamDiT instead of FIFO-Diffusion | StreamDiT-4B · VBench | Subject Consistency: 0.9412 -> 0.9622 | Improved |
| multistep distillation to 8 steps without CFG | StreamDiT-4B · VBench | Quality Score: 0.8185 -> 0.8163 | No clear effect |
| multistep distillation to 8 steps without CFG | StreamDiT-4B · VBench | Dynamic Degree: 0.5240 -> 0.7040 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FIFO-Diffusion — FIFO-Diffusion: Generating Infinite Videos from Text without Training (NeurIPS 2024) · better on 4 of 7 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- StreamingT2V: Consistent, Dynamic, and Extendable Long Video Generation from Text (CVPR 2025 · similarity 0.89)
- From Slow Bidirectional to Fast Autoregressive Video Diffusion Models (CVPR 2025 · similarity 0.87)
- Looking Backward: Streaming Video-to-Video Translation with Feature Banks (ICLR 2025 · similarity 0.87)
- LinGen: Towards High-Resolution Minute-Length Text-to-Video Generation with Linear Computational Complexity (CVPR 2025 · similarity 0.87)
- Grid Diffusion Models for Text-to-Video Generation (CVPR 2024 · similarity 0.87)
- STREAM: Spatio-TempoRal Evaluation and Analysis Metric for Video Generative Models (ICLR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
