File size: 5,820 Bytes
aa099ef | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 | ---
license: apple-amlr
pipeline_tag: text-to-video
base_model:
- Wan-AI/Wan2.1-T2V-1.3B
- zhuhz22/Causal-Forcing
---
# FastFrames: Video Generation at Interactive Latency
[](https://apple-aiml-research.github.io/ml-fastframes/)
[](https://arxiv.org/abs/2611.23226)
[](https://huggingface.co/apple/FastFrames-S)
[](https://huggingface.co/apple/FastFrames-L)
This repository contains the official implementation of *FastFrames: Video Generation at
Interactive Latency*. We release [FastFrames-S](https://huggingface.co/apple/FastFrames-S)
(1.3B) and [FastFrames-L](https://huggingface.co/apple/FastFrames-L) (14B). Both are
autoregressive, two-step video diffusion models that generate 480p video at interactive
latency. We also release an optimized version of LingBot 1.3B that is 16x faster than the
official model and allows fluid camera-conditioned control.
- [Latency profiling](#latency-profiling)
- [Interactive demo with LingBot](#interactive-demo-with-lingbot)
- [T2V generation](#t2v-generation)
- [VBench evaluation](#vbench-evaluation)
- [Training](#training)
- [Citation](#citation)
[](assets/fastframes_demo.mp4)
## Installation
Linux x86-64 with an NVIDIA GPU and a CUDA 12.8-compatible driver. From the repository root:
```bash
uv sync
```
Download FastFrames and third-party weights using the [checkpoint guide](https://github.com/apple-aiml-research/ml-fastframes/blob/main/docs/installation.md#checkpoints).
## Latency Profiling
```bash
uv run python evaluation/latency/benchmark.py ff-s
uv run python evaluation/latency/benchmark.py ff-l --gpus 8
```
Example component timings for **FastFrames-S, 1 GPU, 480p**, with a full KV cache.
| Component | Calls | Latency per call | Total latency |
|---|---:|---:|---:|
| DiT evaluation, t=1000 | 1 | 16.1 ms | 16.1 ms |
| DiT evaluation, t=833 | 1 | 16.2 ms | 16.2 ms |
| KV refinement | 1 | 2.2 ms | 2.2 ms |
| Pixel decoder | 1 | 0.9 ms | 0.9 ms |
| **Total** | | | **35.4 ms** (paper: 36 ms) |
See the [profiling guide](https://github.com/apple-aiml-research/ml-fastframes/blob/main/docs/inference.md#profiling) for details and baselines.
## Interactive Demo with LingBot
Complete the [demo setup](https://github.com/apple-aiml-research/ml-fastframes/blob/main/docs/lingbot.md#interactive-demo-13b), then launch one of:
```bash
CUDA_VISIBLE_DEVICES=0 uv run --extra lingbot --extra demo python apps/lingbot/demo.py --gpus 1
CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 uv run --extra lingbot --extra demo python apps/lingbot/demo.py --gpus 8
```
and open `http://localhost:8000`.
## T2V Generation
```bash
uv run python scripts/generate.py --model ff-s --checkpoint-dir checkpoints \
--prompt "A corgi surfing a wave at sunset" --seed 0 --output outputs/corgi.mp4
```
Outputs 832 × 480 video, 81 frames at 16 fps. Use `--model ff-l` for FastFrames-L.
To generate the videos on our project page (20+ videos):
```bash
uv run python scripts/generate_project_page.py --model ff-s --output-dir outputs/project_page/ff-s
uv run python scripts/generate_project_page.py --model ff-l --output-dir outputs/project_page/ff-l
```
See the [generation guide](https://github.com/apple-aiml-research/ml-fastframes/blob/main/docs/inference.md#generation) for details.
## VBench Evaluation
Follow the [evaluation guide](https://github.com/apple-aiml-research/ml-fastframes/blob/main/docs/inference.md#evaluation) to reproduce Table 1's
Quality / Semantic / Total scores.
## Training
### [Causal Forcing++ -> FastFrames-S](https://github.com/apple-aiml-research/ml-fastframes/blob/main/docs/training.md#causal-forcing---fastframes-s)
FastFrames-S keeps the two-step frame-wise Causal Forcing++ generator frozen and trains the
networks around it.
### [Wan2.1-14B -> FastFrames-L](https://github.com/apple-aiml-research/ml-fastframes/blob/main/docs/training.md#wan21-14b---fastframes-l)
We first distill Wan2.1-14B into a two-step causal model following the Causal Forcing++ recipe,
then train its pixel decoder and KV refinement network as for FastFrames-S. We release all
checkpoints: 1) a causal teacher trained on latents generated by Wan2.1-14B; 2) the
consistency-distilled model; 3) the DMD-distilled model; 4) the final generator, pixel decoder
and KV refinement network.
## Citation
If you find this work useful in your research, please cite:
```bibtex
@article{chou2026fastframes,
title = {FastFrames: Video Generation at Interactive Latency},
author = {Chou, Gene and Yang, Guandao and Luan, Fujun and Mescheder, Lars and
Dong, Wei and Richter, Stephan R. and Koltun, Vladlen},
journal = {arXiv preprint arXiv:2611.23226},
year = {2026}
}
```
## License and Acknowledgments
LingBot-World v2 code and weights follow their upstream [CC BY-NC-SA 4.0 license](https://github.com/apple-aiml-research/ml-fastframes/blob/main/LICENSE_LINGBOT).
All other FastFrames code is licensed under the [Apple Sample Code License](https://github.com/apple-aiml-research/ml-fastframes/blob/main/LICENSE).
Also see [ACKNOWLEDGEMENTS](https://github.com/apple-aiml-research/ml-fastframes/blob/main/ACKNOWLEDGEMENTS) for third-party code.
We thank the authors of [Wan2.1](https://github.com/Wan-Video/Wan2.1),
[Self-Forcing](https://github.com/guandeh17/Self-Forcing), and
[Causal Forcing++](https://github.com/thu-ml/Causal-Forcing) for releasing their code, on which the
FastFrames code and models build.
|