Causal-Forcing#

Causal-Forcing uses Causal ODE or Causal Consistency Distillation to drive asymmetric DMD as a theoretically correct initialization for real-time interactive video generation.

Teaser image source: Causal-Forcing project page.

Requirements#

Minimum VRAM: ~24 GB.
PyTorch: >= 2.9.

Installation#

# from the repo root
uv sync --project integrations_v2/causal_forcing

Running the method#

To run Causal-Forcing, launch its v2 T2V application:

uv run --project integrations_v2/causal_forcing \
    flashdreams-run-v2 \
    t2v-causal-forcing-wan2.1-t2v-1.3b-chunkwise \
    --output-path artifacts/t2v-causal-forcing-wan2.1-t2v-1.3b-chunkwise.mp4 -- \
    --prompt "A cinematic closeup and detailed portrait of a reindeer standing in a snowy forest at sunset. The lighting is gorgeous and soft, with a golden backlight creating a warm and dreamy effect. Soft bokeh and lens flares add a magical touch, enhancing the cinematic quality of the image. The reindeer has a gentle expression, its fur glistening in the fading light. The background features a serene snowy landscape with tall trees silhouetted against the orange and pink hues of the setting sun. The color grade is rich and magical, capturing the essence of a winter wonderland at twilight. A close-up shot from a slightly elevated angle." \
    --total-blocks 21

For multi-GPU inference, run the same command under torchrun (taking 4 GPUs as an example):

uv run --project integrations_v2/causal_forcing \
    torchrun --nproc_per_node=4 --no-python flashdreams-run-v2 \
    t2v-causal-forcing-wan2.1-t2v-1.3b-chunkwise \
    --output-path artifacts/t2v-causal-forcing-wan2.1-t2v-1.3b-chunkwise.mp4 -- \
    --prompt "A cinematic closeup and detailed portrait of a reindeer standing in a snowy forest at sunset. The lighting is gorgeous and soft, with a golden backlight creating a warm and dreamy effect. Soft bokeh and lens flares add a magical touch, enhancing the cinematic quality of the image. The reindeer has a gentle expression, its fur glistening in the fading light. The background features a serene snowy landscape with tall trees silhouetted against the orange and pink hues of the setting sun. The color grade is rich and magical, capturing the essence of a winter wonderland at twilight. A close-up shot from a slightly elevated angle." \
    --total-blocks 21

The package also exposes the following pipeline configs for direct use:

Method	Description
`causal-forcing-wan2.1-t2v-1.3b-chunkwise`	Causal-Forcing chunkwise Wan 2.1 1.3B T2V (`len_t=3`).
`causal-forcing-wan2.1-t2v-1.3b-framewise`	Causal-Forcing framewise Wan 2.1 1.3B T2V (`len_t=1`).
`causal-forcing-wan2.1-i2v-1.3b-framewise`	Causal-Forcing framewise Wan 2.1 1.3B I2V (`len_t=1`).

To inspect all supported CLI arguments and their default values, run:

uv run --project integrations_v2/causal_forcing \
    flashdreams-run-v2 t2v-causal-forcing-wan2.1-t2v-1.3b-chunkwise -- --help

Some generated samples from the above commands:

prompt: "A cinematic closeup and detailed portrait of a reindeer standing in a snowy forest at sunset. The lighting is gorgeous and soft, with a golden backlight creating a warm and dreamy effect. Soft bokeh and lens flares add a magical touch, enhancing the cinematic quality of the image. The reindeer has a gentle expression, its fur glistening in the fading light. The background features a serene snowy landscape with tall trees silhouetted against the orange and pink hues of the setting sun. The color grade is rich and magical, capturing the essence of a winter wonderland at twilight. A close-up shot from a slightly elevated angle."

prompt: "A cinematic closeup and detailed portrait of a reindeer standing in a snowy forest at sunset. The lighting is gorgeous and soft, with a golden backlight creating a warm and dreamy effect. Soft bokeh and lens flares add a magical touch, enhancing the cinematic quality of the image. The reindeer has a gentle expression, its fur glistening in the fading light. The background features a serene snowy landscape with tall trees silhouetted against the orange and pink hues of the setting sun. The color grade is rich and magical, capturing the essence of a winter wonderland at twilight. A close-up shot from a slightly elevated angle."
image: https://raw.githubusercontent.com/thu-ml/Causal-Forcing/refs/heads/main/prompts/i2v/26-15/000001.png

Citation#

If you use Causal-Forcing, please cite the original work:

@article{zhu2026causal,
  title={Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation},
  author={Zhu, Hongzhou and Zhao, Min and He, Guande and Su, Hang and Li, Chongxuan and Zhu, Jun},
  journal={arXiv preprint arXiv:2602.02214},
  year={2026}
}