ControlVideo: Training-free Controllable Text-to-Video Generation

By Yabo Zhang, Yuxiang Wei, Dongsheng Jiang, Xiaopeng Zhang, Wangmeng Zuo, Qi Tian. International Conference on Learning Representations, 2023

369 citations recorded by the source corpus.

Text-driven diffusion models have unlocked unprecedented abilities in image generation, whereas their video counterpart still lags behind due to the excessive training cost of temporal modeling. Besides the training burden, the generated videos also suffer from appearance inconsistency and structural flickers, especially in long video synthesis. To address these challenges, we design a \emph{training-free} framework called \textbf{ControlVideo} to enable natural and efficient text-to-video generation. ControlVideo, adapted from ControlNet, leverages coarsely structural consistency from input motion sequences, and introduces three modules to improve video generation. Firstly, to ensure appearance coherence between frames, ControlVideo adds fully cross-frame interaction in self-attention modules. Secondly, to mitigate the flicker effect, it introduces an interleaved-frame smoother that employs frame interpolation on alternated frames. Finally, to produce long videos efficiently, it utilizes a hierarchical sampler that separately synthesizes each short clip with holistic coherency. Empowered with these modules, ControlVideo outperforms the state-of-the-arts on extensive motion-prompt pairs quantitatively and qualitatively. Notably, thanks to the efficient designs, it generates both short and long videos within several minutes using one NVIDIA 2080Ti. Code is available at https://github.com/YBYBZhang/ControlVideo.

Subfields: Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition, Control and Systems Engineering

View Paper: https://www.semanticscholar.org/paper/529191401a8a5f0a8bdb2a1c01301d76af585a3a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Graphics and Computer-Aided Design

1. Ying Shan
2. Yu Qiao
3. Ziwei Liu
4. Yong Zhang
5. Xintao Wang

In Computer Vision and Pattern Recognition

1. Yu Qiao
2. Luc Van Gool
3. Chen Change Loy
4. Ying Shan
5. Ziwei Liu
6. Limin Wang
7. Xiaoou Tang
8. Xintao Wang

In Control and Systems Engineering

1. Yu Qiao
2. Limin Wang
3. Ying Shan
4. Ziwei Liu
5. Luc Van Gool
6. Cordelia Schmid

Citations in the corpus

• VBench: Comprehensive Benchmark Suite for Video Generative Models (2023, background): “Other guidance modalities are also available, including image-to-video [14, 16, 23, 122], video-to-video [12, 67, 80, 82, 121], and a variety of con-trol maps [15, 51, 58, 73, 103, 125, 126] such as pose, depth, and sketch.”
• VACE: All-in-One Video Creation and Editing (2025, positive): “…[8] and M3DDM [17] are compared for outpainting; 3) For controllable task, we use Control-A-Video [10], VideoComposer [68], and ControlVideo [79] under depth conditions, and compare Text2Video-Zero [31], ControlVideo [79], and Follow-Your-Pose [40] under pose conditions, as well as…”
• CameraCtrl: Enabling Camera Control for Text-to-Video Generation (2024, neutral): “To provide enhanced guidance, some works adopt other signals, e.g. , depth/skeleton sequence, to precisely control the scene/human motion in the generated videos (Guo et al., 2023a; Chen et al., 2023c; Zhang et al., 2023c; Khachatryan et al., 2023; Hu et al., 2023; Xu et al., 2023).”
• DynamiCrafter: Animating Open-domain Images with Video Diffusion Priors (2023, methodology): “Similar to adding controls in T2I [48, 61, 88, 93], introducing control signals in T2V, such as structure [17, 83], pose [44, 94], and Canny edge [39], has been increasingly receiving much attention.”