Search papers, labs, and topics across Lattice.
CT-1, a novel Vision-Language-Camera model, is introduced to improve camera-controllable video generation by accurately estimating camera trajectories using spatial reasoning. The model leverages vision-language modules and a Diffusion Transformer, incorporating a Wavelet-based Regularization Loss in the frequency domain to learn complex camera trajectory distributions. Trained on a newly curated dataset, CT-200K, CT-1 achieves a 25.7% improvement in camera control accuracy, demonstrating its ability to generate faithful and high-quality videos with spatially aware camera control.
Forget manual camera trajectories: CT-1 learns to generate them automatically, bridging the gap between spatial reasoning and high-quality video synthesis.
Camera-controllable video generation aims to synthesize videos with flexible and physically plausible camera movements. However, existing methods either provide imprecise camera control from text prompts or rely on labor-intensive manual camera trajectory parameters, limiting their use in automated scenarios. To address these issues, we propose a novel Vision-Language-Camera model, termed CT-1 (Camera Transformer 1), a specialized model designed to transfer spatial reasoning knowledge to video generation by accurately estimating camera trajectories. Built upon vision-language modules and a Diffusion Transformer model, CT-1 employs a Wavelet-based Regularization Loss in the frequency domain to effectively learn complex camera trajectory distributions. These trajectories are integrated into a video diffusion model to enable spatially aware camera control that aligns with user intentions. To facilitate the training of CT-1, we design a dedicated data curation pipeline and construct CT-200K, a large-scale dataset containing over 47M frames. Experimental results demonstrate that our framework successfully bridges the gap between spatial reasoning and video synthesis, yielding faithful and high-quality camera-controllable videos and improving camera control accuracy by 25.7% over prior methods.
