OmniVCus is a video generation system that personalizes videos based on subject images while accepting multiple control conditions such as text prompts and reference poses. The implementation uses diffusion models and transformer architectures to enable feedforward video customization, trained on custom datasets and released with model weights on Hugging Face.