Although our method, like most that leverage pose-conditioned diffusion models, works best on in-distribution, structured captures and synthetic data, we also evaluate it on in-the-wild captures as well.