Any-to-3D Generation via Hybrid Diffusion Supervision
Any-to-3D Generation via Hybrid Diffusion Supervision
Recent progress in 3D object generation has been fueled by the strong priors offered by diffusion models. However, existing models are tailored to specific tasks, accommodating only one modality at a time and necessitating retraining to change modalities. Given an image-to-3D model and a text prompt, a naive approach is …