DaReNeRF introduces a direction-aware representation to address the limitations of traditional wavelet-based representations in capturing high-fidelity dynamic scenes. The method achieves state-of-the-art performance with reduced training time and model size across both dynamic and static scenes. By leveraging trainable masks, DaReNeRF maintains sparsity while delivering superior reconstruction quality compared to DWT-based solutions.
A otro idioma
del contenido fuente
arxiv.org
Ideas clave extraídas de
by Ange Lou,Ben... a las arxiv.org 03-05-2024
https://arxiv.org/pdf/2403.02265.pdfConsultas más profundas