Open mtk380 opened 1 year ago
Hi, thank you very much for your question.
I roughly tested the inference time on my machine (Nvidia RTX 3090). For the synthetic scenes, rendering a full-resolution view will take ~5s. For the forward-facing scenes, it will take ~15s.
This RecolorNeRF implementation is based on TensoRF, in which no customized CUDA kernels or advanced structures are designed for rendering acceleration. Thus, real-time color editing is not in our consideration. Nevertheless, our approach can work without restrictions to any backbones. To achieve real-time performance, we may suggest adapting our approach to other backbones, e.g., PlenOctrees and KiloNeRF.
Thanks for your great work! I'm interested in the reference time for one image. If I change the color in Real-time, how much should the maximum resolution be?