Multiview Rendering

[hasan1292]

Can you please give more instructions about how to do multiview rendering? Single view rendering is done automatically in the code. Need to know what to do in order to do multi-view rendering?

[Friedrich-M]

Sure! I'll work on adding more detailed instructions for multi-view rendering right away. Thank you for the helpful suggestion!

[Friedrich-M]

Hey, I have updated the instructions. For rendering mode, you could add "--render_only 1" argument. "--render_train 1" indicates rendering training views; "--render_test 1" indicates rendering test views; "--render_path 1" means rendering moving camera results.

Hope it could help you!

[hasan1292]

Hi, Thank you so much for your prompt response. I really appreciate it.

I was able to generate the image based on your command.

video.mp4

I would like to ask few of more questions:

1. Your training works well on DyNerf data. But what if I have to use DyCheck1 or DyCheck2 dataset. Will I use the same training and rendering command?
2. Can you please provide links to DyCheck and Google Immersive dataset like you provided DyNeRF dataset link?

[Friedrich-M]

For question 1, we use NeRFStudio codebase for the DyCheck dataset, you could dive into the nerfplayer-nerfstudio directory of this repo and follow the instructions below for training and rendering.

For question 2, the Dycheck dataset link is https://drive.google.com/drive/folders/1cBw3CUKu2sWQfc_1LbFZGbpdQyTFzDEX and immersive is https://github.com/augmentedperception/deepview_video_dataset.