some issue about quantitative results of novel-view synthe Table.3

[lifftel]

Thanks for your great work. what training config do you use to compute Table 3: Novel View Synthesis in the paper? I run 392_mono_4gpu.yaml to train our model and the final results are 0.104, 27.34, and 0.926 which are less than your report results.

or you run 392_4gpu.yaml config? and what start_frame, end_frame, and subsampling_rate you use to compute these metrics?

[taconite]

Hi, thanks for your interest in our work!

Table 3 in the paper reports 392_4gpus.yaml, which is the standard 4-view training setup. mono means monocular setup - we did not report quantitative results for the monocular setup.

For the novel view evaluation of 392, the frame range is [0, 300) (exclusive on the right) and the sampling rate is 30.

[taconite]

Close the issue due to inactivity.