Why you use 1088x640 instead of the original 1920x1080 for Tanks&Temples?

[immortalCO]

Hi,

We evaluated your code on the Tanks&Temples dataset and found that the images outputted by your model have a size (resolution) of 1088x640‬, with a different ratio from the original image size of 1920x1080. It's also shown in your config file that you manually set it as 1088x640‬.

For Table 9 in your paper, you compared the results with NSVF, and the numbers of NSVF were the same as those in NSVF paper, which also claims that "The image resolution is 1920 × 1080" for Tanks&Temples.

Why did you choose the resolution of 1088x640? Were the metrics in Table 9 in your paper calculated with images with a resolution of 1088x640 or 1920x1080?

Thanks.

[Xharlie]

Hi Jukun, we are using 1080Ti by the time and the memory is very small comparing with FB's clusters used by NSVF. 1088x640‬ is the largest limit my gpu can handle. If the resolution will affect number, I think I can make adjustment by isolating our results as a single table. If you want to compare with us, you can also downsample to 1088x640‬ both gt test view and the generated test view.

[immortalCO]

Thank you for your clarification! I fully understand that GPU memory is a very constraintful issue in this case.

Actually, I would suggest that instead of putting your results in a separate table, you could just upsample the 1088x640 output image to 1920x1080, so that it could be directly compared with the 1920x1080 ground truth, and the numbers and experiment settings will be then consistent with NSVF.