search

guanyingc / DeepHDRVideo

HDR Video Reconstruction: A Coarse-to-fine Network and A Real-world Benchmark Dataset (ICCV 2021)
https://guanyingc.github.io/DeepHDRVideo
Other
98 stars 19 forks source link

About the Tog13 datasets #12

Closed syujung closed 2 years ago

syujung commented 2 years ago

Dear guanying， The file (TOG13 dataset) you provided in google drive already contains the hdr images , I want to know if those hdr images are predictd by the network you proposed in paper ？

guanyingc commented 2 years ago

Hi, those HDR images are results of the Kalantari 13's method.

syujung commented 2 years ago

Ok, I noted you used the pre-alignment when testing on TOG13 datasets，and you do not use pre-alignment when testing on datasets you proposed ，I wonder if the pre-alignment is necessary for testing on TOG13 datasets？ And if the algorithm is robust to handle the global motion，do we need to prealign the images before we put images into the network ？

syujung commented 2 years ago

Thank you again for your patient reply！ Your reply helped me a lot！

guanyingc commented 2 years ago

We do not pre-align when testing on our dataset because we are evaluating the robustness to global motion. I think global image alignment is helpful as it reduce the difficulty in HDR fusion.