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

Some questions about HDR-VDP-2. #3

Closed Tx000 closed 3 years ago

Tx000 commented 3 years ago

Thanks for releasing the code and dataset! I have some questions about HDR-VDP-2: First, I found that you adopted 'sRGB-display' as the mode of HDR-VDP-2. I think this mode is for inputting images in the pixel domain. In this mode, HDR-VDP-2 will map input images from the pixel domain to the linear domain. However, the input images in your code are normalized HDR (linear) images. Second, other modes in HDR-VDP-2 require absolute luminance values rather than normalized values. I wonder how to compute this metric. Thank you!

guanyingc commented 3 years ago

Hi, thanks for your question.

  1. I am kind of forgetting the details when computing the HDR-VDP-2 metric. I checked the code and found that I do use the 'sRGb-display' mode, which might be improper.
  2. I think you need raw exposure time and gain to get the absolute luminance value. They are included in the metadata of the captured real data. However, I haven't used them. We just used the relative exposure.
Tx000 commented 3 years ago

Thanks for your responses!

syujung commented 2 years ago

@guanyingc Hello , if we transfer the linear images to pixel domain(use gamma ) before computing the HDR-VDP(use sRGB-display mode) , can we get the correct metric values? And can we use the exposure time to get the absolute luminance (use other modes {luminance ,luma-display, etc) to compute HDR-VDP ? Which mode might be better ?

guanyingc commented 2 years ago

Hi, in fact, I am not very sure, but you can give it a try.

Since we are comparing the results of different methods, it might not be a big problem if the evaluation process is the same for all methods.