One question about the brightness of the outcome from pertained model

SystemErrorWang / White-box-Cartoonization

Official tensorflow implementation for CVPR2020 paper “Learning to Cartoonize Using White-box Cartoon Representations”

3.95k stars 737 forks source link

One question about the brightness of the outcome from pertained model #68

Open zhaowt61 opened 3 years ago

zhaowt61 commented 3 years ago

Hi Dr. Wang, thanks for your excellent work! I noticed that the brightness of the outcome from the pertained model is significantly higher than the one I reproduced using exactly the same dataset as yours. I am wondering whether you have any pre-processing procedures such as increasing the exposure/brightness of the anime dataset?

SystemErrorWang commented 3 years ago

@zhaowt61 as is shown in figure5 of the paper, pre-process may help increase the final results of the model. Also, you can adopt post process to adjust the color/brightness/exposure

zhaowt61 commented 3 years ago

Hi Dr. Wang, thanks for your reply, could I possibly have you email for further communication? Thank you so much! @SystemErrorWang

SystemErrorWang commented 3 years ago

@zhaowt61 feel free to contact me at secretwang@outlook.com

zhaowt61 commented 3 years ago

@SystemErrorWang Thank you so much, Dr. Wang! 😄