yifanai / video2anime

Turn your videos (and selfies) into anime with generative adversarial network (GAN)
Other
26 stars 14 forks source link

Did you do some temporal constraint on training? #7

Closed Keysmis closed 10 months ago

Keysmis commented 4 years ago

hello~, did you do some temporal constraint on training? i find that female videos has not a lot flicker~~

yifanai commented 4 years ago

Hi, training was done by https://github.com/taki0112/UGATIT. I don't think they used temporal constraint during training, because the model was trained using only images. What do you mean exactly by flicker? It might be due to lighting when the video is recorded.