If the model is automatically downsizing the image, is there any advantage to using bigger than the input image? Just seems like it would take up more space. I have heard some people say that if you use too big of images, you can end up with blurry results. I have 512x512 and 256x256 images.
Search found 3 matches
- Mon Apr 26, 2021 8:19 pm
- Forum: Training Support
- Topic: Cant use images smaller than 256x256, just stops with error 3221225620
- Replies: 3
- Views: 6454
- Mon Apr 26, 2021 7:43 pm
- Forum: Training Support
- Topic: Cant use images smaller than 256x256, just stops with error 3221225620
- Replies: 3
- Views: 6454
Cant use images smaller than 256x256, just stops with error 3221225620
Anytime I try to train with a images that are either 128x128 or 64x64, faceswap just stops with error 3221225620. No crash logs are generated either. On github the topic had come up, but the solution they used was to reinstall tensor flow. That doesnt help at all. https://github.com/deepfakes/facesw...
- Thu Jan 14, 2021 10:13 pm
- Forum: Extract Support
- Topic: Extraction keeps crashing
- Replies: 26
- Views: 28851
Extraction keeps crashing
i9-9900K 128GB Ram GTX-1080 <- never detected so I only use cpu 500W PSU When extracting faces, it always crashes about half way through, no matter what I set any of the settings to. Usually gives me this "cv2.error: OpenCV(4.5.1) C:\Users\appveyor\AppData\Local\Temp\1\pip-req-build-kh7iq4w7\op...