EDIT: It seems to be a resolution issue and/or find_best_frame = true
Additionally, I was using a source image PNG of 512x512 and a training video of 512x512, which is not consistent with the assets provided in the demo (jackie chan and donald trump).
s there a way to edit the config to accomodate for a different resolution. or enable find_best_frame=true?
The Google Colab notebook uses CPU only.
I replaced the source.png and driver.mp4 (named them different files and then replaced their definitions in the config). Haven't changed any configs except for find_best_frame = TRUE
At the last step (make_animation), I am getting the following error:
EDIT: It seems to be a resolution issue and/or
find_best_frame = true
Additionally, I was using a source image PNG of 512x512 and a training video of 512x512, which is not consistent with the assets provided in the demo (jackie chan and donald trump).
s there a way to edit the config to accomodate for a different resolution. or enable find_best_frame=true?
The Google Colab notebook uses CPU only.
I replaced the source.png and driver.mp4 (named them different files and then replaced their definitions in the config). Haven't changed any configs except for
find_best_frame = TRUE
At the last step (make_animation), I am getting the following error:
`Downloading: "https://www.adrianbulat.com/downloads/python-fan/s3fd-619a316812.pth" to /root/.cache/torch/hub/checkpoints/s3fd-619a316812.pth 100% 85.7M/85.7M [00:00<00:00, 176MB/s] Downloading: "https://www.adrianbulat.com/downloads/python-fan/2DFAN4-cd938726ad.zip" to /root/.cache/torch/hub/checkpoints/2DFAN4-cd938726ad.zip 100% 91.9M/91.9M [00:00<00:00, 140MB/s]
TypeError Traceback (most recent call last) in
4 if predict_mode=='relative' and find_best_frame:
5 from demo import find_best_frame as _find
----> 6 i = _find(source_image, driving_video, device.type=='cpu')
7 print ("Best frame: " + str(i))
8 driving_forward = driving_video[i:]
/content/Thin-Plate-Spline-Motion-Model/demo.py in find_best_frame(source, driving, cpu) 107 fa = face_alignment.FaceAlignment(face_alignment.LandmarksType._2D, flip_input=True, 108 device= 'cpu' if cpu else 'cuda') --> 109 kp_source = fa.get_landmarks(255 * source)[0] 110 kp_source = normalize_kp(kp_source) 111 norm = float('inf')
TypeError: 'NoneType' object is not subscriptable`
Can anyone help?