Convert not swapping faces for every frame of input video.

Converting source frames into swapped frames
Forum rules
Read the FAQs and search the forum before posting a new topic.

Please mark any answers that fixed your problems so others can find the solutions.
Post Reply
User avatar
winter0s
Posts: 2
Joined: Mon Oct 28, 2019 10:35 pm

Convert not swapping faces for every frame of input video.

Post by winter0s » Mon Oct 28, 2019 11:02 pm

Hey guys,

A Faceswap rookie here! So, I followed extraction and training guides posted here and ended up with a trained model which turned out to be decent. I used every 6th image for training. When I convert, Faceswap swaps every 6 frame to Face B. What's the point of training with reduced face set then? Should I just train with all the frames from Face A? Or Am I doing it wrong? Any help will be much appreciated :)

Here is my convert setup -


Image

And here is the result-

Code: Select all

Loading...
Setting Faceswap backend to NVIDIA
10/28/2019 17:47:38 INFO     Log level set to: INFO
10/28/2019 17:47:40 INFO     Input Video: D:\_FL\InputA\t_video6334477867065278594.mp4
10/28/2019 17:47:40 INFO     Output Directory: D:\_FL\OUTPUT
10/28/2019 17:47:40 INFO     Reformatting legacy alignments file...
10/28/2019 17:47:40 INFO     Old location: 'D:\_FL\InputA\t_video6334477867065278594_alignments.json', New location: 'D:\_FL\InputA\t_video6334477867065278594_alignments.fsa'
10/28/2019 17:47:40 INFO     Reading alignments from: 'D:\_FL\InputA\t_video6334477867065278594_alignments.fsa'
10/28/2019 17:47:40 INFO     Loading Writer from Ffmpeg plugin...

10/28/2019 17:47:41 INFO     Using configuration saved in state file
10/28/2019 17:47:41 WARNING  From C:\Users\******\MiniConda3\envs\faceswap\lib\site-packages\keras\backend\tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n
10/28/2019 17:47:41 WARNING  From C:\Users\******\MiniConda3\envs\faceswap\lib\site-packages\keras\backend\tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n
10/28/2019 17:47:41 WARNING  From C:\Users\******\MiniConda3\envs\faceswap\lib\site-packages\keras\backend\tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n
10/28/2019 17:47:41 WARNING  From C:\Users\******\MiniConda3\envs\faceswap\lib\site-packages\keras\backend\tensorflow_backend.py:174: The name tf.get_default_session is deprecated. Please use tf.compat.v1.get_default_session instead.\n
10/28/2019 17:47:41 WARNING  From C:\Users\******\MiniConda3\envs\faceswap\lib\site-packages\keras\backend\tensorflow_backend.py:181: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.\n
10/28/2019 17:47:45 INFO     Loaded model from disk: 'D:\_FL\model'
10/28/2019 17:47:45 WARNING  From C:\Users\******\MiniConda3\envs\faceswap\lib\site-packages\keras\optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n
10/28/2019 17:47:45 INFO     Loading Mask from Box_Blend plugin...
10/28/2019 17:47:45 INFO     Loading Mask from Mask_Blend plugin...
10/28/2019 17:47:45 WARNING  Predicted selected, but the model was not trained with a mask. Switching to 'dfl_full'
10/28/2019 17:47:45 INFO     Loading Color from Avg_Color plugin...
10/28/2019 17:47:51 INFO     Outputting to: 'D:\_FL\OUTPUT\t_video6334477867065278594_converted.mp4'

10/28/2019 17:48:12 INFO     Muxing Audio...
10/28/2019 17:48:12 ERROR    There was a problem muxing audio. The output video has been created but you will need to mux audio yourself either with the EFFMpeg tool or an external application.
10/28/2019 17:48:12 INFO     -------------------------
10/28/2019 17:48:12 INFO     Images found:        727
10/28/2019 17:48:12 INFO     Faces detected:      144
10/28/2019 17:48:12 INFO     -------------------------
10/28/2019 17:48:12 INFO     Process Succesfully Completed. Shutting Down...
Process exited.

User avatar
bryanlyon
Site Admin
Posts: 74
Joined: Fri Jul 12, 2019 12:49 am
Answers: 9
Location: San Francisco
Has thanked: 3 times
Been thanked: 14 times
Contact:

Re: Convert not swapping faces for every frame of input video.

Post by bryanlyon » Mon Oct 28, 2019 11:22 pm

When you convert you need to use the full alignments file that you would have made from following the extract guide, not the subset file you made in the same guide.

You don't have to train with the same files you are converting with.

User avatar
winter0s
Posts: 2
Joined: Mon Oct 28, 2019 10:35 pm

Re: Convert not swapping faces for every frame of input video.

Post by winter0s » Tue Oct 29, 2019 12:59 am

Thanks! I will give that a shot.

Post Reply