opt: Namespace(A2L_GMM_ndim=75, APC_frame_history=0, APC_hidden_size=512, APC_residual=False, APC_rnn_layers=3, FPS=60, LSTM_dropout=0, LSTM_hidden_size=256, LSTM_layers=3, LSTM_output_size=80, LSTM_residual=False, LSTM_sequence_length=60, audioRF_future=0, audioRF_history=60, audio_encoder='APC', batch_size=32, checkpoints_dir='./checkpoints/', dataroot='default_path', dataset_mode='audiovisual', dataset_names='default_name', eval=False, feature_decoder='LSTM', feature_dtype='pts3d', frame_future=18, frame_jump_stride=4, gpu_ids='0', ispts_norm=1, load_epoch='500', loss='L2', max_dataset_size=inf, model='audio2feature', name='Audio2Feature', num_threads=0, only_mouth=1, phase='test', predict_length=1, sample_rate=16000, sequence_length=240, serial_batches=False, suffix='', task='Audio2Feature', time_frame_length=1, use_delta_pts=1, verbose=False) ----------------- Options --------------- A2L_GMM_ndim: 75 APC_frame_history: 0 APC_hidden_size: 512 APC_residual: False APC_rnn_layers: 3 FPS: 60 LSTM_dropout: 0 LSTM_hidden_size: 256 LSTM_layers: 3 LSTM_output_size: 80 LSTM_residual: False LSTM_sequence_length: 60 audioRF_future: 0 audioRF_history: 60 audio_encoder: APC batch_size: 32 checkpoints_dir: ./checkpoints/ dataroot: default_path dataset_mode: audiovisual dataset_names: default_name eval: False feature_decoder: LSTM feature_dtype: pts3d frame_future: 18 frame_jump_stride: 4 gpu_ids: 0 isTrain: False [default: None] ispts_norm: 1 load_epoch: 500 loss: L2 max_dataset_size: inf model: audio2feature name: Audio2Feature num_threads: 0 only_mouth: 1 phase: test predict_length: 1 sample_rate: 16000 sequence_length: 240 serial_batches: False suffix: task: Audio2Feature time_frame_length: 1 use_delta_pts: 1 verbose: False ----------------- End ------------------- ----------------- Options --------------- A2H_GMM_ncenter: 1 A2H_GMM_ndim: 12 A2H_GMM_sigma_min: 0.03 A2H_wavenet_cond: True A2H_wavenet_cond_channels: 512 A2H_wavenet_dilation_channels: 128 A2H_wavenet_input_channels: 12 A2H_wavenet_kernel_size: 2 A2H_wavenet_residual_blocks: 2 A2H_wavenet_residual_channels: 128 A2H_wavenet_residual_layers: 7 A2H_wavenet_skip_channels: 256 A2H_wavenet_use_bias: True APC_frame_history: 60 APC_hidden_size: 512 APC_residual: False APC_rnn_layers: 3 FPS: 60 audioRF_future: 0 audioRF_history: 60 audio_encoder: APC audio_windows: 2 audiofeature_input_channels: 80 batch_size: 32 checkpoints_dir: ./checkpoints/ dataroot: path dataset_mode: audiovisual dataset_names: name eval: False feature_decoder: WaveNet frame_future: 15 frame_jump_stride: 1 gpu_ids: 0 isTrain: False [default: None] load_epoch: 500 loss: GMM max_dataset_size: inf model: audio2headpose name: Audio2Headpose num_threads: 0 phase: test predict_length: 5 sample_rate: 16000 sequence_length: 240 serial_batches: False suffix: task: Audio2Headpose time_frame_length: 1 verbose: False ----------------- End ------------------- ------------ Options ------------- batch_size: 1 checkpoints_dir: ./checkpoints/ dataroot: ./data/ dataset_mode: face dataset_names: ['name'] debug: False display_id: 0 display_winsize: 512 fineSize: 512 fp16: 0 gpu_ids: [0] input_nc: 1 isH5: 1 isMask: 0 isTrain: False loadSize: 512 load_pretrain: local_rank: 0 max_dataset_size: inf model: feature2face n_blocks_E: 3 n_downsample_E: 3 n_downsample_G: 8 name: TestRender ngf: 64 ngf_E: 16 no_flip: 1 num_threads: 0 output_nc: 3 phase: test resize_or_crop: scaleWidth serial_batches: False suffix: .jpg task: Feature2Face test_dataset_names: ['name'] tf_log: True verbose: False -------------- End ---------------- ---------- Loading Model: APC------------- ---------- Loading Model: Audio2Feature ------------- initialize network with normal model [Audio2FeatureModel] was created loading the model from ./data/May/checkpoints/Audio2Feature.pkl ---------- Networks initialized ------------- [Network Audio2Feature] Total number of parameters : 3.064 M ----------------------------------------------- ---------- Loading Model: Audio2Headpose ------------- initialize network with normal model [Audio2HeadposeModel] was created loading the model from ./data/May/checkpoints/Audio2Headpose.pkl ---------- Networks initialized ------------- [Network Audio2Headpose] Total number of parameters : 4.267 M ----------------------------------------------- ---------- Loading Model: Feature2Face ------------- dataset [FaceDataset] was created ---------- Generator networks initialized ------------- ------------------------------------------------------- initialize network with normal model [Feature2FaceModel] was created loading the model from ./data/May/checkpoints/Feature2Face.pkl ---------- Networks initialized ------------- [Network Feature2Face_G] Total number of parameters : 121.790 M ----------------------------------------------- Processing audio: 00083 ... 1. Computing APC features... 2. Manifold projection... LLE projection: 0%| | 0/1374 [00:00