Commit Graph

34 Commits

Author SHA1 Message Date
Felipe Daragon
59b6233882 Fix GPU detection on Apple Silicon 2025-05-14 02:05:34 +01:00
Felipe Daragon
a33d858916 GPU detection fix 2025-05-01 01:15:31 +01:00
Felipe Daragon
44fd699d1e Adds reface ratio option 2025-04-27 01:06:35 +01:00
Felipe Daragon
ba17490555 Add support for TIFF and improve output 2025-04-26 16:22:10 +01:00
Felipe Daragon
ef87bffe7d Use local temp dir during face extraction 2025-04-26 13:28:11 +01:00
Felipe Daragon
99c91d20b2 Supress CUDA path error in CPU mode 2025-04-16 21:14:48 +01:00
Felipe Daragon
a6c9eb3efe Add NeoRefacer code and updates 2025-04-10 22:08:59 +01:00
xaviviro
e7907c45b6 Revert "a hack to for onnxruntime to use cuda" 2023-06-27 20:39:24 +02:00
Phan Tuấn Anh
4d3bdf8ee1 comments to understand why torch 2023-06-26 14:28:38 +02:00
Phan Tuấn Anh
91144be667 import torch 2023-06-26 14:19:44 +02:00
Andras Fekete
19d039d145 Split out first_face vs multi-face calls 2023-06-19 13:21:11 -04:00
Andras Fekete
8e9035f712 5% Speed up 2023-06-19 10:50:58 -04:00
Andras Fekete
1388fc651f Remove already swapped face from list
This gives a speedup on my machine, but will have the side effect of preferring faces earlier in the list
2023-06-19 10:27:45 -04:00
Andras Fekete
a439a59296 Process only 1k frames at a time to minimize memory footprint 2023-06-16 09:04:25 -04:00
Xavi Vinaixa
5bc01cf583 Fixed bug with hw accel ffmpeg encoders by testing them beforehand #12. Added command arg —colab_performance to maximize performance in Google Colab. Colab notebook updated. 2023-06-07 17:22:27 +02:00
Xavi Vinaixa
ef9b64ad92 Fixed bug for videos without audio, reactivated HW acceleration for ffmpeg 2023-06-06 12:54:22 +02:00
Xavi Vinaixa
f2e4dd93f4 Disabled TensorrtExecutionProvider due to a bug, it will now run with CUDAExecutionProvider on NVIDIA until we resolve error #9 #10 2023-06-06 12:39:20 +02:00
Xavi Vinaixa
4e833ec3d1 Disabled HW accel for FFMPEG need impromevents 2023-06-06 12:36:47 +02:00
Xavi Vinaixa
d9884e41fe Disabled TensorrtExecutionProvider due to a bug, it will now run with CUDAExecutionProvider on NVIDIA until we resolve error #9 #10 2023-06-06 12:26:11 +02:00
Xavi Vinaixa
bbc10893b4 Commit message: Added support for hardware accelerators in ffmpeg. #5 2023-06-06 11:27:53 +02:00
Xavi Vinaixa
f5606ca4c4 Multithreading with TensorrtExecutionProvider #5. 2023-06-06 09:56:34 +02:00
Xavi Vinaixa
8008e2e766 TensorrtExecutionProvider 2023-06-06 08:14:28 +02:00
Xavi Vinaixa
a8d2941293 Incredible speed boost with TensorrtExecutionProvider #5. 2023-06-06 08:10:51 +02:00
Xavi Vinaixa
4a24df4755 Incredible speed boost with TensorrtExecutionProvider #5. 2023-06-06 08:05:58 +02:00
Xavi Vinaixa
7e628a3b83 Incredible speed boost with TensorrtExecutionProvider #5. 2023-06-06 08:04:33 +02:00
Xavi Vinaixa
cd1f6cd2df Added command line arguments for --share_gradio #7. Implemented multithreaded parallel processing and CoreML optimization, still pending CUDA optimization #5. 2023-06-06 07:51:39 +02:00
Xavi Vinaixa
b4146fa26d multithreading 2023-06-06 07:51:39 +02:00
Xavi Vinaixa
9df3a83d7e Fixed errors in tqdm implementation for progress tracking 2023-06-05 09:50:57 +02:00
Xavi Vinaixa
101691ca95 Fixed errors in tqdm implementation for progress tracking 2023-06-05 09:31:55 +02:00
Xavi Vinaixa
1c03fe3f1b Replaced ProgressBar with the standard tqdm library for enhanced user experience #3 2023-06-05 09:18:23 +02:00
Xavi Vinaixa
50ab0e1fbe Replaced ProgressBar with the standard tqdm library for enhanced user experience #3 2023-06-05 09:04:02 +02:00
Xavi Vinaixa
51a170a7eb Compatibility with Google Colab 2023-06-05 08:29:47 +02:00
Xavi Vinaixa
1269df361d Add rendering speed info and ETA to console output #3 2023-06-05 07:09:17 +02:00
Xavi
4bb1905efd First code commit 2023-06-03 08:04:06 +02:00