github/DeepFaceLab
1
0
Fork 0
mirror of https://github.com/iperov/DeepFaceLab.git synced 2025-08-21 22:13:20 -07:00
Code Issues Projects Releases Packages Wiki Activity Actions

Commit graph

  • dab857a445
         Update MergeMasked.py Joakim Fogelström 2020-06-14 23:39:01 +02:00
  • 2e650a7122 . Colombo 2020-06-12 10:45:57 +04:00
  • 7f11713730 fix Colombo 2020-06-11 21:29:39 +04:00
  • d99afebbdc new telegram group link Colombo 2020-06-11 17:39:36 +04:00
  • 82f405ed49 Trainer: fixed "Choose image for the preview history". Now you can switch between subpreviews using 'space' key. Fixed "Write preview history". Now it writes all subpreviews in separated folders https://i.imgur.com/IszifCJ.jpg also the last preview saved as _last.jpg before the first file https://i.imgur.com/Ls1AOK4.jpg thus you can easily check the changes with the first file in photo viewer Colombo 2020-06-10 10:20:13 +04:00
  • 5c315cab68 XSegEditor: added text label of total labeled images Changed frame line design Changed loading frame design Colombo 2020-06-09 22:04:15 +04:00
  • dc43f5a891 add second headswap example Colombo 2020-06-09 16:01:47 +04:00
  • 93a930fd52 upd TalosOfCrete 2020-06-08 17:58:16 -05:00
  • 81df55dfa1 Conciseness TalosOfCrete 2020-06-08 17:42:24 -05:00
  • 4f9aa3bf06 Fix TalosOfCrete 2020-06-08 14:05:18 -05:00
  • 082394193a Fixes TalosOfCrete 2020-06-08 13:22:53 -05:00
  • 13b53a6e14 Optional 2 TalosOfCrete 2020-06-08 12:53:56 -05:00
  • 00a7a2357a Made separable optional TalosOfCrete 2020-06-08 12:53:29 -05:00
  • c2b5f80e3f upd readme download link Colombo 2020-06-08 15:03:40 +04:00
  • 1d3d417aee upd readme Colombo 2020-06-08 14:53:19 +04:00
  • af98407f06 SAEHD: lr_dropout now can be ‘n’, ‘y’, ‘cpu’. ‘n’ and ’y’ are the same as before. ‘cpu’ mean enabled on CPU. This allows not to use extra VRAM, sacrificing 20% time of iteration. Colombo 2020-06-08 14:18:33 +04:00
  • 4acaf08ebf fix Colombo 2020-06-08 14:07:47 +04:00
  • 7b5b13bbc7 Misc. fixes TalosOfCrete 2020-06-07 19:49:46 -05:00
  • a099350eb6 fix TalosOfCrete 2020-06-07 17:47:32 -05:00
  • 6a993b930c Merge branch 'master' of https://github.com/TalosOfCrete/DeepFaceLab TalosOfCrete 2020-06-07 17:44:53 -05:00
  • 4e46708303 Added depthwise separable convolutions TalosOfCrete 2020-06-07 17:44:48 -05:00
  • cfd7803e0d leras : DepthwiseConv2D Colombo 2020-06-07 20:59:12 +04:00
  • 85ede1a7e7 . Colombo 2020-06-07 12:26:23 +04:00
  • ddf7363eda remove unused code Colombo 2020-06-06 20:02:21 +04:00
  • 235315dd70 fix Colombo 2020-06-06 19:52:07 +04:00
  • 1e003170bd reduced chance of "The paging file is too small for this operation to complete." Colombo 2020-06-06 19:38:11 +04:00
  • 61f93063b2 fix Colombo 2020-06-04 16:44:29 +04:00
  • addc96fe3e fix Colombo 2020-06-04 16:42:36 +04:00
  • 4ed320a86b
         fix nan in multiply (#769) Ilya Sergeev 2020-06-04 15:39:46 +03:00
  • 0f0d8dd407 fix nan in multiply Ilya Sergeev 2020-06-04 15:30:54 +03:00
  • 65bc9273e6 upd readme Colombo 2020-06-04 15:38:20 +04:00
  • c94c9106be upd readme Colombo 2020-06-04 14:44:47 +04:00
  • 903657ff91 upd readme Colombo 2020-06-04 14:39:50 +04:00
  • f91a604c6d Manual extractor: now you can specify face rectangle manually using ‘R Mouse button’. It is useful for small, blurry, undetectable faces, animal faces. https://i.imgur.com/8kmVgg8.jpg Warning: such frames must be used only with XSeg workflow ! Landmarks cannot be placed on the face precisely, and they are actually used for positioning the red frame. Try to keep the red frame the same as the adjacent frames. Colombo 2020-06-02 12:31:24 +04:00
  • fbe70767b2
         upload Nicholas Scott 2020-06-02 02:43:46 -05:00
  • a36359f480 Fix TalosOfCrete 2020-06-01 16:39:04 -05:00
  • a383f157fe Removed Separable Conv2D (for now); added experimental extractors TalosOfCrete 2020-06-01 16:36:51 -05:00
  • e77865ce18 XSegEditor: fix Colombo 2020-06-01 13:44:29 +04:00
  • 6f25ebbc9e XSegEditor: added button "view XSeg mask overlay face" Colombo 2020-05-31 18:00:39 +04:00
  • de8288b098 Replaced all Conv2D with DepthwiseSeparableConv2D; reverted to old archi design TalosOfCrete 2020-05-25 17:51:25 -05:00
  • c6652ac006 add arxiv paper Colombo 2020-05-24 15:10:40 +04:00
  • 130d72cb8b upd Colombo 2020-05-22 22:44:00 +04:00
  • 0a1e33b6fc Update DeepFakeArchi.py TalosOfCrete 2020-05-21 23:09:41 -05:00
  • 30b4c4c679 Temporary fix TalosOfCrete 2020-05-21 21:00:09 -05:00
  • c29982222b Update DeepFakeArchi.py TalosOfCrete 2020-05-21 18:52:09 -05:00
  • 72e2f771a2 upd readme Colombo 2020-05-20 09:58:55 +04:00
  • bd3806b320
         Update README.md Daniel Fernandez 2020-05-16 00:09:45 -05:00
  • 22f7a02244
         Update README.md Daniel Fernandez 2020-05-16 00:08:53 -05:00
  • 27ad85f590
         Update README.md Daniel Fernandez 2020-05-16 00:07:44 -05:00
  • ecf7eb083a
         Added SeparableConv2d.py Nicholas Scott 2020-05-15 21:58:36 -05:00
  • acfc78bfd5 Initial fork commit TalosOfCrete 2020-05-15 21:52:01 -05:00
  • 68e591cca4
         Merge pull request #3 from iperov/master Krzysztof Brzęczyszczykiewicz 2020-05-14 19:39:48 +07:00
  • 0c18b1f011 changed uhd arhi: you have to retrain uhd arhis from scratch. Colombo 2020-05-06 19:21:23 +04:00
  • dcf380345b fix Colombo 2020-05-02 14:04:32 +04:00
  • a03b4506db Don't pass alpha channel of image. Alpha channel may contained if no face was detected in original file and was copied with alpha channel. Alexander Court 2020-05-01 18:40:29 +02:00
  • f935bf0465 . Colombo 2020-04-28 19:04:44 +04:00
  • 7b49220ee9 fix Colombo 2020-04-20 18:40:12 +04:00
  • 4fe45adc7c . Colombo 2020-04-20 10:24:51 +04:00
  • 3a9e851339 fix bug Colombo 2020-04-18 18:29:55 +04:00
  • 08d0521298
         Merge pull request #2 from iperov/master Krzysztof Brzęczyszczykiewicz 2020-04-17 07:40:17 +07:00
  • 69b9732991
         Update MergeMasked.py Felipe Ducau 2020-04-15 22:28:31 -03:00
  • d63a682204
         Update InteractiveMergerSubprocessor.py Felipe Ducau 2020-04-15 22:25:14 -03:00
  • b5b93bbcca
         Adding parallel frame info Felipe Ducau 2020-04-15 21:58:53 -03:00
  • a3c91271d0 upd README.md : google drive is replaced by mega Colombo 2020-04-16 00:08:42 +04:00
  • baff59e421 . Colombo 2020-04-15 22:58:59 +04:00
  • ea607edfc9 XSegEditor: added view lock at the center by holding shift. Colombo 2020-04-15 22:12:54 +04:00
  • 6bbc607312 color transfer sot : optimization Colombo 2020-04-15 18:13:03 +04:00
  • f8580928ed SampleLoader: fix error Colombo 2020-04-15 16:55:51 +04:00
  • c42b1b2124 . Colombo 2020-04-15 10:44:12 +04:00
  • cd8919d95b Merger: optimizations Colombo 2020-04-14 10:25:54 +04:00
  • e549624eeb upd README : remove 'github releases' Colombo 2020-04-13 22:33:17 +04:00
  • 2cc0e64572 Merger: added mask mode ‘learned-prd + learned-dst’ – produces largest area of both dst and predicted masks Colombo 2020-04-13 22:04:24 +04:00
  • d1af3b51cd . Colombo 2020-04-13 21:37:55 +04:00
  • 835f3adb8e XSegEditor : polygon is now transparent while editing Colombo 2020-04-13 21:37:45 +04:00
  • 4a3b94832f upd Colombo 2020-04-13 21:02:51 +04:00
  • 95847637e5 upd Colombo 2020-04-13 21:02:03 +04:00
  • 8ddc7954da upd README.md : added mini tutorial Colombo 2020-04-13 20:57:51 +04:00
  • 80d4eef1fe . Colombo 2020-04-11 23:38:35 +04:00
  • 23961879b4 Merger: color transfer ‘sot-m’ : reduced color flickering consuming x5 more time to process Colombo 2020-04-11 23:33:55 +04:00
  • 25324267d6 Merger: color transfer ‘sot-m’ : reduced color flickering Colombo 2020-04-11 21:33:47 +04:00
  • f72df02950 upd DF.wf.288res.384.92.72.22 Colombo 2020-04-08 18:20:57 +04:00
  • 935d940ace fix Colombo 2020-04-07 22:06:32 +04:00
  • d8a792232c upd readme Colombo 2020-04-07 10:25:11 +04:00
  • 5b4d023712 added 5.XSeg) data_dst/data_src mask for XSeg trainer - remove.bat removes labeled xseg polygons from the extracted frames Colombo 2020-04-06 21:45:46 +04:00
  • b6b3936bcd update readme Colombo 2020-04-06 20:15:46 +04:00
  • f821ab350f fix Colombo 2020-04-06 19:06:37 +04:00
  • afdb1ef85d fix for 16+ CPU Colombo 2020-04-06 18:42:38 +04:00
  • 453237bfd7 upd README.md Colombo 2020-04-06 17:55:33 +04:00
  • 215277b376 XSegEditor: fix Colombo 2020-04-05 14:14:56 +04:00
  • 33b0aadb4e Decreased amount of RAM used by Sample Generator. Colombo 2020-04-05 13:52:32 +04:00
  • 5345c56459
         Do not save unfinished polygons rocketsvm 2020-04-05 10:30:30 +05:00
  • 8e9e346c9d Extractor: now face_type can be choosed in console dialog Colombo 2020-04-04 10:08:23 +04:00
  • 8aa87c3080 upd README.md Colombo 2020-04-04 09:41:10 +04:00
  • 4db11aa133 upd README.md Colombo 2020-04-04 09:40:18 +04:00
  • 2b7364005d Added new face type : head Now you can replace the head. Example: https://www.youtube.com/watch?v=xr5FHd0AdlQ Requirements: Post processing skill in Adobe After Effects or Davinci Resolve. Usage: 1) Find suitable dst footage with the monotonous background behind head 2) Use “extract head” script 3) Gather rich src headset from only one scene (same color and haircut) 4) Mask whole head for src and dst using XSeg editor 5) Train XSeg 6) Apply trained XSeg mask for src and dst headsets 7) Train SAEHD using ‘head’ face_type as regular deepfake model with DF archi. You can use pretrained model for head. Minimum recommended resolution for head is 224. 8) Extract multiple tracks, using Merger: a. Raw-rgb b. XSeg-prd mask c. XSeg-dst mask 9) Using AAE or DavinciResolve, do: a. Hide source head using XSeg-prd mask: content-aware-fill, clone-stamp, background retraction, or other technique b. Overlay new head using XSeg-dst mask Colombo 2020-04-04 09:28:06 +04:00
  • 383d4d3736 fix Colombo 2020-04-03 16:24:34 +04:00
  • 0fb912e91f Trainer: added --silent-start cmd option Colombo 2020-04-02 13:05:04 +04:00
  • 2fe86faf01 fix Win10 input dialog Colombo 2020-04-02 10:27:39 +04:00
  • dbb0988927 upd README.md Colombo 2020-04-01 22:17:59 +04:00
  • 652ef75099 upd README.md Colombo 2020-04-01 16:08:06 +04:00