jones/facefusion
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Next (#477)

Browse Source 
* Add real_hatgan_x4 model

* Mark it as NEXT

* Force download to be executed and exit

* Fix frame per second interpolation

* 5 to 68 landmark (#456)

* changes

* changes

* Adjust model url

* Cleanup 5 to 68 landmark convertion

* Move everything to face analyser

* Introduce matrix only face helper

* Revert facefusion.ini

* Adjust limit due false positive analysis

* changes (#457)

* Use pixel format yuv422p to merge video

* Fix some code

* Minor cleanup

* Add gpen_bfr_1024 and gpen_bfr_2048

* Revert it back to yuv420p due compatibility issues

* Add debug back to ffmpeg

* Add debug back to ffmpeg

* Migrate to conda (#461)

* Migrate from venv to conda

* Migrate from venv to conda

* Message when conda is not activated

* Use release for every slider (#463)

* Use release event handler for every slider

* Move more sliders to release handler

* Move more sliders to release handler

* Add get_ui_components() to simplify code

* Revert some changes on frame slider

* Add the first iteration of a frame colorizer

* Support for the DDColor model

* Improve model file handling

* Improve model file handling part2

* Remove deoldify

* Remove deoldify

* Voice separator (#468)

* changes

* changes

* changes

* changes

* changes

* changes

* Rename audio extractor to voice extractor

* Cosmetic changes

* Cosmetic changes

* Fix fps lowering and boosting

* Fix fps lowering and boosting

* Fix fps lowering and boosting

* Some refactoring for audio.py and some astype() here and there (#470)

* Some refactoring for audio.py and some astype() here and there

* Fix lint

* Spacing

* Add mp3 to benchmark suite for lip syncer testing

* Improve naming

* Adjust chunk size

* Use higher quality

* Revert "Use higher quality"

This reverts commit d32f28757251ecc0f48214073adf54f3631b1289.

* Improve naming in ffmpeg.py

* Simplify code

* Better fps calculation

* Fix naming here and there

* Add back real esrgan x2

* Remove trailing comma

* Update wording and README

* Use semaphore to prevent frame colorizer memory issues

* Revert "Remove deoldify"

This reverts commit bd8034cbc71fe701f78dddec3057dc98593b2162.

* Remove unused type from frame colorizer

* Adjust naming

* Add missing clear of model initializer

* Change nvenc preset mappping to support old FFMPEG 4

* Update onnxruntime to 1.17.1

* Fix lint

* Prepare 2.5.0

* Fix Gradio overrides

* Add Deoldify Artistic back

* Feat/audio refactoring (#476)

* Improve audio naming and variables

* Improve audio naming and variables

* Refactor voice extractor like crazy

* Refactor voice extractor like crazy

* Remove spaces

* Update the usage

---------

Co-authored-by: Harisreedhar <46858047+harisreedhar@users.noreply.github.com>
This commit is contained in:
Henry Ruhs
2024-04-09 15:40:55 +02:00
committed by GitHub
parent 6e67d7bff6
commit 4ccf4c24c7
45 changed files with 1007 additions and 405 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
facefusion/face_helper.py
View File

@@ -43,16 +43,21 @@ WARP_TEMPLATES : WarpTemplateSet =\
}
def warp_face_by_face_landmark_5(temp_vision_frame : VisionFrame, face_landmark_5 : FaceLandmark5, warp_template : WarpTemplate, crop_size : Size) -> Tuple[VisionFrame, Matrix]:
def estimate_matrix_by_face_landmark_5(face_landmark_5 : FaceLandmark5, warp_template : WarpTemplate, crop_size : Size) -> Matrix:
normed_warp_template = WARP_TEMPLATES.get(warp_template) * crop_size
affine_matrix = cv2.estimateAffinePartial2D(face_landmark_5, normed_warp_template, method = cv2.RANSAC, ransacReprojThreshold = 100)[0]
return affine_matrix
def warp_face_by_face_landmark_5(temp_vision_frame : VisionFrame, face_landmark_5 : FaceLandmark5, warp_template : WarpTemplate, crop_size : Size) -> Tuple[VisionFrame, Matrix]:
affine_matrix = estimate_matrix_by_face_landmark_5(face_landmark_5, warp_template, crop_size)
crop_vision_frame = cv2.warpAffine(temp_vision_frame, affine_matrix, crop_size, borderMode = cv2.BORDER_REPLICATE, flags = cv2.INTER_AREA)
return crop_vision_frame, affine_matrix
def warp_face_by_bounding_box(temp_vision_frame : VisionFrame, bounding_box : BoundingBox, crop_size : Size) -> Tuple[VisionFrame, Matrix]:
source_points = numpy.array([ [ bounding_box[0], bounding_box[1] ], [bounding_box[2], bounding_box[1] ], [ bounding_box[0], bounding_box[3] ] ], dtype = numpy.float32)
target_points = numpy.array([ [ 0, 0 ], [ crop_size[0], 0 ], [ 0, crop_size[1] ] ], dtype = numpy.float32)
source_points = numpy.array([ [ bounding_box[0], bounding_box[1] ], [bounding_box[2], bounding_box[1] ], [ bounding_box[0], bounding_box[3] ] ]).astype(numpy.float32)
target_points = numpy.array([ [ 0, 0 ], [ crop_size[0], 0 ], [ 0, crop_size[1] ] ]).astype(numpy.float32)
affine_matrix = cv2.getAffineTransform(source_points, target_points)
if bounding_box[2] - bounding_box[0] > crop_size[0] or bounding_box[3] - bounding_box[1] > crop_size[1]:
interpolation_method = cv2.INTER_AREA
@@ -112,14 +117,14 @@ def distance_to_face_landmark_5(points : numpy.ndarray[Any, Any], distance : num
return face_landmark_5
def convert_face_landmark_68_to_5(landmark_68 : FaceLandmark68) -> FaceLandmark5:
def convert_face_landmark_68_to_5(face_landmark_68 : FaceLandmark68) -> FaceLandmark5:
face_landmark_5 = numpy.array(
[
numpy.mean(landmark_68[36:42], axis = 0),
numpy.mean(landmark_68[42:48], axis = 0),
landmark_68[30],
landmark_68[48],
landmark_68[54]
numpy.mean(face_landmark_68[36:42], axis = 0),
numpy.mean(face_landmark_68[42:48], axis = 0),
face_landmark_68[30],
face_landmark_68[48],
face_landmark_68[54]
])
return face_landmark_5