Skip to content

Industry leading face manipulation platform

patch-3.7.1 (#1178)

Henry Ruhs3f81a8a78454
diff --git a/facefusion/metadata.py b/facefusion/metadata.py
index 4cc03d69..13086cc6 100644
--- a/facefusion/metadata.py
+++ b/facefusion/metadata.py
@@ -4,7 +4,7 @@ METADATA =\
{
'name': 'FaceFusion',
'description': 'Industry leading face manipulation platform',
- 'version': '3.7.0',
+ 'version': '3.7.1',
'license': 'OpenRAIL-AS',
'author': 'Henry Ruhs',
'url': 'https://facefusion.io'
diff --git a/facefusion/program.py b/facefusion/program.py
index 540b2862..ae2d9667 100755
--- a/facefusion/program.py
+++ b/facefusion/program.py
@@ -177,7 +177,7 @@ def create_frame_extraction_program() -> ArgumentParser:
def create_frame_distribution_program() -> ArgumentParser:
program = ArgumentParser(add_help = False)
group_frame_distribution = program.add_argument_group('frame distribution')
- group_frame_distribution.add_argument('--target-frame-amount', help = translator.get('help.target_frame_amount'), type = int, default = config.get_int_value('frame_distribution', 'target_frame_amount', '5'), choices = facefusion.choices.target_frame_amount_range, metavar = create_int_metavar(facefusion.choices.target_frame_amount_range))
+ group_frame_distribution.add_argument('--target-frame-amount', help = translator.get('help.target_frame_amount'), type = int, default = config.get_int_value('frame_distribution', 'target_frame_amount', '2'), choices = facefusion.choices.target_frame_amount_range, metavar = create_int_metavar(facefusion.choices.target_frame_amount_range))
job_store.register_step_keys([ 'target_frame_amount' ])
return program
diff --git a/facefusion/video_manager.py b/facefusion/video_manager.py
index 8923afb6..b63203cc 100644
--- a/facefusion/video_manager.py
+++ b/facefusion/video_manager.py
@@ -19,6 +19,12 @@ def get_video_capture(video_path : str) -> cv2.VideoCapture:
return VIDEO_POOL_SET.get('capture').get(video_path)
+def conditional_set_video_frame_position(video_capture : cv2.VideoCapture, frame_position : int) -> bool:
+ if not video_capture.get(cv2.CAP_PROP_POS_FRAMES) == frame_position:
+ return video_capture.set(cv2.CAP_PROP_POS_FRAMES, frame_position)
+ return True
+
+
def get_video_writer(video_path : str) -> cv2.VideoWriter:
if video_path not in VIDEO_POOL_SET.get('writer'):
video_writer = cv2.VideoWriter()
diff --git a/facefusion/vision.py b/facefusion/vision.py
index 50b2a8d9..79ddb536 100644
--- a/facefusion/vision.py
+++ b/facefusion/vision.py
@@ -10,7 +10,7 @@ from facefusion.common_helper import is_windows
from facefusion.filesystem import get_file_extension, is_image, is_video
from facefusion.thread_helper import thread_lock, thread_semaphore
from facefusion.types import ColorMode, Duration, Fps, Mask, Orientation, Resolution, Scale, VisionFrame
-from facefusion.video_manager import get_video_capture
+from facefusion.video_manager import conditional_set_video_frame_position, get_video_capture
def read_static_images(image_paths : List[str], color_mode : ColorMode = 'rgb') -> List[VisionFrame]:
@@ -80,11 +80,11 @@ def read_video_frame(video_path : str, frame_number : int = 0) -> Optional[Visio
video_capture = get_video_capture(video_path)
if video_capture and video_capture.isOpened():
- frame_total = video_capture.get(cv2.CAP_PROP_FRAME_COUNT)
- frame_position = min(frame_total, frame_number)
+ video_frame_total = int(video_capture.get(cv2.CAP_PROP_FRAME_COUNT))
+ video_frame_position = min(video_frame_total, frame_number)
with thread_semaphore():
- video_capture.set(cv2.CAP_PROP_POS_FRAMES, frame_position)
+ conditional_set_video_frame_position(video_capture, video_frame_position)
has_vision_frame, vision_frame = video_capture.read()
if has_vision_frame:
@@ -105,13 +105,13 @@ def read_video_chunk(video_path : str, chunk_number : int, chunk_size : int) ->
video_capture = get_video_capture(video_path)
if video_capture and video_capture.isOpened():
- frame_total = int(video_capture.get(cv2.CAP_PROP_FRAME_COUNT))
- frame_position = chunk_number * chunk_size
+ video_frame_total = int(video_capture.get(cv2.CAP_PROP_FRAME_COUNT))
+ video_frame_position = chunk_number * chunk_size
with thread_semaphore():
- video_capture.set(cv2.CAP_PROP_POS_FRAMES, frame_position)
+ conditional_set_video_frame_position(video_capture, video_frame_position)
- for frame_number in range(frame_position, min(frame_position + chunk_size, frame_total)):
+ for frame_number in range(video_frame_position, min(video_frame_position + chunk_size, video_frame_total)):
has_vision_frame, vision_frame = video_capture.read()
if has_vision_frame:
@@ -120,9 +120,9 @@ def read_video_chunk(video_path : str, chunk_number : int, chunk_size : int) ->
return video_frame_chunk
-def select_video_frames(video_path : str, frame_number : int = 0, frame_offset : int = 5) -> List[VisionFrame]:
+def select_video_frames(video_path : str, frame_number : int = 0, frame_offset : int = 2) -> List[VisionFrame]:
vision_frames = []
- chunk_size = (frame_offset * 2 + 1) * 4
+ chunk_size = frame_offset * 2 + 1
if is_video(video_path):
with thread_lock():
diff --git a/facefusion/workflows/image_to_image.py b/facefusion/workflows/image_to_image.py
index 766a9dd9..9af47856 100644
--- a/facefusion/workflows/image_to_image.py
+++ b/facefusion/workflows/image_to_image.py
@@ -63,10 +63,11 @@ def prepare_image() -> ErrorCode:
def process_image() -> ErrorCode:
temp_image_path = get_temp_file_path(state_manager.get_item('target_path'))
- reference_vision_frame = read_static_image(temp_image_path)
+ reference_vision_frame = read_static_image(state_manager.get_item('target_path'))
source_vision_frames = read_static_images(state_manager.get_item('source_paths'))
source_audio_frame = create_empty_audio_frame()
source_voice_frame = create_empty_audio_frame()
+ target_vision_frame = read_static_image(state_manager.get_item('target_path'))
temp_vision_frame = read_static_image(temp_image_path, 'rgba')
temp_vision_mask = extract_vision_mask(temp_vision_frame)
@@ -79,7 +80,7 @@ def process_image() -> ErrorCode:
'source_vision_frames': source_vision_frames,
'source_audio_frame': source_audio_frame,
'source_voice_frame': source_voice_frame,
- 'target_vision_frames': [ temp_vision_frame[:, :, :3] ],
+ 'target_vision_frames': [ target_vision_frame ],
'temp_vision_frame': temp_vision_frame[:, :, :3],
'temp_vision_mask': temp_vision_mask
})