Merge branch 'main' of https://github.com/hacksider/Deep-Live-Cam

fix for GfpGAN and inswapper model path retrieval bug

.github/ISSUE_TEMPLATE/bug_report.md

@@ -1,38 +1,26 @@
----
-name: Bug report
-about: Create a report to help us improve
-title: ''
-labels: ''
-assignees: ''
-
----
+***[Remove this]The issue would be closed without notice and be considered spam if the template is not followed.***
 
 **Describe the bug**
 A clear and concise description of what the bug is.
 
-**To Reproduce**
-Steps to reproduce the behavior:
-1. Go to '...'
-2. Click on '....'
-3. Scroll down to '....'
-4. See error
-
-**Expected behavior**
-A clear and concise description of what you expected to happen.
-
 **Screenshots**
 If applicable, add screenshots to help explain your problem.
 
-**Desktop (please complete the following information):**
- - OS: [e.g. iOS]
- - Browser [e.g. chrome, safari]
- - Version [e.g. 22]
+**Error Message**
 
-**Smartphone (please complete the following information):**
- - Device: [e.g. iPhone6]
- - OS: [e.g. iOS8.1]
- - Browser [e.g. stock browser, safari]
+`<The error message in terminal>`
+
+**Desktop (please complete the following information):**
+ - OS: [e.g. Windows]
  - Version [e.g. 22]
+ - GPU
+ - CPU
 
 **Additional context**
 Add any other context about the problem here.
+
+**Confirmation (Mandatory)**
+- [ ] I have followed the template
+- [ ] This is not a query about how to increase performance
+- [ ] I have checked the issues page, and this is not a duplicate
+

.gitignore

@@ -24,3 +24,4 @@ models/GFPGANv1.4.pth
 models/DMDNet.pth
 faceswap/
 .vscode/
+switch_states.json

README.md

@@ -4,11 +4,16 @@
   Real-time face swap and video deepfake with a single click and only a single image.
 </p>
 
+<p align="center">
+<a href="https://trendshift.io/repositories/11395" target="_blank"><img src="https://trendshift.io/api/badge/repositories/11395" alt="hacksider%2FDeep-Live-Cam | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</p>
+
 <p align="center">
   <img src="media/demo.gif" alt="Demo GIF">
   <img src="media/avgpcperformancedemo.gif" alt="Performance Demo GIF">
 </p>
 
+
 ## Disclaimer
 
 This software is intended as a productive contribution to the AI-generated media industry. It aims to assist artists with tasks like animating custom characters or using them as models for clothing, etc.
@@ -395,9 +400,9 @@ This is an open-source project developed in our free time. Updates may be delaye
 - [GosuDRM](https://github.com/GosuDRM) : for open version of roop
 - [pereiraroland26](https://github.com/pereiraroland26) : Multiple faces support
 - [vic4key](https://github.com/vic4key) : For supporting/contributing on this project
-- [KRSHH](https://github.com/KRSHH) : For updating the UI
+- [KRSHH](https://github.com/KRSHH) : For his contributions
 - and [all developers](https://github.com/hacksider/Deep-Live-Cam/graphs/contributors) behind libraries used in this project.
-- Foot Note: [This is originally roop-cam, see the full history of the code here.](https://github.com/hacksider/roop-cam) Please be informed that the base author of the code is [s0md3v](https://github.com/s0md3v/roop)
+- Foot Note: Please be informed that the base author of the code is [s0md3v](https://github.com/s0md3v/roop)
 
 ## Contributions
 ![Alt](https://repobeats.axiom.co/api/embed/fec8e29c45dfdb9c5916f3a7830e1249308d20e1.svg "Repobeats analytics image")

models/instructions.txt

@@ -1,4 +1,4 @@
 just put the models in this folder -
 
 https://huggingface.co/hacksider/deep-live-cam/resolve/main/inswapper_128_fp16.onnx?download=true
-https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/GFPGANv1.4.pth
+https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/GFPGANv1.4.pth

modules/globals.py

@@ -16,31 +16,28 @@ source_path = None
 target_path = None
 output_path = None
 frame_processors: List[str] = []
-keep_fps = True  # Initialize with default value
-keep_audio = True  # Initialize with default value
-keep_frames = False  # Initialize with default value
-many_faces = False  # Initialize with default value
-map_faces = False  # Initialize with default value
-color_correction = False  # Initialize with default value
-nsfw_filter = False  # Initialize with default value
+keep_fps = True
+keep_audio = True
+keep_frames = False
+many_faces = False
+map_faces = False
+color_correction = False  # New global variable for color correction toggle
+nsfw_filter = False
 video_encoder = None
 video_quality = None
-live_mirror = False  # Initialize with default value
-live_resizable = False  # Initialize with default value
+live_mirror = False
+live_resizable = False
 max_memory = None
 execution_providers: List[str] = []
 execution_threads = None
 headless = None
 log_level = "error"
-fp_ui: Dict[str, bool] = {"face_enhancer": False}  # Initialize with default value
+fp_ui: Dict[str, bool] = {"face_enhancer": False}
 camera_input_combobox = None
 webcam_preview_running = False
-show_fps = False  # Initialize with default value
+show_fps = False
 mouth_mask = False
 show_mouth_mask_box = False
-mask_down_size = 0.5
-mask_size = 1.0
 mask_feather_ratio = 8
-opacity_switch = False
-face_opacity = 100
-selected_camera = None
+mask_down_size = 0.50
+mask_size = 1

modules/metadata.py

@@ -1,3 +1,3 @@
 name = 'Deep Live Cam'
-version = '1.6.0'
+version = '1.7.0'
 edition = 'Portable'

modules/processors/frame/face_enhancer.py

@@ -9,23 +9,36 @@ import modules.processors.frame.core
 from modules.core import update_status
 from modules.face_analyser import get_one_face
 from modules.typing import Frame, Face
-from modules.utilities import conditional_download, resolve_relative_path, is_image, is_video
+from modules.utilities import (
+    conditional_download,
+    is_image,
+    is_video,
+)
 
 FACE_ENHANCER = None
 THREAD_SEMAPHORE = threading.Semaphore()
 THREAD_LOCK = threading.Lock()
-NAME = 'DLC.FACE-ENHANCER'
+NAME = "DLC.FACE-ENHANCER"
 
+abs_dir = os.path.dirname(os.path.abspath(__file__))
+models_dir = os.path.join(os.path.dirname(os.path.dirname(os.path.dirname(abs_dir))), 'models')
 
 def pre_check() -> bool:
-    download_directory_path = resolve_relative_path('..\models')
-    conditional_download(download_directory_path, ['https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/GFPGANv1.4.pth'])
+    download_directory_path = models_dir
+    conditional_download(
+        download_directory_path,
+        [
+            "https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/GFPGANv1.4.pth"
+        ],
+    )
     return True
 
 
 def pre_start() -> bool:
-    if not is_image(modules.globals.target_path) and not is_video(modules.globals.target_path):
-        update_status('Select an image or video for target path.', NAME)
+    if not is_image(modules.globals.target_path) and not is_video(
+        modules.globals.target_path
+    ):
+        update_status("Select an image or video for target path.", NAME)
         return False
     return True
 
@@ -35,21 +48,14 @@ def get_face_enhancer() -> Any:
 
     with THREAD_LOCK:
         if FACE_ENHANCER is None:
-            if os.name == 'nt':
-                model_path = resolve_relative_path('..\models\GFPGANv1.4.pth')
-                # todo: set models path https://github.com/TencentARC/GFPGAN/issues/399
-            else:
-                model_path = resolve_relative_path('../models/GFPGANv1.4.pth')
-            FACE_ENHANCER = gfpgan.GFPGANer(model_path=model_path, upscale=1) # type: ignore[attr-defined]
+            model_path = os.path.join(models_dir, 'GFPGANv1.4.pth')
+            FACE_ENHANCER = gfpgan.GFPGANer(model_path=model_path, upscale=1)  # type: ignore[attr-defined]
     return FACE_ENHANCER
 
 
 def enhance_face(temp_frame: Frame) -> Frame:
     with THREAD_SEMAPHORE:
-        _, _, temp_frame = get_face_enhancer().enhance(
-            temp_frame,
-            paste_back=True
-        )
+        _, _, temp_frame = get_face_enhancer().enhance(temp_frame, paste_back=True)
     return temp_frame
 
 
@@ -60,7 +66,9 @@ def process_frame(source_face: Face, temp_frame: Frame) -> Frame:
     return temp_frame
 
 
-def process_frames(source_path: str, temp_frame_paths: List[str], progress: Any = None) -> None:
+def process_frames(
+    source_path: str, temp_frame_paths: List[str], progress: Any = None
+) -> None:
     for temp_frame_path in temp_frame_paths:
         temp_frame = cv2.imread(temp_frame_path)
         result = process_frame(None, temp_frame)
@@ -77,3 +85,10 @@ def process_image(source_path: str, target_path: str, output_path: str) -> None:
 
 def process_video(source_path: str, temp_frame_paths: List[str]) -> None:
     modules.processors.frame.core.process_video(None, temp_frame_paths, process_frames)
+
+
+def process_frame_v2(temp_frame: Frame) -> Frame:
+    target_face = get_one_face(temp_frame)
+    if target_face:
+        temp_frame = enhance_face(temp_frame)
+    return temp_frame

modules/processors/frame/face_swapper.py

@@ -10,19 +10,21 @@ from modules.face_analyser import get_one_face, get_many_faces, default_source_f
 from modules.typing import Face, Frame
 from modules.utilities import (
     conditional_download,
-    resolve_relative_path,
     is_image,
     is_video,
 )
 from modules.cluster_analysis import find_closest_centroid
+import os
 
 FACE_SWAPPER = None
 THREAD_LOCK = threading.Lock()
 NAME = "DLC.FACE-SWAPPER"
 
+abs_dir = os.path.dirname(os.path.abspath(__file__))
+models_dir = os.path.join(os.path.dirname(os.path.dirname(os.path.dirname(abs_dir))), 'models')
 
 def pre_check() -> bool:
-    download_directory_path = resolve_relative_path("../models")
+    download_directory_path = abs_dir
     conditional_download(
         download_directory_path,
         [
@@ -54,7 +56,7 @@ def get_face_swapper() -> Any:
 
     with THREAD_LOCK:
         if FACE_SWAPPER is None:
-            model_path = resolve_relative_path("../models/inswapper_128_fp16.onnx")
+            model_path = os.path.join(models_dir, 'inswapper_128_fp16.onnx')
             FACE_SWAPPER = insightface.model_zoo.get_model(
                 model_path, providers=modules.globals.execution_providers
             )
@@ -62,38 +64,37 @@ def get_face_swapper() -> Any:
 
 
 def swap_face(source_face: Face, target_face: Face, temp_frame: Frame) -> Frame:
-    swapped_frame = get_face_swapper().get(
+    face_swapper = get_face_swapper()
+
+    # Apply the face swap
+    swapped_frame = face_swapper.get(
         temp_frame, target_face, source_face, paste_back=True
     )
 
-    # Apply opacity if enabled
-    if modules.globals.opacity_switch:
-        opacity = modules.globals.face_opacity / 100
-        swapped_frame = cv2.addWeighted(
-            swapped_frame, opacity, temp_frame, 1 - opacity, 0
+    if modules.globals.mouth_mask:
+        # Create a mask for the target face
+        face_mask = create_face_mask(target_face, temp_frame)
+
+        # Create the mouth mask
+        mouth_mask, mouth_cutout, mouth_box, lower_lip_polygon = (
+            create_lower_mouth_mask(target_face, temp_frame)
         )
 
-    # Apply mouth mask if enabled
-    if modules.globals.mouth_mask:
-        face_mask = create_face_mask(target_face, temp_frame)
-        mouth_mask_data = create_lower_mouth_mask(target_face, temp_frame)
-        mouth_mask, mouth_cutout, mouth_box, lower_lip_polygon = mouth_mask_data
+        # Apply the mouth area
+        swapped_frame = apply_mouth_area(
+            swapped_frame, mouth_cutout, mouth_box, face_mask, lower_lip_polygon
+        )
 
-        if mouth_box is not None:
-            swapped_frame = apply_mouth_area(
-                swapped_frame, mouth_cutout, mouth_box, face_mask, lower_lip_polygon
+        if modules.globals.show_mouth_mask_box:
+            mouth_mask_data = (mouth_mask, mouth_cutout, mouth_box, lower_lip_polygon)
+            swapped_frame = draw_mouth_mask_visualization(
+                swapped_frame, target_face, mouth_mask_data
             )
 
-            if modules.globals.show_mouth_mask_box:
-                swapped_frame = draw_mouth_mask_visualization(
-                    swapped_frame, target_face, mouth_mask_data
-                )
-
     return swapped_frame
 
 
 def process_frame(source_face: Face, temp_frame: Frame) -> Frame:
-    # Ensure the frame is in RGB format if color correction is enabled
     if modules.globals.color_correction:
         temp_frame = cv2.cvtColor(temp_frame, cv2.COLOR_BGR2RGB)
 
@@ -151,6 +152,7 @@ def process_frame_v2(temp_frame: Frame, temp_frame_path: str = "") -> Frame:
                     for frame in target_frame:
                         for target_face in frame["faces"]:
                             temp_frame = swap_face(source_face, target_face, temp_frame)
+
     else:
         detected_faces = get_many_faces(temp_frame)
         if modules.globals.many_faces:
@@ -252,65 +254,14 @@ def process_video(source_path: str, temp_frame_paths: List[str]) -> None:
     )
 
 
-def create_face_mask(face: Face, frame: Frame) -> np.ndarray:
-    mask = np.zeros(frame.shape[:2], dtype=np.uint8)
-    landmarks = face.landmark_2d_106
-    if landmarks is not None:
-        landmarks = landmarks.astype(np.int32)
-
-        right_side_face = landmarks[0:16]
-        left_side_face = landmarks[17:32]
-        right_eye = landmarks[33:42]
-        right_eye_brow = landmarks[43:51]
-        left_eye = landmarks[87:96]
-        left_eye_brow = landmarks[97:105]
-
-        right_eyebrow_top = np.min(right_eye_brow[:, 1])
-        left_eyebrow_top = np.min(left_eye_brow[:, 1])
-        eyebrow_top = min(right_eyebrow_top, left_eyebrow_top)
-
-        face_top = np.min([right_side_face[0, 1], left_side_face[-1, 1]])
-        forehead_height = face_top - eyebrow_top
-        extended_forehead_height = int(forehead_height * 5.0)
-
-        forehead_left = right_side_face[0].copy()
-        forehead_right = left_side_face[-1].copy()
-        forehead_left[1] -= extended_forehead_height
-        forehead_right[1] -= extended_forehead_height
-
-        face_outline = np.vstack(
-            [
-                [forehead_left],
-                right_side_face,
-                left_side_face[::-1],
-                [forehead_right],
-            ]
-        )
-
-        padding = int(np.linalg.norm(right_side_face[0] - left_side_face[-1]) * 0.05)
-
-        hull = cv2.convexHull(face_outline)
-        hull_padded = []
-        for point in hull:
-            x, y = point[0]
-            center = np.mean(face_outline, axis=0)
-            direction = np.array([x, y]) - center
-            direction = direction / np.linalg.norm(direction)
-            padded_point = np.array([x, y]) + direction * padding
-            hull_padded.append(padded_point)
-
-        hull_padded = np.array(hull_padded, dtype=np.int32)
-        cv2.fillConvexPoly(mask, hull_padded, 255)
-        mask = cv2.GaussianBlur(mask, (5, 5), 3)
-
-    return mask
-
-
-def create_lower_mouth_mask(face: Face, frame: Frame) -> tuple:
+def create_lower_mouth_mask(
+    face: Face, frame: Frame
+) -> (np.ndarray, np.ndarray, tuple, np.ndarray):
     mask = np.zeros(frame.shape[:2], dtype=np.uint8)
     mouth_cutout = None
     landmarks = face.landmark_2d_106
     if landmarks is not None:
+        #                  0  1  2  3  4  5  6  7  8  9  10 11 12 13 14 15 16 17 18 19 20
         lower_lip_order = [
             65,
             66,
@@ -334,61 +285,181 @@ def create_lower_mouth_mask(face: Face, frame: Frame) -> tuple:
             2,
             65,
         ]
-        lower_lip_landmarks = landmarks[lower_lip_order].astype(np.float32)
+        lower_lip_landmarks = landmarks[lower_lip_order].astype(
+            np.float32
+        )  # Use float for precise calculations
 
+        # Calculate the center of the landmarks
         center = np.mean(lower_lip_landmarks, axis=0)
-        expansion_factor = 1 + modules.globals.mask_down_size
+
+        # Expand the landmarks outward
+        expansion_factor = (
+            1 + modules.globals.mask_down_size
+        )  # Adjust this for more or less expansion
         expanded_landmarks = (lower_lip_landmarks - center) * expansion_factor + center
 
-        toplip_indices = [20, 0, 1, 2, 3, 4, 5]
-        toplip_extension = modules.globals.mask_size * 0.5
+        # Extend the top lip part
+        toplip_indices = [
+            20,
+            0,
+            1,
+            2,
+            3,
+            4,
+            5,
+        ]  # Indices for landmarks 2, 65, 66, 62, 70, 69, 18
+        toplip_extension = (
+            modules.globals.mask_size * 0.5
+        )  # Adjust this factor to control the extension
         for idx in toplip_indices:
             direction = expanded_landmarks[idx] - center
             direction = direction / np.linalg.norm(direction)
             expanded_landmarks[idx] += direction * toplip_extension
 
-        chin_indices = [11, 12, 13, 14, 15, 16]
-        chin_extension = 2 * 0.2
+        # Extend the bottom part (chin area)
+        chin_indices = [
+            11,
+            12,
+            13,
+            14,
+            15,
+            16,
+        ]  # Indices for landmarks 21, 22, 23, 24, 0, 8
+        chin_extension = 2 * 0.2  # Adjust this factor to control the extension
         for idx in chin_indices:
             expanded_landmarks[idx][1] += (
                 expanded_landmarks[idx][1] - center[1]
             ) * chin_extension
 
+        # Convert back to integer coordinates
         expanded_landmarks = expanded_landmarks.astype(np.int32)
 
+        # Calculate bounding box for the expanded lower mouth
         min_x, min_y = np.min(expanded_landmarks, axis=0)
         max_x, max_y = np.max(expanded_landmarks, axis=0)
 
-        padding = int((max_x - min_x) * 0.1)
+        # Add some padding to the bounding box
+        padding = int((max_x - min_x) * 0.1)  # 10% padding
         min_x = max(0, min_x - padding)
         min_y = max(0, min_y - padding)
         max_x = min(frame.shape[1], max_x + padding)
         max_y = min(frame.shape[0], max_y + padding)
 
+        # Ensure the bounding box dimensions are valid
         if max_x <= min_x or max_y <= min_y:
             if (max_x - min_x) <= 1:
                 max_x = min_x + 1
             if (max_y - min_y) <= 1:
                 max_y = min_y + 1
 
+        # Create the mask
         mask_roi = np.zeros((max_y - min_y, max_x - min_x), dtype=np.uint8)
         cv2.fillPoly(mask_roi, [expanded_landmarks - [min_x, min_y]], 255)
+
+        # Apply Gaussian blur to soften the mask edges
         mask_roi = cv2.GaussianBlur(mask_roi, (15, 15), 5)
+
+        # Place the mask ROI in the full-sized mask
         mask[min_y:max_y, min_x:max_x] = mask_roi
+
+        # Extract the masked area from the frame
         mouth_cutout = frame[min_y:max_y, min_x:max_x].copy()
 
-        return mask, mouth_cutout, (min_x, min_y, max_x, max_y), expanded_landmarks
+        # Return the expanded lower lip polygon in original frame coordinates
+        lower_lip_polygon = expanded_landmarks
 
-    return mask, mouth_cutout, None, None
+    return mask, mouth_cutout, (min_x, min_y, max_x, max_y), lower_lip_polygon
+
+
+def draw_mouth_mask_visualization(
+    frame: Frame, face: Face, mouth_mask_data: tuple
+) -> Frame:
+    landmarks = face.landmark_2d_106
+    if landmarks is not None and mouth_mask_data is not None:
+        mask, mouth_cutout, (min_x, min_y, max_x, max_y), lower_lip_polygon = (
+            mouth_mask_data
+        )
+
+        vis_frame = frame.copy()
+
+        # Ensure coordinates are within frame bounds
+        height, width = vis_frame.shape[:2]
+        min_x, min_y = max(0, min_x), max(0, min_y)
+        max_x, max_y = min(width, max_x), min(height, max_y)
+
+        # Adjust mask to match the region size
+        mask_region = mask[0 : max_y - min_y, 0 : max_x - min_x]
+
+        # Remove the color mask overlay
+        # color_mask = cv2.applyColorMap((mask_region * 255).astype(np.uint8), cv2.COLORMAP_JET)
+
+        # Ensure shapes match before blending
+        vis_region = vis_frame[min_y:max_y, min_x:max_x]
+        # Remove blending with color_mask
+        # if vis_region.shape[:2] == color_mask.shape[:2]:
+        #     blended = cv2.addWeighted(vis_region, 0.7, color_mask, 0.3, 0)
+        #     vis_frame[min_y:max_y, min_x:max_x] = blended
+
+        # Draw the lower lip polygon
+        cv2.polylines(vis_frame, [lower_lip_polygon], True, (0, 255, 0), 2)
+
+        # Remove the red box
+        # cv2.rectangle(vis_frame, (min_x, min_y), (max_x, max_y), (0, 0, 255), 2)
+
+        # Visualize the feathered mask
+        feather_amount = max(
+            1,
+            min(
+                30,
+                (max_x - min_x) // modules.globals.mask_feather_ratio,
+                (max_y - min_y) // modules.globals.mask_feather_ratio,
+            ),
+        )
+        # Ensure kernel size is odd
+        kernel_size = 2 * feather_amount + 1
+        feathered_mask = cv2.GaussianBlur(
+            mask_region.astype(float), (kernel_size, kernel_size), 0
+        )
+        feathered_mask = (feathered_mask / feathered_mask.max() * 255).astype(np.uint8)
+        # Remove the feathered mask color overlay
+        # color_feathered_mask = cv2.applyColorMap(feathered_mask, cv2.COLORMAP_VIRIDIS)
+
+        # Ensure shapes match before blending feathered mask
+        # if vis_region.shape == color_feathered_mask.shape:
+        #     blended_feathered = cv2.addWeighted(vis_region, 0.7, color_feathered_mask, 0.3, 0)
+        #     vis_frame[min_y:max_y, min_x:max_x] = blended_feathered
+
+        # Add labels
+        cv2.putText(
+            vis_frame,
+            "Lower Mouth Mask",
+            (min_x, min_y - 10),
+            cv2.FONT_HERSHEY_SIMPLEX,
+            0.5,
+            (255, 255, 255),
+            1,
+        )
+        cv2.putText(
+            vis_frame,
+            "Feathered Mask",
+            (min_x, max_y + 20),
+            cv2.FONT_HERSHEY_SIMPLEX,
+            0.5,
+            (255, 255, 255),
+            1,
+        )
+
+        return vis_frame
+    return frame
 
 
 def apply_mouth_area(
-    frame: Frame,
+    frame: np.ndarray,
     mouth_cutout: np.ndarray,
     mouth_box: tuple,
     face_mask: np.ndarray,
     mouth_polygon: np.ndarray,
-) -> Frame:
+) -> np.ndarray:
     min_x, min_y, max_x, max_y = mouth_box
     box_width = max_x - min_x
     box_height = max_y - min_y
@@ -413,10 +484,12 @@ def apply_mouth_area(
 
         color_corrected_mouth = apply_color_transfer(resized_mouth_cutout, roi)
 
+        # Use the provided mouth polygon to create the mask
         polygon_mask = np.zeros(roi.shape[:2], dtype=np.uint8)
         adjusted_polygon = mouth_polygon - [min_x, min_y]
         cv2.fillPoly(polygon_mask, [adjusted_polygon], 255)
 
+        # Apply feathering to the polygon mask
         feather_amount = min(
             30,
             box_width // modules.globals.mask_feather_ratio,
@@ -435,6 +508,7 @@ def apply_mouth_area(
             color_corrected_mouth * combined_mask + roi * (1 - combined_mask)
         ).astype(np.uint8)
 
+        # Apply face mask to blended result
         face_mask_3channel = (
             np.repeat(face_mask_roi[:, :, np.newaxis], 3, axis=2) / 255.0
         )
@@ -447,47 +521,92 @@ def apply_mouth_area(
     return frame
 
 
-def apply_color_transfer(source: np.ndarray, target: np.ndarray) -> np.ndarray:
+def create_face_mask(face: Face, frame: Frame) -> np.ndarray:
+    mask = np.zeros(frame.shape[:2], dtype=np.uint8)
+    landmarks = face.landmark_2d_106
+    if landmarks is not None:
+        # Convert landmarks to int32
+        landmarks = landmarks.astype(np.int32)
+
+        # Extract facial features
+        right_side_face = landmarks[0:16]
+        left_side_face = landmarks[17:32]
+        right_eye = landmarks[33:42]
+        right_eye_brow = landmarks[43:51]
+        left_eye = landmarks[87:96]
+        left_eye_brow = landmarks[97:105]
+
+        # Calculate forehead extension
+        right_eyebrow_top = np.min(right_eye_brow[:, 1])
+        left_eyebrow_top = np.min(left_eye_brow[:, 1])
+        eyebrow_top = min(right_eyebrow_top, left_eyebrow_top)
+
+        face_top = np.min([right_side_face[0, 1], left_side_face[-1, 1]])
+        forehead_height = face_top - eyebrow_top
+        extended_forehead_height = int(forehead_height * 5.0)  # Extend by 50%
+
+        # Create forehead points
+        forehead_left = right_side_face[0].copy()
+        forehead_right = left_side_face[-1].copy()
+        forehead_left[1] -= extended_forehead_height
+        forehead_right[1] -= extended_forehead_height
+
+        # Combine all points to create the face outline
+        face_outline = np.vstack(
+            [
+                [forehead_left],
+                right_side_face,
+                left_side_face[
+                    ::-1
+                ],  # Reverse left side to create a continuous outline
+                [forehead_right],
+            ]
+        )
+
+        # Calculate padding
+        padding = int(
+            np.linalg.norm(right_side_face[0] - left_side_face[-1]) * 0.05
+        )  # 5% of face width
+
+        # Create a slightly larger convex hull for padding
+        hull = cv2.convexHull(face_outline)
+        hull_padded = []
+        for point in hull:
+            x, y = point[0]
+            center = np.mean(face_outline, axis=0)
+            direction = np.array([x, y]) - center
+            direction = direction / np.linalg.norm(direction)
+            padded_point = np.array([x, y]) + direction * padding
+            hull_padded.append(padded_point)
+
+        hull_padded = np.array(hull_padded, dtype=np.int32)
+
+        # Fill the padded convex hull
+        cv2.fillConvexPoly(mask, hull_padded, 255)
+
+        # Smooth the mask edges
+        mask = cv2.GaussianBlur(mask, (5, 5), 3)
+
+    return mask
+
+
+def apply_color_transfer(source, target):
+    """
+    Apply color transfer from target to source image
+    """
     source = cv2.cvtColor(source, cv2.COLOR_BGR2LAB).astype("float32")
     target = cv2.cvtColor(target, cv2.COLOR_BGR2LAB).astype("float32")
 
     source_mean, source_std = cv2.meanStdDev(source)
     target_mean, target_std = cv2.meanStdDev(target)
 
+    # Reshape mean and std to be broadcastable
     source_mean = source_mean.reshape(1, 1, 3)
     source_std = source_std.reshape(1, 1, 3)
     target_mean = target_mean.reshape(1, 1, 3)
     target_std = target_std.reshape(1, 1, 3)
 
+    # Perform the color transfer
     source = (source - source_mean) * (target_std / source_std) + target_mean
 
     return cv2.cvtColor(np.clip(source, 0, 255).astype("uint8"), cv2.COLOR_LAB2BGR)
-
-
-def draw_mouth_mask_visualization(
-    frame: Frame, face: Face, mouth_mask_data: tuple
-) -> Frame:
-    landmarks = face.landmark_2d_106
-    if landmarks is not None and mouth_mask_data is not None:
-        mask, mouth_cutout, (min_x, min_y, max_x, max_y), lower_lip_polygon = (
-            mouth_mask_data
-        )
-
-        vis_frame = frame.copy()
-
-        # Draw the lower lip polygon
-        cv2.polylines(vis_frame, [lower_lip_polygon], True, (0, 255, 0), 2)
-
-        # Add labels
-        cv2.putText(
-            vis_frame,
-            "Mouth Mask",
-            (min_x, min_y - 10),
-            cv2.FONT_HERSHEY_SIMPLEX,
-            0.5,
-            (255, 255, 255),
-            1,
-        )
-
-        return vis_frame
-    return frame

requirements.txt

@@ -7,6 +7,7 @@ onnx==1.16.0
 insightface==0.7.3
 psutil==5.9.8
 tk==0.1.0
+customtkinter==5.2.2
 pillow==9.5.0
 torch==2.0.1+cu118; sys_platform != 'darwin'
 torch==2.0.1; sys_platform == 'darwin'
@@ -20,4 +21,3 @@ protobuf==4.23.2
 tqdm==4.66.4
 gfpgan==1.3.8
 tkinterdnd2==0.4.2
-customtkinter==5.2.2