Merge pull request #1713 from TeachDian/fix-1705-wsl-onnxruntime-gpu

320 was over optimized, put back to 640

.gitignore

@@ -26,3 +26,4 @@ faceswap/
 .vscode/
 switch_states.json
 /models
+install.bat

README.md

@@ -1,4 +1,4 @@
-<h1 align="center">Deep-Live-Cam 2.0.4c</h1>
+<h1 align="center">Deep-Live-Cam 2.1</h1>
 
 <p align="center">
   Real-time face swap and video deepfake with a single click and only a single image.
@@ -30,11 +30,11 @@ By using this software, you agree to these terms and commit to using it in a man
 
 Users are expected to use this software responsibly and legally. If using a real person's face, obtain their consent and clearly label any output as a deepfake when sharing online. We are not responsible for end-user actions.
 
-## Exclusive v2.6d Quick Start - Pre-built (Windows/Mac Silicon)
+## Exclusive v2.7 beta Quick Start - Pre-built (Windows/Mac Silicon/CPU)
 
   <a href="https://deeplivecam.net/index.php/quickstart"> <img src="media/Download.png" width="285" height="77" />
 
-##### This is the fastest build you can get if you have a discrete NVIDIA or AMD GPU or Mac Silicon, And you'll receive special priority support.
+##### This is the fastest build you can get if you have a discrete NVIDIA or AMD GPU, CPU or Mac Silicon, And you'll receive special priority support. 2.7 beta is the best you can have with 30+ extra features than the open source version.
  
 ###### These Pre-builts are perfect for non-technical users or those who don't have time to, or can't manually install all the requirements. Just a heads-up: this is an open-source project, so you can also install it manually. 
 
@@ -124,7 +124,7 @@ cd Deep-Live-Cam
 
 **3. Download the Models**
 
-1. [GFPGANv1.4](https://huggingface.co/hacksider/deep-live-cam/resolve/main/GFPGANv1.4.pth)
+1. [GFPGANv1.4](https://huggingface.co/hacksider/deep-live-cam/resolve/main/GFPGANv1.4.onnx)
 2. [inswapper\_128\_fp16.onnx](https://huggingface.co/hacksider/deep-live-cam/resolve/main/inswapper_128_fp16.onnx)
 
 Place these files in the "**models**" folder.
@@ -309,6 +309,9 @@ python run.py --execution-provider openvino
 -   Use a screen capture tool like OBS to stream.
 -   To change the face, select a new source image.
 
+## Download all models in this huggingface link
+- [**Download models here**](https://huggingface.co/hacksider/deep-live-cam/tree/main)
+
 ## Command Line Arguments (Unmaintained)
 
 ```
@@ -338,23 +341,16 @@ Looking for a CLI mode? Using the -s/--source argument will make the run program
 
 ## Press
 
-**We are always open to criticism and are ready to improve, that's why we didn't cherry-pick anything.**
-
- - [*"Deep-Live-Cam goes viral, allowing anyone to become a digital doppelganger"*](https://arstechnica.com/information-technology/2024/08/new-ai-tool-enables-real-time-face-swapping-on-webcams-raising-fraud-concerns/) - Ars Technica
- - [*"Thanks Deep Live Cam, shapeshifters are among us now"*](https://dataconomy.com/2024/08/15/what-is-deep-live-cam-github-deepfake/) - Dataconomy
- - [*"This free AI tool lets you become anyone during video-calls"*](https://www.newsbytesapp.com/news/science/deep-live-cam-ai-impersonation-tool-goes-viral/story) - NewsBytes
- - [*"OK, this viral AI live stream software is truly terrifying"*](https://www.creativebloq.com/ai/ok-this-viral-ai-live-stream-software-is-truly-terrifying) - Creative Bloq
- - [*"Deepfake AI Tool Lets You Become Anyone in a Video Call With Single Photo"*](https://petapixel.com/2024/08/14/deep-live-cam-deepfake-ai-tool-lets-you-become-anyone-in-a-video-call-with-single-photo-mark-zuckerberg-jd-vance-elon-musk/) - PetaPixel
- - [*"Deep-Live-Cam Uses AI to Transform Your Face in Real-Time, Celebrities Included"*](https://www.techeblog.com/deep-live-cam-ai-transform-face/) - TechEBlog
- - [*"An AI tool that "makes you look like anyone" during a video call is going viral online"*](https://telegrafi.com/en/a-tool-that-makes-you-look-like-anyone-during-a-video-call-is-going-viral-on-the-Internet/) - Telegrafi
- - [*"This Deepfake Tool Turning Images Into Livestreams is Topping the GitHub Charts"*](https://decrypt.co/244565/this-deepfake-tool-turning-images-into-livestreams-is-topping-the-github-charts) - Emerge
- - [*"New Real-Time Face-Swapping AI Allows Anyone to Mimic Famous Faces"*](https://www.digitalmusicnews.com/2024/08/15/face-swapping-ai-real-time-mimic/) - Digital Music News
- - [*"This real-time webcam deepfake tool raises alarms about the future of identity theft"*](https://www.diyphotography.net/this-real-time-webcam-deepfake-tool-raises-alarms-about-the-future-of-identity-theft/) - DIYPhotography
- - [*"That's Crazy, Oh God. That's Fucking Freaky Dude... That's So Wild Dude"*](https://www.youtube.com/watch?time_continue=1074&v=py4Tc-Y8BcY) - SomeOrdinaryGamers
- - [*"Alright look look look, now look chat, we can do any face we want to look like chat"*](https://www.youtube.com/live/mFsCe7AIxq8?feature=shared&t=2686) - IShowSpeed
- - [*"They do a pretty good job matching poses, expression and even the lighting"*](https://www.youtube.com/watch?v=wnCghLjqv3s&t=551s) - TechLinked (LTT)
- - [*"Als Sean Connery an der Redaktionskonferenz teilnahm"*](https://www.golem.de/news/deepfakes-als-sean-connery-an-der-redaktionskonferenz-teilnahm-2408-188172.html) - Golem.de (German)
- - [*"What the F***! Why do I look like Vinny Jr? I look exactly like Vinny Jr!? No, this shit is crazy! Bro This is F*** Crazy! "*](https://youtu.be/JbUPRmXRUtE?t=3964) - IShowSpeed
+ - [**Ars Technica**](https://arstechnica.com/information-technology/2024/08/new-ai-tool-enables-real-time-face-swapping-on-webcams-raising-fraud-concerns/) - *"Deep-Live-Cam goes viral, allowing anyone to become a digital doppelganger"*
+ - [**Yahoo!**](https://www.yahoo.com/tech/ok-viral-ai-live-stream-080041056.html) - *"OK, this viral AI live stream software is truly terrifying"*
+ - [**CNN Brasil**](https://www.cnnbrasil.com.br/tecnologia/ia-consegue-clonar-rostos-na-webcam-entenda-funcionamento/) - *"AI can clone faces on webcam; understand how it works"*
+ - [**Bloomberg Technoz**](https://www.bloombergtechnoz.com/detail-news/71032/kenalan-dengan-teknologi-deep-live-cam-bisa-jadi-alat-menipu) - *"Get to know Deep Live Cam technology, it can be used as a tool for deception."*
+ - [**TrendMicro**](https://www.trendmicro.com/vinfo/gb/security/news/cyber-attacks/ai-vs-ai-deepfakes-and-ekyc) - *"AI vs AI: DeepFakes and eKYC"*
+ - [**PetaPixel**](https://petapixel.com/2024/08/14/deep-live-cam-deepfake-ai-tool-lets-you-become-anyone-in-a-video-call-with-single-photo-mark-zuckerberg-jd-vance-elon-musk/) - *"Deepfake AI Tool Lets You Become Anyone in a Video Call With Single Photo"*
+ - [**SomeOrdinaryGamers**](https://www.youtube.com/watch?time_continue=1074&v=py4Tc-Y8BcY) - *"That's Crazy, Oh God. That's Fucking Freaky Dude... That's So Wild Dude"*
+ - [**IShowSpeed**](https://www.youtube.com/live/mFsCe7AIxq8?feature=shared&t=2686) - *"Alright look look look, now look chat, we can do any face we want to look like chat"*
+ - [**TechLinked (Linus Tech Tips)**](https://www.youtube.com/watch?v=wnCghLjqv3s&t=551s) - *"They do a pretty good job matching poses, expression and even the lighting"*
+ - [**IShowSpeed**](https://youtu.be/JbUPRmXRUtE?t=3964) - *"What the F***! Why do I look like Vinny Jr? I look exactly like Vinny Jr!? No, this shit is crazy! Bro This is F*** Crazy!"*
 
 
 ## Credits
@@ -368,6 +364,7 @@ Looking for a CLI mode? Using the -s/--source argument will make the run program
 -   [vic4key](https://github.com/vic4key): For supporting/contributing to this project
 -   [kier007](https://github.com/kier007): for improving the user experience
 -   [qitianai](https://github.com/qitianai): for multi-lingual support
+-   [laurigates](https://github.com/laurigates): Decoupling stuffs to make everything faster!
 -   and [all developers](https://github.com/hacksider/Deep-Live-Cam/graphs/contributors) behind libraries used in this project.
 -   Footnote: Please be informed that the base author of the code is [s0md3v](https://github.com/s0md3v/roop)
 -   All the wonderful users who helped make this project go viral by starring the repo ❤️

modules/core.py

@@ -39,7 +39,7 @@ def parse_args() -> None:
     program.add_argument('-s', '--source', help='select an source image', dest='source_path')
     program.add_argument('-t', '--target', help='select an target image or video', dest='target_path')
     program.add_argument('-o', '--output', help='select output file or directory', dest='output_path')
-    program.add_argument('--frame-processor', help='pipeline of frame processors', dest='frame_processor', default=['face_swapper'], choices=['face_swapper', 'face_enhancer'], nargs='+')
+    program.add_argument('--frame-processor', help='pipeline of frame processors', dest='frame_processor', default=['face_swapper'], choices=['face_swapper', 'face_enhancer', 'face_enhancer_gpen256', 'face_enhancer_gpen512'], nargs='+')
     program.add_argument('--keep-fps', help='keep original fps', dest='keep_fps', action='store_true', default=False)
     program.add_argument('--keep-audio', help='keep original audio', dest='keep_audio', action='store_true', default=True)
     program.add_argument('--keep-frames', help='keep temporary frames', dest='keep_frames', action='store_true', default=False)
@@ -86,11 +86,9 @@ def parse_args() -> None:
     modules.globals.execution_threads = args.execution_threads
     modules.globals.lang = args.lang
 
-    #for ENHANCER tumbler:
-    if 'face_enhancer' in args.frame_processor:
-        modules.globals.fp_ui['face_enhancer'] = True
-    else:
-        modules.globals.fp_ui['face_enhancer'] = False
+    #for ENHANCER tumblers:
+    for enhancer_key in ('face_enhancer', 'face_enhancer_gpen256', 'face_enhancer_gpen512'):
+        modules.globals.fp_ui[enhancer_key] = enhancer_key in args.frame_processor
 
     # translate deprecated args
     if args.source_path_deprecated:

modules/face_analyser.py

@@ -28,9 +28,9 @@ def get_face_analyser() -> Any:
                 FACE_ANALYSER = insightface.app.FaceAnalysis(
                     name='buffalo_l',
                     providers=modules.globals.execution_providers,
-                    allowed_modules=['detection', 'recognition']
+                    allowed_modules=['detection', 'recognition', 'landmark_2d_106']
                 )
-                FACE_ANALYSER.prepare(ctx_id=0, det_size=(320, 320))
+                FACE_ANALYSER.prepare(ctx_id=0, det_size=(640, 640))
     return FACE_ANALYSER
 
 

modules/globals.py

@@ -50,7 +50,7 @@ headless: bool | None = None         # Run without UI?
 log_level: str = "error"             # Logging level (e.g., 'debug', 'info', 'warning', 'error')
 
 # Face Processor UI Toggles (Example)
-fp_ui: Dict[str, bool] = {"face_enhancer": False}
+fp_ui: Dict[str, bool] = {"face_enhancer": False, "face_enhancer_gpen256": False, "face_enhancer_gpen512": False}
 
 # Face Swapper Specific Options
 face_swapper_enabled: bool = True # General toggle for the swapper processor
@@ -63,6 +63,7 @@ show_mouth_mask_box: bool = False  # Visualize the mouth mask area (for debuggin
 mask_feather_ratio: int = 12       # Denominator for feathering calculation (higher = smaller feather)
 mask_down_size: float = 0.1        # Expansion factor for lower lip mask (relative)
 mask_size: float = 1.0             # Expansion factor for upper lip mask (relative)
+mouth_mask_size: float = 0.0       # Mouth mask size (0-100; 0=off, 100=mouth to chin)
 
 # --- START: Added for Frame Interpolation ---
 enable_interpolation: bool = True # Toggle temporal smoothing

modules/metadata.py

@@ -1,3 +1,3 @@
 name = 'Deep-Live-Cam'
-version = '2.0.3c'
+version = '2.1'
 edition = 'GitHub Edition'

modules/paths.py

@@ -0,0 +1,6 @@
+"""Shared path constants for the Deep-Live-Cam project."""
+
+import os
+
+ROOT_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+MODELS_DIR = os.path.join(ROOT_DIR, "models")

modules/processors/frame/_onnx_enhancer.py

@@ -0,0 +1,145 @@
+"""Shared ONNX-based face enhancement utilities for GPEN-BFR models.
+
+Provides session creation, pre/post processing, and the core
+enhance-face-via-ONNX pipeline.
+"""
+
+import os
+import platform
+import threading
+from typing import Any
+
+import cv2
+import numpy as np
+import onnxruntime
+
+import modules.globals
+
+IS_APPLE_SILICON = platform.system() == "Darwin" and platform.machine() == "arm64"
+
+# Limit concurrent ONNX calls to avoid VRAM exhaustion on multi-face frames
+THREAD_SEMAPHORE = threading.Semaphore(min(max(1, (os.cpu_count() or 1)), 8))
+
+
+def create_onnx_session(model_path: str) -> onnxruntime.InferenceSession:
+    """Create an ONNX Runtime session using the configured execution providers."""
+    providers = modules.globals.execution_providers
+    session = onnxruntime.InferenceSession(model_path, providers=providers)
+    return session
+
+
+def warmup_session(session: onnxruntime.InferenceSession) -> None:
+    """Run a dummy inference pass to trigger JIT / compile caching."""
+    try:
+        input_feed = {
+            inp.name: np.zeros(
+                [d if isinstance(d, int) and d > 0 else 1 for d in inp.shape],
+                dtype=np.float32,
+            )
+            for inp in session.get_inputs()
+        }
+        session.run(None, input_feed)
+    except Exception as e:
+        print(f"ONNX enhancer warmup skipped (non-fatal): {e}")
+
+
+def preprocess_face(face_img: np.ndarray, input_size: int) -> np.ndarray:
+    """Resize, normalize, and convert a BGR face crop to ONNX input blob.
+
+    GPEN-BFR expects [1, 3, H, W] float32 in RGB, normalized to [-1, 1].
+    """
+    resized = cv2.resize(face_img, (input_size, input_size), interpolation=cv2.INTER_LINEAR)
+    rgb = cv2.cvtColor(resized, cv2.COLOR_BGR2RGB)
+    blob = rgb.astype(np.float32) / 255.0 * 2.0 - 1.0
+    blob = np.transpose(blob, (2, 0, 1))[np.newaxis, ...]
+    return blob
+
+
+def postprocess_face(output: np.ndarray) -> np.ndarray:
+    """Convert ONNX output [1, 3, H, W] float32 back to BGR uint8 image."""
+    img = output[0].transpose(1, 2, 0)
+    img = ((img + 1.0) / 2.0 * 255.0)
+    img = np.clip(img, 0, 255).astype(np.uint8)
+    img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
+    return img
+
+
+def _get_face_affine(face: Any, input_size: int):
+    """Compute affine transform to align a face to GPEN input space.
+
+    Returns (M, inv_M) — forward and inverse affine matrices.
+    """
+    template = np.array([
+        [0.31556875, 0.4615741],
+        [0.68262291, 0.4615741],
+        [0.50009375, 0.6405054],
+        [0.34947187, 0.8246919],
+        [0.65343645, 0.8246919],
+    ], dtype=np.float32) * input_size
+
+    landmarks = None
+    if hasattr(face, "kps") and face.kps is not None:
+        landmarks = face.kps.astype(np.float32)
+    elif hasattr(face, "landmark_2d_106") and face.landmark_2d_106 is not None:
+        lm106 = face.landmark_2d_106
+        landmarks = np.array([
+            lm106[38],  # left eye
+            lm106[88],  # right eye
+            lm106[86],  # nose tip
+            lm106[52],  # left mouth
+            lm106[61],  # right mouth
+        ], dtype=np.float32)
+
+    if landmarks is None or len(landmarks) < 5:
+        return None, None
+
+    M = cv2.estimateAffinePartial2D(landmarks, template, method=cv2.LMEDS)[0]
+    if M is None:
+        return None, None
+    inv_M = cv2.invertAffineTransform(M)
+    return M, inv_M
+
+
+def enhance_face_onnx(
+    frame: np.ndarray,
+    face: Any,
+    session: onnxruntime.InferenceSession,
+    input_size: int,
+) -> np.ndarray:
+    """Enhance a single face in the frame using an ONNX face restoration model."""
+    M, inv_M = _get_face_affine(face, input_size)
+    if M is None:
+        return frame
+
+    face_crop = cv2.warpAffine(
+        frame, M, (input_size, input_size),
+        flags=cv2.INTER_LINEAR, borderMode=cv2.BORDER_REPLICATE,
+    )
+
+    blob = preprocess_face(face_crop, input_size)
+    with THREAD_SEMAPHORE:
+        output = session.run(None, {session.get_inputs()[0].name: blob})[0]
+    enhanced = postprocess_face(output)
+
+    # Create mask for blending (feathered edges)
+    mask = np.ones((input_size, input_size), dtype=np.float32)
+    border = max(1, input_size // 16)
+    mask[:border, :] = np.linspace(0, 1, border)[:, np.newaxis]
+    mask[-border:, :] = np.linspace(1, 0, border)[:, np.newaxis]
+    mask[:, :border] = np.minimum(mask[:, :border], np.linspace(0, 1, border)[np.newaxis, :])
+    mask[:, -border:] = np.minimum(mask[:, -border:], np.linspace(1, 0, border)[np.newaxis, :])
+
+    h, w = frame.shape[:2]
+    warped_enhanced = cv2.warpAffine(
+        enhanced, inv_M, (w, h),
+        flags=cv2.INTER_LINEAR, borderValue=(0, 0, 0),
+    )
+    warped_mask = cv2.warpAffine(
+        mask, inv_M, (w, h),
+        flags=cv2.INTER_LINEAR, borderValue=0,
+    )
+
+    mask_3ch = warped_mask[:, :, np.newaxis]
+    result = (warped_enhanced.astype(np.float32) * mask_3ch +
+              frame.astype(np.float32) * (1.0 - mask_3ch))
+    return np.clip(result, 0, 255).astype(np.uint8)

modules/processors/frame/core.py

@@ -17,8 +17,17 @@ FRAME_PROCESSORS_INTERFACE = [
     'process_video'
 ]
 
+ALLOWED_PROCESSORS = {
+    'face_swapper',
+    'face_enhancer',
+    'face_enhancer_gpen256',
+    'face_enhancer_gpen512'
+}
 
 def load_frame_processor_module(frame_processor: str) -> Any:
+    if frame_processor not in ALLOWED_PROCESSORS:
+        print(f"Frame processor {frame_processor} is not allowed")
+        sys.exit()
     try:
         frame_processor_module = importlib.import_module(f'modules.processors.frame.{frame_processor}')
         for method_name in FRAME_PROCESSORS_INTERFACE:

modules/processors/frame/face_enhancer_gpen256.py

@@ -0,0 +1,125 @@
+"""GPEN-BFR-256 face enhancer — ONNX-based face restoration at 256x256."""
+
+from typing import Any, List
+import os
+import threading
+
+import cv2
+import numpy as np
+
+import modules.globals
+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 (
+    is_image,
+    is_video,
+)
+from modules.processors.frame._onnx_enhancer import (
+    create_onnx_session,
+    warmup_session,
+    enhance_face_onnx,
+)
+
+NAME = "DLC.FACE-ENHANCER-GPEN256"
+INPUT_SIZE = 256
+MODEL_URL = "https://github.com/harisreedhar/Face-Upscalers-ONNX/releases/download/GPEN-BFR/GPEN-BFR-256.onnx"
+MODEL_FILE = "GPEN-BFR-256.onnx"
+
+ENHANCER = None
+THREAD_LOCK = threading.Lock()
+
+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:
+    model_path = os.path.join(models_dir, MODEL_FILE)
+    if not os.path.exists(model_path):
+        update_status(f"Downloading {MODEL_FILE}...", NAME)
+        from modules.utilities import conditional_download
+        conditional_download(models_dir, [MODEL_URL])
+    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)
+        return False
+    return True
+
+
+def get_enhancer() -> Any:
+    global ENHANCER
+    with THREAD_LOCK:
+        if ENHANCER is None:
+            model_path = os.path.join(models_dir, MODEL_FILE)
+            if not os.path.exists(model_path):
+                from modules.utilities import conditional_download
+                conditional_download(models_dir, [MODEL_URL])
+            if not os.path.exists(model_path):
+                raise FileNotFoundError(f"Model file not found: {model_path}")
+            print(f"{NAME}: Loading ONNX model from {model_path}")
+            ENHANCER = create_onnx_session(model_path)
+            warmup_session(ENHANCER)
+            print(f"{NAME}: Model loaded successfully.")
+    return ENHANCER
+
+
+def enhance_face(temp_frame: Frame, face: Face) -> Frame:
+    try:
+        session = get_enhancer()
+    except Exception as e:
+        print(f"{NAME}: {e}")
+        return temp_frame
+    try:
+        return enhance_face_onnx(temp_frame, face, session, INPUT_SIZE)
+    except Exception as e:
+        print(f"{NAME}: Error during face enhancement: {e}")
+        return temp_frame
+
+
+def process_frame(source_face: Face | None, temp_frame: Frame) -> Frame:
+    target_face = get_one_face(temp_frame)
+    if target_face is None:
+        return temp_frame
+    return enhance_face(temp_frame, target_face)
+
+
+def process_frame_v2(temp_frame: Frame) -> Frame:
+    target_face = get_one_face(temp_frame)
+    if target_face:
+        temp_frame = enhance_face(temp_frame, target_face)
+    return temp_frame
+
+
+def process_frames(
+    source_path: str | None, temp_frame_paths: List[str], progress: Any = None
+) -> None:
+    for temp_frame_path in temp_frame_paths:
+        temp_frame = cv2.imread(temp_frame_path)
+        if temp_frame is None:
+            if progress:
+                progress.update(1)
+            continue
+        result = process_frame(None, temp_frame)
+        cv2.imwrite(temp_frame_path, result)
+        if progress:
+            progress.update(1)
+
+
+def process_image(source_path: str | None, target_path: str, output_path: str) -> None:
+    target_frame = cv2.imread(target_path)
+    if target_frame is None:
+        print(f"{NAME}: Error: Failed to read target image {target_path}")
+        return
+    result_frame = process_frame(None, target_frame)
+    cv2.imwrite(output_path, result_frame)
+    print(f"{NAME}: Enhanced image saved to {output_path}")
+
+
+def process_video(source_path: str | None, temp_frame_paths: List[str]) -> None:
+    modules.processors.frame.core.process_video(source_path, temp_frame_paths, process_frames)

modules/processors/frame/face_enhancer_gpen512.py

@@ -0,0 +1,125 @@
+"""GPEN-BFR-512 face enhancer — ONNX-based face restoration at 512x512."""
+
+from typing import Any, List
+import os
+import threading
+
+import cv2
+import numpy as np
+
+import modules.globals
+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 (
+    is_image,
+    is_video,
+)
+from modules.processors.frame._onnx_enhancer import (
+    create_onnx_session,
+    warmup_session,
+    enhance_face_onnx,
+)
+
+NAME = "DLC.FACE-ENHANCER-GPEN512"
+INPUT_SIZE = 512
+MODEL_URL = "https://github.com/harisreedhar/Face-Upscalers-ONNX/releases/download/GPEN-BFR/GPEN-BFR-512.onnx"
+MODEL_FILE = "GPEN-BFR-512.onnx"
+
+ENHANCER = None
+THREAD_LOCK = threading.Lock()
+
+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:
+    model_path = os.path.join(models_dir, MODEL_FILE)
+    if not os.path.exists(model_path):
+        update_status(f"Downloading {MODEL_FILE}...", NAME)
+        from modules.utilities import conditional_download
+        conditional_download(models_dir, [MODEL_URL])
+    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)
+        return False
+    return True
+
+
+def get_enhancer() -> Any:
+    global ENHANCER
+    with THREAD_LOCK:
+        if ENHANCER is None:
+            model_path = os.path.join(models_dir, MODEL_FILE)
+            if not os.path.exists(model_path):
+                from modules.utilities import conditional_download
+                conditional_download(models_dir, [MODEL_URL])
+            if not os.path.exists(model_path):
+                raise FileNotFoundError(f"Model file not found: {model_path}")
+            print(f"{NAME}: Loading ONNX model from {model_path}")
+            ENHANCER = create_onnx_session(model_path)
+            warmup_session(ENHANCER)
+            print(f"{NAME}: Model loaded successfully.")
+    return ENHANCER
+
+
+def enhance_face(temp_frame: Frame, face: Face) -> Frame:
+    try:
+        session = get_enhancer()
+    except Exception as e:
+        print(f"{NAME}: {e}")
+        return temp_frame
+    try:
+        return enhance_face_onnx(temp_frame, face, session, INPUT_SIZE)
+    except Exception as e:
+        print(f"{NAME}: Error during face enhancement: {e}")
+        return temp_frame
+
+
+def process_frame(source_face: Face | None, temp_frame: Frame) -> Frame:
+    target_face = get_one_face(temp_frame)
+    if target_face is None:
+        return temp_frame
+    return enhance_face(temp_frame, target_face)
+
+
+def process_frame_v2(temp_frame: Frame) -> Frame:
+    target_face = get_one_face(temp_frame)
+    if target_face:
+        temp_frame = enhance_face(temp_frame, target_face)
+    return temp_frame
+
+
+def process_frames(
+    source_path: str | None, temp_frame_paths: List[str], progress: Any = None
+) -> None:
+    for temp_frame_path in temp_frame_paths:
+        temp_frame = cv2.imread(temp_frame_path)
+        if temp_frame is None:
+            if progress:
+                progress.update(1)
+            continue
+        result = process_frame(None, temp_frame)
+        cv2.imwrite(temp_frame_path, result)
+        if progress:
+            progress.update(1)
+
+
+def process_image(source_path: str | None, target_path: str, output_path: str) -> None:
+    target_frame = cv2.imread(target_path)
+    if target_frame is None:
+        print(f"{NAME}: Error: Failed to read target image {target_path}")
+        return
+    result_frame = process_frame(None, target_frame)
+    cv2.imwrite(output_path, result_frame)
+    print(f"{NAME}: Enhanced image saved to {output_path}")
+
+
+def process_video(source_path: str | None, temp_frame_paths: List[str]) -> None:
+    modules.processors.frame.core.process_video(source_path, temp_frame_paths, process_frames)

modules/processors/frame/face_masking.py

@@ -6,24 +6,31 @@ from modules.gpu_processing import gpu_gaussian_blur, gpu_resize, gpu_cvt_color
 
 def apply_color_transfer(source, target):
     """
-    Apply color transfer from target to source image
+    Apply color transfer from target to source image using LAB color space.
+    Uses float32 throughout for performance (sufficient precision for 8-bit images).
     """
-    source = cv2.cvtColor(source, cv2.COLOR_BGR2LAB).astype("float32")
-    target = cv2.cvtColor(target, cv2.COLOR_BGR2LAB).astype("float32")
+    # Convert to float32 [0,1] range for proper LAB conversion
+    source_f32 = source.astype(np.float32) / 255.0
+    target_f32 = target.astype(np.float32) / 255.0
 
-    source_mean, source_std = cv2.meanStdDev(source)
-    target_mean, target_std = cv2.meanStdDev(target)
+    source_lab = cv2.cvtColor(source_f32, cv2.COLOR_BGR2LAB)
+    target_lab = cv2.cvtColor(target_f32, cv2.COLOR_BGR2LAB)
 
-    # 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)
+    source_mean, source_std = cv2.meanStdDev(source_lab)
+    target_mean, target_std = cv2.meanStdDev(target_lab)
 
-    # Perform the color transfer
-    source = (source - source_mean) * (target_std / source_std) + target_mean
+    # Reshape mean and std to be broadcastable (already float64 from meanStdDev, cast to f32)
+    source_mean = source_mean.reshape(1, 1, 3).astype(np.float32)
+    source_std = np.maximum(source_std.reshape(1, 1, 3), 1e-6).astype(np.float32)
+    target_mean = target_mean.reshape(1, 1, 3).astype(np.float32)
+    target_std = target_std.reshape(1, 1, 3).astype(np.float32)
 
-    return cv2.cvtColor(np.clip(source, 0, 255).astype("uint8"), cv2.COLOR_LAB2BGR)
+    # Perform the color transfer in LAB space
+    result_lab = (source_lab - source_mean) * (target_std / source_std) + target_mean
+
+    # Convert back to BGR and uint8
+    result_bgr = cv2.cvtColor(result_lab, cv2.COLOR_LAB2BGR)
+    return np.clip(result_bgr * 255.0, 0, 255).astype(np.uint8)
 
 def create_face_mask(face: Face, frame: Frame) -> np.ndarray:
     mask = np.zeros(frame.shape[:2], dtype=np.uint8)
@@ -48,16 +55,14 @@ def create_face_mask(face: Face, frame: Frame) -> np.ndarray:
         # Create a slightly larger convex hull for padding
         face_outline = landmarks[0:33]
         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)
+        # Vectorized hull padding — expand each point outward from center
+        center = np.mean(face_outline, axis=0, dtype=np.float32)
+        hull_pts = hull.reshape(-1, 2).astype(np.float32)
+        directions = hull_pts - center
+        norms = np.linalg.norm(directions, axis=1, keepdims=True)
+        norms = np.maximum(norms, 1e-6)  # avoid division by zero
+        directions /= norms
+        hull_padded = (hull_pts + directions * padding).astype(np.int32)
 
         # Fill the padded convex hull
         cv2.fillConvexPoly(mask, hull_padded, 255)
@@ -77,8 +82,8 @@ def create_lower_mouth_mask(
 
     landmarks = face.landmark_2d_106
     if landmarks is not None:
-        # Use outer mouth landmarks (52-63) to capture the lips only
-        lower_lip_order = list(range(52, 64))
+        # Use outer mouth landmarks (52-71) to capture the full mouth area
+        lower_lip_order = list(range(52, 72))
         
         if max(lower_lip_order) >= landmarks.shape[0]:
             return mask, mouth_cutout, mouth_box, lower_lip_polygon
@@ -89,13 +94,16 @@ def create_lower_mouth_mask(
         center = np.mean(lower_lip_landmarks, axis=0)
 
         # Expand the landmarks outward using the mouth_mask_size
-        # Use a more conservative expansion to avoid affecting face shape
-        expansion_factor = (
-            1 + modules.globals.mask_down_size * modules.globals.mouth_mask_size
-        )
-        expanded_landmarks = (lower_lip_landmarks - center) * expansion_factor + center
+        mouth_mask_size = getattr(modules.globals, "mouth_mask_size", 0.0) # 0-100 slider
+        expansion_factor = 1 + (mouth_mask_size / 100.0) * 2.5
 
-        # Removed specific top/chin extensions to preserve face shape
+        # Expand with extra downward bias toward chin
+        offsets = lower_lip_landmarks - center
+        chin_bias = 1 + (mouth_mask_size / 100.0) * 1.5
+        scale_y = np.where(offsets[:, 1] > 0, expansion_factor * chin_bias, expansion_factor)
+        expanded_landmarks = lower_lip_landmarks.copy()
+        expanded_landmarks[:, 0] = center[0] + offsets[:, 0] * expansion_factor
+        expanded_landmarks[:, 1] = center[1] + offsets[:, 1] * scale_y
 
         # Convert back to integer coordinates
         expanded_landmarks = expanded_landmarks.astype(np.int32)
@@ -468,26 +476,28 @@ def apply_mask_area(
             box_height // modules.globals.mask_feather_ratio,
         )
         feathered_mask = cv2.GaussianBlur(
-            polygon_mask.astype(float), (0, 0), feather_amount
+            polygon_mask.astype(np.float32), (0, 0), feather_amount
         )
-        feathered_mask = feathered_mask / feathered_mask.max()
+        max_val = feathered_mask.max()
+        if max_val > 1e-6:
+            feathered_mask *= np.float32(1.0 / max_val)
 
         # Apply additional smoothing to the mask edges
         feathered_mask = cv2.GaussianBlur(feathered_mask, (5, 5), 1)
 
         face_mask_roi = face_mask[min_y:max_y, min_x:max_x]
-        combined_mask = feathered_mask * (face_mask_roi / 255.0)
+        combined_mask = feathered_mask * (face_mask_roi.astype(np.float32) * np.float32(1.0 / 255.0))
 
-        combined_mask = combined_mask[:, :, np.newaxis]
+        combined_mask_3ch = combined_mask[:, :, np.newaxis]
+        inv_mask = np.float32(1.0) - combined_mask_3ch
         blended = (
-            color_corrected_area * combined_mask + roi * (1 - combined_mask)
+            color_corrected_area * combined_mask_3ch + roi * inv_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
-        )
-        final_blend = blended * face_mask_3channel + roi * (1 - face_mask_3channel)
+        face_mask_f32 = face_mask_roi[:, :, np.newaxis].astype(np.float32) * np.float32(1.0 / 255.0)
+        face_mask_3channel = np.broadcast_to(face_mask_f32, blended.shape)
+        final_blend = blended * face_mask_3channel + roi * (np.float32(1.0) - face_mask_3channel)
 
         frame[min_y:max_y, min_x:max_x] = final_blend.astype(np.uint8)
     except Exception as e:

modules/processors/frame/face_swapper.py

@@ -136,8 +136,12 @@ def swap_face(source_face: Face, target_face: Face, temp_frame: Frame) -> Frame:
     if not hasattr(source_face, 'normed_embedding') or source_face.normed_embedding is None:
         return temp_frame
 
-    # Store a copy of the original frame before swapping for opacity blending
-    original_frame = temp_frame.copy()
+    # Store a copy of the original frame before swapping for opacity blending and mouth mask
+    opacity = getattr(modules.globals, "opacity", 1.0)
+    opacity = max(0.0, min(1.0, opacity))
+    mouth_mask_enabled = getattr(modules.globals, "mouth_mask", False)
+    # Always copy if mouth mask is enabled (we need the unmodified original for mouth cutout)
+    original_frame = temp_frame.copy() if (opacity < 1.0 or mouth_mask_enabled) else temp_frame
 
     # Pre-swap Input Check with optimization
     if temp_frame.dtype != np.uint8:
@@ -188,28 +192,28 @@ def swap_face(source_face: Face, target_face: Face, temp_frame: Frame) -> Frame:
     # --- Post-swap Processing (Masking, Opacity, etc.) ---
     # Now, work with the guaranteed uint8 'swapped_frame'
 
-    if getattr(modules.globals, "mouth_mask", False): # Check if mouth_mask is enabled
+    if mouth_mask_enabled: # Check if mouth_mask is enabled
         # Create a mask for the target face
-        face_mask = create_face_mask(target_face, temp_frame) # Use temp_frame (original shape) for mask creation geometry
+        face_mask = create_face_mask(target_face, original_frame) # Use original_frame for mask creation geometry
 
-        # Create the mouth mask using original geometry
+        # Create the mouth mask using the ORIGINAL frame (before swap) for cutout
         mouth_mask, mouth_cutout, mouth_box, lower_lip_polygon = (
-            create_lower_mouth_mask(target_face, temp_frame) # Use temp_frame (original) for cutout
+            create_lower_mouth_mask(target_face, original_frame) # Use original_frame for real mouth cutout
         )
 
         # Apply the mouth area only if mouth_cutout exists
-        if mouth_cutout is not None and mouth_box != (0,0,0,0): # Add check for valid box
-             # Apply mouth area (from original) onto the 'swapped_frame'
+        if mouth_cutout is not None and mouth_box != (0,0,0,0):
+            # Apply mouth area (from original) onto the 'swapped_frame'
             swapped_frame = apply_mouth_area(
                 swapped_frame, mouth_cutout, mouth_box, face_mask, lower_lip_polygon
             )
 
+            # Draw bounding box only while slider is being dragged
             if getattr(modules.globals, "show_mouth_mask_box", False):
-                        mouth_mask_data = (mouth_mask, mouth_cutout, mouth_box, lower_lip_polygon)
-                        # Draw visualization on the swapped_frame *before* opacity blending
-                        swapped_frame = draw_mouth_mask_visualization(
-                            swapped_frame, target_face, mouth_mask_data
-                        )
+                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
+                )
         
     # --- Poisson Blending ---
     if getattr(modules.globals, "poisson_blend", False):
@@ -240,19 +244,13 @@ def swap_face(source_face: Face, target_face: Face, temp_frame: Frame) -> Frame:
                 except Exception as e:
                     print(f"Poisson blending failed: {e}")
         
-            # Apply opacity blend between the original frame and the swapped frame
-    opacity = getattr(modules.globals, "opacity", 1.0)
-    # Ensure opacity is within valid range [0.0, 1.0]
-    opacity = max(0.0, min(1.0, opacity))
+    # Apply opacity blend between the original frame and the swapped frame
+    if opacity >= 1.0:
+        return swapped_frame.astype(np.uint8)
 
     # Blend the original_frame with the (potentially mouth-masked) swapped_frame
-    # Ensure both frames are uint8 before blending
     final_swapped_frame = gpu_add_weighted(original_frame.astype(np.uint8), 1 - opacity, swapped_frame.astype(np.uint8), opacity, 0)
-
-    # Ensure final frame is uint8 after blending (addWeighted should preserve it, but belt-and-suspenders)
-    final_swapped_frame = final_swapped_frame.astype(np.uint8)
-
-    return final_swapped_frame
+    return final_swapped_frame.astype(np.uint8)
 
 
 # --- START: Mac M1-M5 Optimized Face Detection ---
@@ -363,10 +361,8 @@ def apply_post_processing(current_frame: Frame, swapped_face_bboxes: List[np.nda
                 pass
             PREVIOUS_FRAME_RESULT = processed_frame.copy()
     else:
-         # If interpolation is off or weight is invalid, just use the current frame
-         # Update state with the current (potentially sharpened) frame
-         # Reset previous frame state if interpolation was just turned off or weight is invalid
-         PREVIOUS_FRAME_RESULT = processed_frame.copy()
+         # Interpolation is off or weight is invalid — no need to cache
+         PREVIOUS_FRAME_RESULT = None
 
 
     return final_frame
@@ -756,9 +752,9 @@ def create_lower_mouth_mask(
         return mask, mouth_cutout, mouth_box, lower_lip_polygon
 
     try: # Wrap main logic in try-except
-        # Use outer mouth landmarks (52-63) to capture the lips only
-        # This avoids including the chin/jawline, preserving the face shape from the swap
-        lower_lip_order = list(range(52, 64))
+        # Use outer mouth landmarks (52-71) to capture the full mouth area
+        # This covers both upper and lower lips for proper mouth preservation
+        lower_lip_order = list(range(52, 72))
 
         # Check if all indices are valid for the loaded landmarks (already partially done by < 106 check)
         if max(lower_lip_order) >= landmarks.shape[0]:
@@ -778,9 +774,18 @@ def create_lower_mouth_mask(
             return mask, mouth_cutout, mouth_box, lower_lip_polygon
 
 
-        mask_down_size = getattr(modules.globals, "mask_down_size", 0.1) # Default 0.1
-        expansion_factor = 1 + mask_down_size
-        expanded_landmarks = (lower_lip_landmarks - center) * expansion_factor + center
+        mouth_mask_size = getattr(modules.globals, "mouth_mask_size", 0.0) # 0-100 slider
+        # 0=tight lip outline, 50=covers mouth area, 100=mouth to chin
+        expansion_factor = 1 + (mouth_mask_size / 100.0) * 2.5
+
+        # Expand landmarks from center, with extra downward bias toward chin
+        offsets = lower_lip_landmarks - center
+        # Add extra downward expansion for points below center (toward chin)
+        chin_bias = 1 + (mouth_mask_size / 100.0) * 1.5  # extra vertical stretch downward
+        scale_y = np.where(offsets[:, 1] > 0, expansion_factor * chin_bias, expansion_factor)
+        expanded_landmarks = lower_lip_landmarks.copy()
+        expanded_landmarks[:, 0] = center[0] + offsets[:, 0] * expansion_factor
+        expanded_landmarks[:, 1] = center[1] + offsets[:, 1] * scale_y
 
         # Ensure landmarks are finite after adjustments
         if not np.all(np.isfinite(expanded_landmarks)):
@@ -887,8 +892,8 @@ def draw_mouth_mask_visualization(
         print(f"Error drawing polygon for visualization: {e}") # Optional debug
         pass
 
-    # Optional: Draw bounding box (red rectangle)
-    # cv2.rectangle(vis_frame, (min_x, min_y), (max_x, max_y), (0, 0, 255), 1)
+    # Draw bounding box (red rectangle)
+    cv2.rectangle(vis_frame, (min_x, min_y), (max_x, max_y), (0, 0, 255), 2)
 
     # Optional: Add labels
     label_pos_y = min_y - 10 if min_y > 20 else max_y + 15 # Adjust position based on box location
@@ -968,85 +973,34 @@ def apply_mouth_area(
             # print("Warning: Mouth cutout is invalid after resize attempt.")
             return frame
 
-        # --- Color Correction Step ---
-        # Apply color transfer from ROI (swapped face region) to the original mouth cutout
-        # This helps match lighting/color before blending
-        color_corrected_mouth = resized_mouth_cutout # Default to resized if correction fails
-        try:
-           # Ensure both images are 3 channels for color transfer
-           if len(resized_mouth_cutout.shape) == 3 and resized_mouth_cutout.shape[2] == 3 and \
-              len(roi.shape) == 3 and roi.shape[2] == 3:
-                 color_corrected_mouth = apply_color_transfer(resized_mouth_cutout, roi)
-           else:
-               # print("Warning: Cannot apply color transfer, images not BGR.")
-               pass
-        except cv2.error as ct_e: # Handle potential errors in color transfer
-           # print(f"Warning: Color transfer failed: {ct_e}. Using uncorrected mouth cutout.") # Optional debug
-           pass
-        except Exception as ct_gen_e:
-           # print(f"Warning: Unexpected error during color transfer: {ct_gen_e}")
-           pass
-        # --- End Color Correction ---
-
-
         # --- Mask Creation ---
-        # Create a mask based *specifically* on the mouth_polygon, relative to the ROI
+        # Create a mask based on the mouth_polygon, relative to the ROI
         polygon_mask_roi = np.zeros(roi.shape[:2], dtype=np.uint8)
-        # Adjust polygon coordinates relative to the ROI's top-left corner
         adjusted_polygon = mouth_polygon - [min_x, min_y]
-        # Draw the filled polygon on the ROI mask
         cv2.fillPoly(polygon_mask_roi, [adjusted_polygon.astype(np.int32)], 255)
 
-        # Feather the polygon mask (Gaussian blur)
-        mask_feather_ratio = getattr(modules.globals, "mask_feather_ratio", 12) # Default 12
-        # Calculate feather amount based on the smaller dimension of the box
-        feather_base_dim = min(box_width, box_height)
-        feather_amount = max(1, min(30, feather_base_dim // max(1, mask_feather_ratio))) # Avoid div by zero
-        # Ensure kernel size is odd and positive
+        # Feather the edges with Gaussian blur for smooth blending
+        feather_amount = max(1, min(30, min(box_width, box_height) // 8))
         kernel_size = 2 * feather_amount + 1
-        feathered_polygon_mask = cv2.GaussianBlur(polygon_mask_roi.astype(float), (kernel_size, kernel_size), 0)
+        feathered_mask = cv2.GaussianBlur(polygon_mask_roi.astype(np.float32), (kernel_size, kernel_size), 0)
 
-        # Normalize feathered mask to [0.0, 1.0] range
-        max_val = feathered_polygon_mask.max()
-        if max_val > 1e-6: # Avoid division by zero
-           feathered_polygon_mask = feathered_polygon_mask / max_val
+        # Normalize to [0.0, 1.0]
+        max_val = feathered_mask.max()
+        if max_val > 1e-6:
+            feathered_mask = feathered_mask / max_val
         else:
-           feathered_polygon_mask.fill(0.0) # Mask is all black if max is near zero
-        # --- End Mask Creation ---
+            feathered_mask.fill(0.0)
 
-
-        # --- Refined Blending ---
-        # Get the corresponding ROI from the *full face mask* (already blurred)
-        # Ensure face_mask is float and normalized [0.0, 1.0]
-        if face_mask.dtype != np.float64 and face_mask.dtype != np.float32:
-            face_mask_float = face_mask.astype(float) / 255.0
-        else: # Assume already float [0,1] if type is float
-            face_mask_float = face_mask
-        face_mask_roi = face_mask_float[min_y:max_y, min_x:max_x]
-
-        # Combine the feathered mouth polygon mask with the face mask ROI
-        # Use minimum to ensure we only affect area inside both masks (mouth area within face)
-        # This helps blend the edges smoothly with the surrounding swapped face region
-        combined_mask = np.minimum(feathered_polygon_mask, face_mask_roi)
-
-        # Expand mask to 3 channels for blending (ensure it matches image channels)
+        # --- Blending: paste original mouth onto swapped face ---
         if len(frame.shape) == 3 and frame.shape[2] == 3:
-            combined_mask_3channel = combined_mask[:, :, np.newaxis]
+            mask_3ch = feathered_mask[:, :, np.newaxis].astype(np.float32)
+            inv_mask = 1.0 - mask_3ch
 
-            # Ensure data types are compatible for blending (float or double for mask, uint8 for images)
-            color_corrected_mouth_uint8 = color_corrected_mouth.astype(np.uint8)
-            roi_uint8 = roi.astype(np.uint8)
-            combined_mask_float = combined_mask_3channel.astype(np.float64) # Use float64 for precision in mask
+            # Blend: (original_mouth * mask) + (swapped_face * (1 - mask))
+            blended_roi = (resized_mouth_cutout.astype(np.float32) * mask_3ch +
+                           roi.astype(np.float32) * inv_mask)
 
-            # Blend: (original_mouth * combined_mask) + (swapped_face_roi * (1 - combined_mask))
-            blended_roi = (color_corrected_mouth_uint8 * combined_mask_float +
-                           roi_uint8 * (1.0 - combined_mask_float))
-
-            # Place the blended ROI back into the frame
-            frame[min_y:max_y, min_x:max_x] = blended_roi.astype(np.uint8)
-        else:
-            # print("Warning: Cannot apply mouth mask blending, frame is not 3-channel BGR.")
-            pass # Don't modify frame if it's not BGR
+            frame[min_y:max_y, min_x:max_x] = np.clip(blended_roi, 0, 255).astype(np.uint8)
 
     except Exception as e:
         print(f"Error applying mouth area: {e}") # Optional debug

modules/ui.py

@@ -3,7 +3,6 @@ import webbrowser
 import customtkinter as ctk
 from typing import Callable, Tuple
 import cv2
-from cv2_enumerate_cameras import enumerate_cameras  # Add this import
 from modules.gpu_processing import gpu_cvt_color, gpu_resize, gpu_flip
 from PIL import Image, ImageOps
 import time
@@ -11,6 +10,8 @@ import json
 import queue
 import threading
 import numpy as np
+import requests
+import tempfile
 import modules.globals
 import modules.metadata
 from modules.face_analyser import (
@@ -32,12 +33,36 @@ from modules.utilities import (
 )
 from modules.video_capture import VideoCapturer
 from modules.gettext import LanguageManager
+from modules.ui_tooltip import ToolTip
 from modules import globals
 import platform
 
 if platform.system() == "Windows":
     from pygrabber.dshow_graph import FilterGraph
 
+# --- Tk 9.0 compatibility patch ---
+# In Tk 9.0, Menu.index("end") returns "" instead of raising TclError
+# when the menu is empty. CustomTkinter's CTkOptionMenu doesn't handle
+# this, causing crashes. This patch adds the missing guard.
+try:
+    from customtkinter.windows.widgets.core_widget_classes import DropdownMenu as _DropdownMenu
+
+    _original_add_menu_commands = _DropdownMenu._add_menu_commands
+
+    def _patched_add_menu_commands(self, *args, **kwargs):
+        try:
+            end_index = self._menu.index("end")
+            if end_index == "" or end_index is None:
+                return
+        except Exception:
+            pass
+        _original_add_menu_commands(self, *args, **kwargs)
+
+    _DropdownMenu._add_menu_commands = _patched_add_menu_commands
+except (ImportError, AttributeError):
+    pass  # CustomTkinter version doesn't have this class path
+# --- End Tk 9.0 patch ---
+
 ROOT = None
 POPUP = None
 POPUP_LIVE = None
@@ -112,6 +137,7 @@ def save_switch_states():
         "show_fps": modules.globals.show_fps,
         "mouth_mask": modules.globals.mouth_mask,
         "show_mouth_mask_box": modules.globals.show_mouth_mask_box,
+        "mouth_mask_size": modules.globals.mouth_mask_size,
     }
     with open("switch_states.json", "w") as f:
         json.dump(switch_states, f)
@@ -133,10 +159,10 @@ def load_switch_states():
         modules.globals.live_resizable = switch_states.get("live_resizable", False)
         modules.globals.fp_ui = switch_states.get("fp_ui", {"face_enhancer": False})
         modules.globals.show_fps = switch_states.get("show_fps", False)
-        modules.globals.mouth_mask = switch_states.get("mouth_mask", False)
-        modules.globals.show_mouth_mask_box = switch_states.get(
-            "show_mouth_mask_box", False
-        )
+        modules.globals.mouth_mask_size = switch_states.get("mouth_mask_size", 0.0)
+        # mouth_mask is driven by the slider: on if size > 0, off if 0
+        modules.globals.mouth_mask = modules.globals.mouth_mask_size > 0
+        modules.globals.show_mouth_mask_box = False  # always start hidden
     except FileNotFoundError:
         # If the file doesn't exist, use default values
         pass
@@ -168,12 +194,20 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
     select_face_button = ctk.CTkButton(
         root, text=_("Select a face"), cursor="hand2", command=lambda: select_source_path()
     )
-    select_face_button.place(relx=0.1, rely=0.30, relwidth=0.3, relheight=0.1)
+    select_face_button.place(relx=0.1, rely=0.30, relwidth=0.24, relheight=0.1)
+    ToolTip(select_face_button, _("Choose the source face image to swap onto the target"))
+
+    random_face_button = ctk.CTkButton(
+        root, text="🔄", cursor="hand2", width=30, command=lambda: fetch_random_face()
+    )
+    random_face_button.place(relx=0.35, rely=0.30, relwidth=0.05, relheight=0.1)
+    ToolTip(random_face_button, _("Get a random face from thispersondoesnotexist.com"))
 
     swap_faces_button = ctk.CTkButton(
         root, text="↔", cursor="hand2", command=lambda: swap_faces_paths()
     )
     swap_faces_button.place(relx=0.45, rely=0.30, relwidth=0.1, relheight=0.1)
+    ToolTip(swap_faces_button, _("Swap source and target images"))
 
     select_target_button = ctk.CTkButton(
         root,
@@ -182,6 +216,7 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
         command=lambda: select_target_path(),
     )
     select_target_button.place(relx=0.6, rely=0.30, relwidth=0.3, relheight=0.1)
+    ToolTip(select_target_button, _("Choose the target image or video to apply face swap to"))
 
     keep_fps_value = ctk.BooleanVar(value=modules.globals.keep_fps)
     keep_fps_checkbox = ctk.CTkSwitch(
@@ -194,7 +229,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             save_switch_states(),
         ),
     )
-    keep_fps_checkbox.place(relx=0.1, rely=0.5)
+    keep_fps_checkbox.place(relx=0.1, rely=0.42)
+    ToolTip(keep_fps_checkbox, _("Output video keeps the original frame rate"))
 
     keep_frames_value = ctk.BooleanVar(value=modules.globals.keep_frames)
     keep_frames_switch = ctk.CTkSwitch(
@@ -207,20 +243,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             save_switch_states(),
         ),
     )
-    keep_frames_switch.place(relx=0.1, rely=0.55)
-
-    enhancer_value = ctk.BooleanVar(value=modules.globals.fp_ui["face_enhancer"])
-    enhancer_switch = ctk.CTkSwitch(
-        root,
-        text=_("Face Enhancer"),
-        variable=enhancer_value,
-        cursor="hand2",
-        command=lambda: (
-            update_tumbler("face_enhancer", enhancer_value.get()),
-            save_switch_states(),
-        ),
-    )
-    enhancer_switch.place(relx=0.1, rely=0.6)
+    keep_frames_switch.place(relx=0.1, rely=0.47)
+    ToolTip(keep_frames_switch, _("Keep extracted frames on disk after processing"))
 
     keep_audio_value = ctk.BooleanVar(value=modules.globals.keep_audio)
     keep_audio_switch = ctk.CTkSwitch(
@@ -233,7 +257,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             save_switch_states(),
         ),
     )
-    keep_audio_switch.place(relx=0.6, rely=0.5)
+    keep_audio_switch.place(relx=0.6, rely=0.42)
+    ToolTip(keep_audio_switch, _("Copy audio track from the source video to output"))
 
     many_faces_value = ctk.BooleanVar(value=modules.globals.many_faces)
     many_faces_switch = ctk.CTkSwitch(
@@ -246,7 +271,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             save_switch_states(),
         ),
     )
-    many_faces_switch.place(relx=0.6, rely=0.55)
+    many_faces_switch.place(relx=0.6, rely=0.47)
+    ToolTip(many_faces_switch, _("Swap every detected face, not just the primary one"))
 
     color_correction_value = ctk.BooleanVar(value=modules.globals.color_correction)
     color_correction_switch = ctk.CTkSwitch(
@@ -259,7 +285,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             save_switch_states(),
         ),
     )
-    color_correction_switch.place(relx=0.6, rely=0.6)
+    color_correction_switch.place(relx=0.6, rely=0.57)
+    ToolTip(color_correction_switch, _("Fix blue/green color cast from some webcams"))
 
     #    nsfw_value = ctk.BooleanVar(value=modules.globals.nsfw_filter)
     #    nsfw_switch = ctk.CTkSwitch(root, text='NSFW filter', variable=nsfw_value, cursor='hand2', command=lambda: setattr(modules.globals, 'nsfw_filter', nsfw_value.get()))
@@ -277,7 +304,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             close_mapper_window() if not map_faces.get() else None
         ),
     )
-    map_faces_switch.place(relx=0.1, rely=0.65)
+    map_faces_switch.place(relx=0.1, rely=0.52)
+    ToolTip(map_faces_switch, _("Manually assign which source face maps to which target face"))
 
     poisson_blend_value = ctk.BooleanVar(value=modules.globals.poisson_blend)
     poisson_blend_switch = ctk.CTkSwitch(
@@ -290,7 +318,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             save_switch_states(),
         ),
     )
-    poisson_blend_switch.place(relx=0.1, rely=0.7)
+    poisson_blend_switch.place(relx=0.1, rely=0.57)
+    ToolTip(poisson_blend_switch, _("Blend face edges smoothly using Poisson blending"))
 
     show_fps_value = ctk.BooleanVar(value=modules.globals.show_fps)
     show_fps_switch = ctk.CTkSwitch(
@@ -303,48 +332,34 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             save_switch_states(),
         ),
     )
-    show_fps_switch.place(relx=0.6, rely=0.65)
+    show_fps_switch.place(relx=0.6, rely=0.52)
+    ToolTip(show_fps_switch, _("Display frames-per-second counter on the live preview"))
 
+    # mouth_mask and show_mouth_mask_box are auto-controlled by the Mouth Mask slider
     mouth_mask_var = ctk.BooleanVar(value=modules.globals.mouth_mask)
-    mouth_mask_switch = ctk.CTkSwitch(
-        root,
-        text=_("Mouth Mask"),
-        variable=mouth_mask_var,
-        cursor="hand2",
-        command=lambda: setattr(modules.globals, "mouth_mask", mouth_mask_var.get()),
-    )
-    mouth_mask_switch.place(relx=0.1, rely=0.45)
-
     show_mouth_mask_box_var = ctk.BooleanVar(value=modules.globals.show_mouth_mask_box)
-    show_mouth_mask_box_switch = ctk.CTkSwitch(
-        root,
-        text=_("Show Mouth Mask Box"),
-        variable=show_mouth_mask_box_var,
-        cursor="hand2",
-        command=lambda: setattr(
-            modules.globals, "show_mouth_mask_box", show_mouth_mask_box_var.get()
-        ),
-    )
-    show_mouth_mask_box_switch.place(relx=0.6, rely=0.45)
 
     start_button = ctk.CTkButton(
         root, text=_("Start"), cursor="hand2", command=lambda: analyze_target(start, root)
     )
-    start_button.place(relx=0.15, rely=0.86, relwidth=0.2, relheight=0.05)
+    start_button.place(relx=0.15, rely=0.78, relwidth=0.2, relheight=0.04)
+    ToolTip(start_button, _("Begin processing the target image/video with selected face"))
 
     stop_button = ctk.CTkButton(
         root, text=_("Destroy"), cursor="hand2", command=lambda: destroy()
     )
-    stop_button.place(relx=0.4, rely=0.86, relwidth=0.2, relheight=0.05)
+    stop_button.place(relx=0.4, rely=0.78, relwidth=0.2, relheight=0.04)
+    ToolTip(stop_button, _("Stop processing and close the application"))
 
     preview_button = ctk.CTkButton(
         root, text=_("Preview"), cursor="hand2", command=lambda: toggle_preview()
     )
-    preview_button.place(relx=0.65, rely=0.86, relwidth=0.2, relheight=0.05)
+    preview_button.place(relx=0.65, rely=0.78, relwidth=0.2, relheight=0.04)
+    ToolTip(preview_button, _("Show/hide a preview of the processed output"))
 
     # --- Camera Selection ---
     camera_label = ctk.CTkLabel(root, text=_("Select Camera:"))
-    camera_label.place(relx=0.1, rely=0.92, relwidth=0.2, relheight=0.05)
+    camera_label.place(relx=0.1, rely=0.83, relwidth=0.2, relheight=0.03)
 
     available_cameras = get_available_cameras()
     camera_indices, camera_names = available_cameras
@@ -363,7 +378,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             root, variable=camera_variable, values=camera_names
         )
 
-    camera_optionmenu.place(relx=0.35, rely=0.92, relwidth=0.25, relheight=0.05)
+    camera_optionmenu.place(relx=0.35, rely=0.83, relwidth=0.25, relheight=0.03)
+    ToolTip(camera_optionmenu, _("Select which camera to use for live mode"))
 
     live_button = ctk.CTkButton(
         root,
@@ -383,9 +399,52 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             else "disabled"
         ),
     )
-    live_button.place(relx=0.65, rely=0.92, relwidth=0.2, relheight=0.05)
+    live_button.place(relx=0.65, rely=0.83, relwidth=0.2, relheight=0.03)
+    ToolTip(live_button, _("Start real-time face swap using webcam"))
     # --- End Camera Selection ---
 
+    # --- Face Enhancer Dropdown ---
+    enhancer_options = ["None", "GFPGAN", "GPEN-512", "GPEN-256"]
+    enhancer_key_map = {
+        "None": None,
+        "GFPGAN": "face_enhancer",
+        "GPEN-512": "face_enhancer_gpen512",
+        "GPEN-256": "face_enhancer_gpen256",
+    }
+
+    # Determine initial value from current fp_ui state
+    initial_enhancer = "None"
+    if modules.globals.fp_ui.get("face_enhancer", False):
+        initial_enhancer = "GFPGAN"
+    elif modules.globals.fp_ui.get("face_enhancer_gpen512", False):
+        initial_enhancer = "GPEN-512"
+    elif modules.globals.fp_ui.get("face_enhancer_gpen256", False):
+        initial_enhancer = "GPEN-256"
+
+    enhancer_variable = ctk.StringVar(value=initial_enhancer)
+
+    def on_enhancer_change(choice: str):
+        # Disable all enhancers first
+        for key in ["face_enhancer", "face_enhancer_gpen256", "face_enhancer_gpen512"]:
+            update_tumbler(key, False)
+        # Enable the selected one
+        selected_key = enhancer_key_map.get(choice)
+        if selected_key:
+            update_tumbler(selected_key, True)
+        save_switch_states()
+
+    enhancer_label = ctk.CTkLabel(root, text="Face Enhancer:")
+    enhancer_label.place(relx=0.1, rely=0.62, relwidth=0.2, relheight=0.03)
+
+    enhancer_dropdown = ctk.CTkOptionMenu(
+        root,
+        variable=enhancer_variable,
+        values=enhancer_options,
+        command=on_enhancer_change,
+    )
+    enhancer_dropdown.place(relx=0.35, rely=0.62, relwidth=0.3, relheight=0.03)
+    ToolTip(enhancer_dropdown, _("Select a face enhancement model (None = no enhancement)"))
+
     # 1) Define a DoubleVar for transparency (0 = fully transparent, 1 = fully opaque)
     transparency_var = ctk.DoubleVar(value=1.0)
 
@@ -405,9 +464,9 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
             modules.globals.face_swapper_enabled = True
             update_status(f"Transparency set to {percentage}%")
 
-    # 2) Transparency label and slider (placed ABOVE sharpness)
+    # 2) Transparency label and slider
     transparency_label = ctk.CTkLabel(root, text="Transparency:")
-    transparency_label.place(relx=0.15, rely=0.75, relwidth=0.2, relheight=0.05)
+    transparency_label.place(relx=0.15, rely=0.66, relwidth=0.2, relheight=0.03)
 
     transparency_slider = ctk.CTkSlider(
         root,
@@ -423,7 +482,8 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
         border_width=1,
         corner_radius=3,
     )
-    transparency_slider.place(relx=0.35, rely=0.77, relwidth=0.5, relheight=0.02)
+    transparency_slider.place(relx=0.35, rely=0.67, relwidth=0.5, relheight=0.02)
+    ToolTip(transparency_slider, _("Blend between original and swapped face (0% = original, 100% = fully swapped)"))
 
     # 3) Sharpness label & slider
     sharpness_var = ctk.DoubleVar(value=0.0)  # start at 0.0
@@ -432,7 +492,7 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
         update_status(f"Sharpness set to {value:.1f}")
 
     sharpness_label = ctk.CTkLabel(root, text="Sharpness:")
-    sharpness_label.place(relx=0.15, rely=0.80, relwidth=0.2, relheight=0.05)
+    sharpness_label.place(relx=0.15, rely=0.69, relwidth=0.2, relheight=0.03)
 
     sharpness_slider = ctk.CTkSlider(
         root,
@@ -448,17 +508,64 @@ def create_root(start: Callable[[], None], destroy: Callable[[], None]) -> ctk.C
         border_width=1,
         corner_radius=3,
     )
-    sharpness_slider.place(relx=0.35, rely=0.82, relwidth=0.5, relheight=0.02)
+    sharpness_slider.place(relx=0.35, rely=0.70, relwidth=0.5, relheight=0.02)
+    ToolTip(sharpness_slider, _("Sharpen the enhanced face output"))
+
+    # 4) Mouth Mask Size slider
+    mouth_mask_size_var = ctk.DoubleVar(value=modules.globals.mouth_mask_size)
+
+    def on_mouth_mask_size_change(value: float):
+        val = float(value)
+        modules.globals.mouth_mask_size = val
+        # Auto-enable/disable mouth mask based on slider position
+        if val > 0:
+            modules.globals.mouth_mask = True
+            mouth_mask_var.set(True)
+        else:
+            modules.globals.mouth_mask = False
+            mouth_mask_var.set(False)
+            modules.globals.show_mouth_mask_box = False
+
+    def on_mouth_mask_slider_release(event):
+        # Hide bounding box when user releases the slider
+        modules.globals.show_mouth_mask_box = False
+
+    def on_mouth_mask_slider_press(event):
+        # Show bounding box while dragging
+        if modules.globals.mouth_mask_size > 0:
+            modules.globals.show_mouth_mask_box = True
+
+    mouth_mask_size_label = ctk.CTkLabel(root, text="Mouth Mask:")
+    mouth_mask_size_label.place(relx=0.15, rely=0.72, relwidth=0.2, relheight=0.03)
+
+    mouth_mask_size_slider = ctk.CTkSlider(
+        root,
+        from_=0.0,
+        to=100.0,
+        variable=mouth_mask_size_var,
+        command=on_mouth_mask_size_change,
+        fg_color="#E0E0E0",
+        progress_color="#007BFF",
+        button_color="#FFFFFF",
+        button_hover_color="#CCCCCC",
+        height=5,
+        border_width=1,
+        corner_radius=3,
+    )
+    mouth_mask_size_slider.place(relx=0.35, rely=0.73, relwidth=0.5, relheight=0.02)
+    mouth_mask_size_slider.bind("<ButtonPress-1>", on_mouth_mask_slider_press)
+    mouth_mask_size_slider.bind("<ButtonRelease-1>", on_mouth_mask_slider_release)
+    ToolTip(mouth_mask_size_slider, _("0 = use swapped mouth, 100 = expose original mouth to chin area"))
 
     # Status and link at the bottom
     global status_label
     status_label = ctk.CTkLabel(root, text=None, justify="center")
-    status_label.place(relx=0.1, rely=0.96, relwidth=0.8)
+    status_label.place(relx=0.1, rely=0.75, relwidth=0.8)
 
     donate_label = ctk.CTkLabel(
         root, text="Deep Live Cam", justify="center", cursor="hand2"
     )
-    donate_label.place(relx=0.1, rely=0.98, relwidth=0.8)
+    donate_label.place(relx=0.1, rely=0.87, relwidth=0.8)
     donate_label.configure(
         text_color=ctk.ThemeManager.theme.get("URL").get("text_color")
     )
@@ -667,6 +774,26 @@ def update_tumbler(var: str, value: bool) -> None:
         )
 
 
+def fetch_random_face() -> None:
+    PREVIEW.withdraw()
+    try:
+        response = requests.get(
+            "https://thispersondoesnotexist.com/",
+            headers={"User-Agent": "Mozilla/5.0"},
+            timeout=10,
+        )
+        response.raise_for_status()
+        temp_dir = tempfile.gettempdir()
+        temp_path = os.path.join(temp_dir, "deep_live_cam_random_face.jpg")
+        with open(temp_path, "wb") as f:
+            f.write(response.content)
+        modules.globals.source_path = temp_path
+        image = render_image_preview(temp_path, (200, 200))
+        source_label.configure(image=image)
+    except Exception as e:
+        print(f"Failed to fetch random face: {e}")
+
+
 def select_source_path() -> None:
     global RECENT_DIRECTORY_SOURCE, img_ft, vid_ft
 
@@ -922,21 +1049,13 @@ def get_available_cameras():
         camera_indices = []
         camera_names = []
 
-        if platform.system() == "Darwin":  # macOS specific handling
-            # Try to open the default FaceTime camera first
-            cap = cv2.VideoCapture(0)
-            if cap.isOpened():
-                camera_indices.append(0)
-                camera_names.append("FaceTime Camera")
-                cap.release()
-
-            # On macOS, additional cameras typically use indices 1 and 2
-            for i in [1, 2]:
-                cap = cv2.VideoCapture(i)
-                if cap.isOpened():
-                    camera_indices.append(i)
-                    camera_names.append(f"Camera {i}")
-                    cap.release()
+        if platform.system() == "Darwin":
+            # Do NOT probe cameras with cv2.VideoCapture on macOS — probing
+            # invalid indices triggers the OBSENSOR backend and causes SIGSEGV.
+            # Default to indices 0 and 1 (covers FaceTime + one USB camera).
+            # The user can select the correct index from the UI dropdown.
+            camera_indices = [0, 1]
+            camera_names = ["Camera 0", "Camera 1"]
         else:
             # Linux camera detection - test first 10 indices
             for i in range(10):
@@ -974,28 +1093,48 @@ def _capture_thread_func(cap, capture_queue, stop_event):
                 pass
 
 
-# How often to run full face detection. On intermediate frames the last
-# detected face positions are reused, which significantly reduces the
-# per-frame cost of the processing thread.
-DETECT_EVERY_N = 2
+def _detection_thread_func(latest_frame_holder, detection_result, detection_lock, stop_event):
+    """Detection thread: continuously runs face detection on the latest
+    captured frame and stores results in detection_result under detection_lock.
+
+    This decouples face detection (~15-30ms) from face swapping (~5-10ms)
+    so the swap loop never blocks on detection, significantly improving
+    live mode FPS."""
+    while not stop_event.is_set():
+        with detection_lock:
+            frame = latest_frame_holder[0]
+
+        if frame is None:
+            time.sleep(0.005)
+            continue
+
+        if modules.globals.many_faces:
+            many = get_many_faces(frame)
+            with detection_lock:
+                detection_result['target_face'] = None
+                detection_result['many_faces'] = many
+        else:
+            face = get_one_face(frame)
+            with detection_lock:
+                detection_result['target_face'] = face
+                detection_result['many_faces'] = None
 
 
-def _processing_thread_func(capture_queue, processed_queue, stop_event):
-    """Processing thread: takes raw frames from capture_queue, applies face
-    processing, and puts results into processed_queue. Drops processed frames
-    when the output queue is full so the UI always gets the latest result.
+def _processing_thread_func(capture_queue, processed_queue, stop_event,
+                             latest_frame_holder, detection_result, detection_lock):
+    """Processing thread: takes raw frames from capture_queue, reads the
+    latest detection result from the shared detection_result dict, applies
+    face swap/enhancement, and puts results into processed_queue.
 
-    Uses DETECT_EVERY_N to skip expensive face detection on intermediate
-    frames, reusing cached face positions instead."""
+    Face detection runs concurrently in _detection_thread_func — this thread
+    only reads cached results so it never blocks on detection."""
     frame_processors = get_frame_processors_modules(modules.globals.frame_processors)
     source_image = None
+    last_source_path = None
     prev_time = time.time()
     fps_update_interval = 0.5
     frame_count = 0
     fps = 0
-    proc_frame_index = 0
-    cached_target_face = None  # cached single-face result
-    cached_many_faces = None   # cached many-faces result
 
     while not stop_event.is_set():
         try:
@@ -1003,32 +1142,37 @@ def _processing_thread_func(capture_queue, processed_queue, stop_event):
         except queue.Empty:
             continue
 
-        temp_frame = frame.copy()
-        run_detection = (proc_frame_index % DETECT_EVERY_N == 0)
-        proc_frame_index += 1
+        temp_frame = frame
 
         if modules.globals.live_mirror:
             temp_frame = gpu_flip(temp_frame, 1)
 
+        # Publish the mirrored frame for the detection thread to pick up
+        with detection_lock:
+            latest_frame_holder[0] = temp_frame
+
         if not modules.globals.map_faces:
-            if source_image is None and modules.globals.source_path:
+            if modules.globals.source_path and modules.globals.source_path != last_source_path:
+                last_source_path = modules.globals.source_path
                 source_image = get_one_face(cv2.imread(modules.globals.source_path))
 
-            # Update face detection cache on detection frames
-            if run_detection or (cached_target_face is None and cached_many_faces is None):
-                if modules.globals.many_faces:
-                    cached_many_faces = get_many_faces(temp_frame)
-                    cached_target_face = None
-                else:
-                    cached_target_face = get_one_face(temp_frame)
-                    cached_many_faces = None
+            # Read latest detection results (brief lock to avoid blocking detection thread)
+            with detection_lock:
+                cached_target_face = detection_result.get('target_face')
+                cached_many_faces = detection_result.get('many_faces')
 
             for frame_processor in frame_processors:
                 if frame_processor.NAME == "DLC.FACE-ENHANCER":
                     if modules.globals.fp_ui["face_enhancer"]:
                         temp_frame = frame_processor.process_frame(None, temp_frame)
+                elif frame_processor.NAME == "DLC.FACE-ENHANCER-GPEN256":
+                    if modules.globals.fp_ui.get("face_enhancer_gpen256", False):
+                        temp_frame = frame_processor.process_frame(None, temp_frame)
+                elif frame_processor.NAME == "DLC.FACE-ENHANCER-GPEN512":
+                    if modules.globals.fp_ui.get("face_enhancer_gpen512", False):
+                        temp_frame = frame_processor.process_frame(None, temp_frame)
                 elif frame_processor.NAME == "DLC.FACE-SWAPPER":
-                    # Use cached face positions to skip redundant detection
+                    # Use cached face positions from detection thread
                     swapped_bboxes = []
                     if modules.globals.many_faces and cached_many_faces:
                         result = temp_frame.copy()
@@ -1051,6 +1195,10 @@ def _processing_thread_func(capture_queue, processed_queue, stop_event):
                 if frame_processor.NAME == "DLC.FACE-ENHANCER":
                     if modules.globals.fp_ui["face_enhancer"]:
                         temp_frame = frame_processor.process_frame_v2(temp_frame)
+                elif frame_processor.NAME in ("DLC.FACE-ENHANCER-GPEN256", "DLC.FACE-ENHANCER-GPEN512"):
+                    fp_key = frame_processor.NAME.split(".")[-1].lower().replace("-", "_")
+                    if modules.globals.fp_ui.get(fp_key, False):
+                        temp_frame = frame_processor.process_frame_v2(temp_frame)
                 else:
                     temp_frame = frame_processor.process_frame_v2(temp_frame)
 
@@ -1104,6 +1252,14 @@ def create_webcam_preview(camera_index: int):
     processed_queue = queue.Queue(maxsize=2)
     stop_event = threading.Event()
 
+    # Shared state for the detection pipeline.
+    # latest_frame_holder[0] is the most recent raw frame for the detection
+    # thread; detection_result holds the last detected faces for the
+    # processing thread to read.  Both are guarded by detection_lock.
+    detection_lock = threading.Lock()
+    latest_frame_holder = [None]
+    detection_result = {'target_face': None, 'many_faces': None}
+
     # Start capture thread
     cap_thread = threading.Thread(
         target=_capture_thread_func,
@@ -1112,21 +1268,45 @@ def create_webcam_preview(camera_index: int):
     )
     cap_thread.start()
 
+    # Start detection thread — runs face detection asynchronously so the
+    # processing/swap thread never blocks on it
+    det_thread = threading.Thread(
+        target=_detection_thread_func,
+        args=(latest_frame_holder, detection_result, detection_lock, stop_event),
+        daemon=True,
+    )
+    det_thread.start()
+
     # Start processing thread
     proc_thread = threading.Thread(
         target=_processing_thread_func,
-        args=(capture_queue, processed_queue, stop_event),
+        args=(capture_queue, processed_queue, stop_event,
+              latest_frame_holder, detection_result, detection_lock),
         daemon=True,
     )
     proc_thread.start()
 
-    # Main (UI) thread: pull processed frames and update the display
-    while not stop_event.is_set():
+    # Cleanup helper called from the display loop when preview closes
+    def _cleanup():
+        stop_event.set()
+        cap_thread.join(timeout=2.0)
+        det_thread.join(timeout=2.0)
+        proc_thread.join(timeout=2.0)
+        cap.release()
+        PREVIEW.withdraw()
+
+    # Non-blocking display loop using ROOT.after() — avoids blocking the
+    # Tk event loop which could cause UI freezes or re-entrancy issues
+    def _display_next_frame():
+        if stop_event.is_set() or PREVIEW.state() == "withdrawn":
+            _cleanup()
+            return
+
         try:
-            temp_frame = processed_queue.get(timeout=0.03)
+            temp_frame = processed_queue.get_nowait()
         except queue.Empty:
-            ROOT.update()
-            continue
+            ROOT.after(16, _display_next_frame)
+            return
 
         if modules.globals.live_resizable:
             temp_frame = fit_image_to_size(
@@ -1144,17 +1324,11 @@ def create_webcam_preview(camera_index: int):
         )
         image = ctk.CTkImage(image, size=image.size)
         preview_label.configure(image=image)
-        ROOT.update()
 
-        if PREVIEW.state() == "withdrawn":
-            break
+        ROOT.after(16, _display_next_frame)
 
-    # Signal threads to stop and wait for them
-    stop_event.set()
-    cap_thread.join(timeout=2.0)
-    proc_thread.join(timeout=2.0)
-    cap.release()
-    PREVIEW.withdraw()
+    # Kick off the non-blocking display loop
+    ROOT.after(0, _display_next_frame)
 
 
 def create_source_target_popup_for_webcam(

modules/ui_tooltip.py

@@ -0,0 +1,74 @@
+"""Lightweight hover tooltip for CustomTkinter widgets."""
+
+import customtkinter as ctk
+
+
+class ToolTip:
+    """Show a floating tooltip popup when the user hovers over a widget.
+
+    Usage:
+        ToolTip(my_button, "Helpful description text")
+    """
+
+    def __init__(self, widget: ctk.CTkBaseClass, text: str, delay: int = 500):
+        self._widget = widget
+        self._text = text
+        self._delay = delay
+        self._tooltip_window = None
+        self._after_id = None
+
+        widget.bind("<Enter>", self._schedule_show, add="+")
+        widget.bind("<Leave>", self._hide, add="+")
+
+    def _schedule_show(self, event=None):
+        self._cancel()
+        self._after_id = self._widget.after(self._delay, self._show)
+
+    def _show(self):
+        if self._tooltip_window is not None:
+            return
+
+        x = self._widget.winfo_rootx() + 20
+        y = self._widget.winfo_rooty() + self._widget.winfo_height() + 5
+
+        self._tooltip_window = tw = ctk.CTkToplevel(self._widget)
+        tw.withdraw()
+        tw.overrideredirect(True)
+
+        label = ctk.CTkLabel(
+            tw,
+            text=self._text,
+            fg_color="#333333",
+            text_color="#EEEEEE",
+            corner_radius=6,
+            padx=8,
+            pady=4,
+        )
+        label.pack()
+
+        tw.update_idletasks()
+
+        # Clamp to screen bounds
+        screen_w = tw.winfo_screenwidth()
+        screen_h = tw.winfo_screenheight()
+        tip_w = tw.winfo_reqwidth()
+        tip_h = tw.winfo_reqheight()
+
+        if x + tip_w > screen_w:
+            x = screen_w - tip_w - 5
+        if y + tip_h > screen_h:
+            y = self._widget.winfo_rooty() - tip_h - 5
+
+        tw.geometry(f"+{x}+{y}")
+        tw.deiconify()
+
+    def _hide(self, event=None):
+        self._cancel()
+        if self._tooltip_window is not None:
+            self._tooltip_window.destroy()
+            self._tooltip_window = None
+
+    def _cancel(self):
+        if self._after_id is not None:
+            self._widget.after_cancel(self._after_id)
+            self._after_id = None

modules/utilities.py

@@ -15,10 +15,6 @@ import modules.globals
 TEMP_FILE = "temp.mp4"
 TEMP_DIRECTORY = "temp"
 
-# monkey patch ssl for mac
-if platform.system().lower() == "darwin":
-    ssl._create_default_https_context = ssl._create_unverified_context
-
 
 def run_ffmpeg(args: List[str]) -> bool:
     """Run ffmpeg with hardware acceleration and optimized settings."""
@@ -286,8 +282,15 @@ def conditional_download(download_directory_path: str, urls: List[str]) -> None:
             download_directory_path, os.path.basename(url)
         )
         if not os.path.exists(download_file_path):
-            request = urllib.request.urlopen(url)  # type: ignore[attr-defined]
-            total = int(request.headers.get("Content-Length", 0))
+            request = urllib.request.Request(url)
+            
+            # Create a specific SSL context for macOS to avoid globally disabling verification
+            ctx = None
+            if platform.system().lower() == "darwin":
+                ctx = ssl._create_unverified_context()
+                
+            response = urllib.request.urlopen(request, context=ctx)
+            total = int(response.headers.get("Content-Length", 0))
             with tqdm(
                 total=total,
                 desc="Downloading",
@@ -295,7 +298,13 @@ def conditional_download(download_directory_path: str, urls: List[str]) -> None:
                 unit_scale=True,
                 unit_divisor=1024,
             ) as progress:
-                urllib.request.urlretrieve(url, download_file_path, reporthook=lambda count, block_size, total_size: progress.update(block_size))  # type: ignore[attr-defined]
+                with open(download_file_path, "wb") as f:
+                    while True:
+                        buffer = response.read(8192)
+                        if not buffer:
+                            break
+                        f.write(buffer)
+                        progress.update(len(buffer))
 
 
 def resolve_relative_path(path: str) -> str:

requirements.txt

@@ -9,8 +9,8 @@ tk==0.1.0
 customtkinter==5.2.2
 pillow==12.1.1
 onnxruntime-silicon==1.16.3; sys_platform == 'darwin' and platform_machine == 'arm64'
-onnxruntime-gpu==1.24.2; sys_platform != 'darwin'
+onnxruntime-gpu==1.23.2; sys_platform != 'darwin'
 tensorflow; sys_platform != 'darwin'
 opennsfw2==0.10.2
-protobuf==5.29.6
+protobuf==4.25.1
 pygrabber

tkinter_fix.py

@@ -1,3 +1,6 @@
+import os
+os.environ.setdefault('TK_SILENCE_DEPRECATION', '1')
+
 import tkinter
 
 # Only needs to be imported once at the beginning of the application