Fix CoreML execution provider falling back to CPU silently, eliminate
redundant per-frame face detection, and optimize the paste-back blend
to operate on the face bounding box instead of the full frame.
All changes are quality-neutral (pixel-identical output verified) and
benefit non-Mac platforms via the shared detection and paste-back
improvements.
Changes:
- Remove unsupported CoreML options (RequireStaticShapes, MaximumCacheSize)
that caused ORT 1.24 to silently fall back to CPUExecutionProvider
- Add _fast_paste_back(): bbox-restricted erode/blur/blend, skip dead
fake_diff code in insightface's inswapper (computed but never used)
- process_frame() accepts optional pre-detected target_face to avoid
redundant get_one_face() call (~30-40ms saved per frame, all platforms)
- In-memory pipeline detects face once and shares across processors
- Fix get_face_swapper() to fall back to FP16 model when FP32 absent
- Fix pre_start() to accept either model variant (was FP16-only check)
- Make tensorflow import conditional (fixes crash on macOS)
- Add missing tqdm dep, make tensorflow/pygrabber platform-conditional
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Some people just want the opencv error gone. I keep on telling them that it is only for blurs and color conversion. It is the onnx runtime who is running the swap.
The macOS Apple Silicon section installed Python 3.11 but then
referenced Python 3.10 in several places:
- `brew install python-tk@3.10` → python-tk@3.11
- Linux comment "Ensure you use the installed Python 3.10" → 3.11
- CoreML section cross-reference "completed the macOS setup above
using Python 3.10" → 3.11
- `python3.10 run.py` usage command → python3.11
- "You must use Python 3.10" note → 3.11
- `brew reinstall python-tk@3.10` troubleshooting tip → 3.11
- Removed `python@3.11` from the list of conflicting versions to
uninstall (it is the required version, not a conflict)
Fixes#1632
The live webcam preview in ui.py calls process_frame_v2() on all
frame processors, but face_enhancer.py was missing this method.
This caused an AttributeError crash when the GFPGAN face enhancer
was enabled during live mode.
Fixes https://github.com/hacksider/Deep-Live-Cam/issues/1654
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Change default face swapper model to FP32 for better GPU compatibility and avoid NaN issues on certain GPUs.
Revamped `run.py` to adjust PATH variables for dependencies setup and re-added with expanded configuration.
Add ToolTip class (modules/ui_tooltip.py) and wire descriptive hover
tooltips onto every button, switch, slider, and dropdown in the main
window. Tooltips appear after a 500ms hover delay and are clamped to
screen bounds.
This requires no new dependencies — ToolTip uses only customtkinter.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
cv2_enumerate_cameras(CAP_AVFOUNDATION) probes indices 0-99 through
OpenCV's AVFoundation backend, which intermittently segfaults (exit
code 139) when invalid device indices are probed. Replace with a
bounded cv2.VideoCapture loop (range(10)) that safely skips
unavailable indices.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Replace float64 with float32 in apply_mouth_area() blending masks —
float32 provides sufficient precision for 8-bit image blending and
halves memory bandwidth
- Use float32 in apply_mask_area() mask computations
- Vectorize hull padding loop in create_face_mask() (face_masking.py)
replacing per-point Python loop with NumPy array operations
- Fix apply_color_transfer() to use proper [0,1] LAB conversion —
cv2.cvtColor with float32 input expects [0,1] range, not [0,255]
- Pre-compute inverse masks to avoid repeated (1.0 - mask) subtraction
- Use np.broadcast_to instead of np.repeat for face mask expansion
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Move opacity calculation before frame copy to skip the copy when
opacity is 1.0 (common case). Add early return path for full opacity.
Clear PREVIOUS_FRAME_RESULT instead of caching when interpolation
is disabled.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add a dedicated detection thread that runs face detection continuously
on the latest captured frame and publishes results to a shared dict.
The processing/swap thread reads cached detection results instead of
running detection inline, so it never blocks on the 15-30ms detection
cost.
Architecture change: 2 threads → 3 threads
Before: capture → [detect + swap] → display
After: capture → swap (uses cached detections) → display
↘ detect (async, writes to shared cache) ↗
Also replaces the blocking while/ROOT.update() display loop with
ROOT.after()-based scheduling, which avoids Tk event loop re-entrancy
issues and UI freezes.
Closes#1664
Max Buckley