OCR (Pro)
Text-recognition workbench. The values here become the defaults for the OCR nodes in the graph.
OCR · Viewer
Capture, read, verify
Pick a window, take a snapshot or load an image, and run recognition on the exact region you draw on the preview. Live mode keeps re-reading.
- Window · desktop · file source
- Snapshot history
- Live OCR mode
- Pick a window and the text-recognition engine (Tesseract or PaddleOCR).
- Tune the pre-processing until the text is readable and see the live result.
- Pre-processing for difficult text: contrast, sharpness, denoise, deskew (auto-straighten).
- Black/white split — turn a grey image into pure black and white text.
- Smoothing radius — cleans up broken strokes.
- White padding around the image — some engines read better with a border.
- Language settings and engine paths are edited here too. For PaddleOCR point the app at your model files (put them in vendor/paddleocr next to the program, or type in full paths).