Image to Text — OCR with Bold, Italic & Color Detection

For text-based PDFs, bold and italic are read directly from the PDF's font data — this is exact, not a guess. For images and scanned documents, Tesseract's OCR engine estimates bold/italic based on stroke width and character shape analysis — accuracy is typically 85–95% on clear, high-resolution images and lower on blurry or low-quality scans.

For PDFs with a text layer, ToolReign uses a neutral default in v1 (PDF.js does not expose fill color via getTextContent). For images, color is sampled directly from the darkest pixels within each detected word's bounding box — this works reliably for standard dark-text-on-light-background documents but may be less accurate for stylized or low-contrast text.

No. PDF.js and Tesseract.js run in your browser — your file is not sent to our servers. On first OCR, the browser downloads the Tesseract WASM runtime and English language data from a CDN (jsDelivr); later runs reuse the cached copy.

Yes. When a PDF has no usable text layer, ToolReign automatically renders each page and runs OCR, with a notice that scanned mode is active.

Plain text, HTML with inline styles, Markdown with heading heuristics, and full JSON of the extraction data structure.

Images over 10 MB show a warning; images over 25 MB are blocked. PDFs with more than 20 pages show a warning — pages are processed on demand.