Getting started with StyleOCR

StyleOCR turns images and PDFs into structured text and documents (for example Word, Markdown, plain text, or Excel) using on-device OCR and optional vision-language models (VLM) for layout-heavy pages.

Before you begin

• You need permission to read the files you import (scans, exports, photos of documents, etc.).
• Recognition quality depends on resolution, skew, lighting, and language; use Pro or VLM mode when accuracy matters more than speed.
• Large batches can run in the background; tasks can be stopped and resumed from history when supported by your build.

Install the app

Install StyleOCR from the distribution channel you use (for example the macOS build from the product site or app store listing). After launch, sign in to billing only if you use subscription features; core workflows work offline once models are available locally.

Your first task

1. Create or open a task from the main workflow. A task groups one or more images or PDF pages you want to process together.
2. Add files by drag-and-drop or the file picker. Supported inputs include common raster formats and PDF; the app shows what is supported in the drop zone hint.
3. Choose an OCR mode (for example Fast, Pro, or VLM) depending on whether you need speed, higher accuracy, or richer layout and table understanding.
4. Pick an output format — such as Word (.docx), Markdown, plain text, or Excel — and optional features like style extraction, retain images in document, or page break behavior for PDFs.
5. Start OCR and wait for progress. When finished, download or open results from the task view.

After the first run

• Install or switch vision models under Settings → Vision models if you need VLM features or a different OCR engine bundle.
• Explore information extraction presets (for example invoice or ID-style flows) or define your own fields in the task options.
• Enable MCP server under Settings if you want compatible assistants to call your local StyleOCR instance.

Next: read the User guide for modes, models, and integrations.