Text2Folders is intentionally narrow: one screenshot of line-oriented text in, one ZIP of folders out.
What works best
- Clear screenshots (UI lists, not blurry camera photos)
- One item per line (addresses, SKUs, labels)
- High contrast (light text on dark, or the reverse)
Dense paragraphs and multi-column tables are out of scope for v1.
Naming rules
Each accepted OCR line becomes a folder name. Text2Folders trims whitespace, removes unsafe path characters, handles reserved names, drops empty results, and removes case-insensitive duplicates. In the web flow, review and edit recognized lines before finalizing the ZIP.
Why ZIP
A standard ZIP extracts with normal archive tools on Windows, macOS, and Linux, with no proprietary format. Text2Folders temporarily stores the source screenshot and generated ZIP in Cloudflare R2 so jobs can be processed and downloaded. Default file retention is up to 24 hours for anonymous jobs and 168 hours for signed-in jobs; deleting or expiring a job removes its stored objects.
OCR is fallible, and the staging service is not intended for sensitive, regulated, or production-critical workloads.
