Filozy

Four-step archive workflow

Make a scanned document searchable and smaller

A colour scan of a printed page is usually far larger than it needs to be and cannot be searched at all. Greyscale first, recognise the text, then compress.

By the Filozy team, with AI assistance Reviewed How we write and review

One file · 4 connected steps

Start the workflow here

Choose the PDF once. Each completed step passes it directly to the next tool on this device.

The whole job, visible up front

What will happen

  1. 01

    Trim margins

    Trim the blank margin, so the pages the text step redraws are no bigger than they need to be.

    Open Trim margins by itself
  2. 02

    Remove colour

    Drop colour the page never needed, and check the size actually fell.

    Open Grayscale by itself
  3. 03

    Recognize text

    Add a searchable text layer. English only for now; leave the page list empty to keep every page.

    Open OCR by itself
  4. 04

    Compress

    Choose Reduce under Pictures. The text step stores each page as a new lossless picture, and only re-encoding pictures makes that small.

    Open Compress by itself

Common questions

Why did the file get bigger after the text step?
Because the searchable copy redraws each page as a new picture at 200 dots per inch, stored without loss, with the recognised words laid invisibly behind it. A lossless picture of a page is large. The Compress step is where it shrinks, but only if you choose Reduce under Pictures: Keep quality, the default, never re-encodes a picture.
Will the result always be smaller than the scan I started with?
Not always. The finished file is made of pictures that were redrawn and then recompressed, so a black-and-white scan that was already stored compactly can come out larger. Compare the final size with your original, and keep the original either way.
Which languages can it recognise?
English only for now. Each run reports an average confidence: below about 80%, read the text before relying on it. A straight, high-contrast scan at 300 dpi does far more for accuracy than anything afterwards, and Scan to PDF can straighten a photograph of a page before you start.
Can I recognise only some of the pages?
You can, but the searchable copy contains only the pages that were read, so the rest would be missing from the finished file. Leave Pages to recognise empty to keep them all. Keep Also make a searchable PDF ticked too: without it the step makes only a text file, and there is no PDF to carry on with.