Show HN: Local-first fast CPU image to text for screenshots, PDFs, webpages

6/5/2026 at 9:37:03 PM

What's the performance like compared to tesseract? I don't see tesseract mentioned anywhere in the readme, which is surprising considering that's the number one tool most go to for Image > text OCR.

by KetoManx64

6/5/2026 at 9:48:04 PM

No rigorous eval, and I love Tesseract. Here's the example that motivated me to build textsnap (which is in the github's README), parsed with Tesseract:

https://imgur.com/a/i2eQra8

by mrkn1

6/6/2026 at 1:54:47 AM

Very noticable difference and the exact issue I run repeatedly with tesseract! Definitely going to try dropping textsnap into my scripts now. Thanks!!

by KetoManx64

6/6/2026 at 4:40:11 AM

This is awesome! Been needing something like this for some research paper diagrams I've been indexing.

by lavaman131

6/5/2026 at 1:20:38 PM

Curious how it does on multi-page scanned PDFs vs. single screenshots? The ORT vision/decoder split is the part that usually makes or breaks CPU VLM OCR...

by abstract257

6/5/2026 at 2:08:38 PM

I had to extract the image from a PDF for it to work. Then run it on each page image extracted.

by krunck

6/6/2026 at 6:32:00 AM

Thanks

by abstract257

6/5/2026 at 3:17:16 PM

- how well do you think this ll work with code? i mean take code screenshots and convert it into actual code for vscode

by vivzkestrel

6/5/2026 at 3:20:50 PM

Just ran

  textsnap "https://i.ytimg.com/vi/LBNDfxjEYlA/maxresdefault.jpg"

and got this

  $('.count').each(function () {
  $('this').prop('Counter', 0).animate({
    Counter: $('this').text()
  }, {
      duration: 4000,
      easing: 'swing',
      step: 'function (now) {
          $('this").text(Math.ceil(now));
      }
    }); 
  });

by mrkn1

6/5/2026 at 3:36:37 PM

What was the reason for adopting PaddleOCR? Can other OCR models be used as well?

by monosma

6/5/2026 at 3:42:15 PM

No reason other than their Q4 model working reasonably well and fast on my CPU laptop. Should work with any ONNX VLM model

by mrkn1

6/5/2026 at 3:33:23 PM

Roman alphabet only or does this work with other alphabets?

by kouru225

6/5/2026 at 3:41:06 PM

109 languages, including other alphabets.

by mrkn1

6/5/2026 at 12:03:02 PM

Very cool, I'm building my own local-first product as well

by garrett2558

6/5/2026 at 12:28:52 PM

thank you! what is it about?

by mrkn1

6/5/2026 at 2:05:19 PM

Now this is legit cool, keep up the great work.

by BIGFOOT_EXISTS

6/5/2026 at 2:37:57 PM

thank you!

by mrkn1