It seems was in the development program…
am I wrong?
thanks
1 Like
It seems was in the development program…
Please clarify this statement.
Quote from June 2024
It was not planned for DEVONthink 3.
I see DVT4 is now able to recognize handwritten text. That is amazing!
I’m having a bit of trouble understanding how this works. I’ve read the two paragraphs on page 331 of the manual, but when I try to convert a PDF file with handwritten text to a PDF+Text file using the context menu or automatically after importing it, I’m getting really poor results. Can you help me out?
OCR doesn’t support handwriting.
- Make sure you have your AI > Chat settings set up with a vision-capable model (with the eye icon) then use Edit > Summarize via Chat.
- Or set up AI > Transcription settings and use Data > Recognition > Transcript Text & Notes on the image.
Hm, the paragraph on page 331 states
Pro — PDF documents without a text layer are now indexed using macOS, even handwritten text. This only works on PDFs without embedded text. Modified documents lacking a text layer for some pages will have one produced after the first save.
I therefore assumed no external AI is needed. I am still struggling setting one up.
1 Like
Choose Apple Vision Accurate in the Transcription preferences and use the recognition command.
Thanks for the hint. I’ve already set the preferences accordingly.
If I select a PDF or PNG and use Data > Recognition > Transcript Text & Notes nothing happens.
One thing I noticed that might be correlated:
- I can edit the files with PDFExpert
- I cannot edit the files with the build-in Preview App. It says: “The original document can’t be changed, so a duplicate with your changes has been created”
Edit: Seems to be a problem in the Inbox only though, and moving the file into a database doesn’t solve it.
That is not a DEVONthink nor DEVONtech issue. That is an Apple behavior and cannot be changed by us.
Show the settings.
You’re not going to see anything happen. You have chosen to transcribe to searchable text. This is discussed in the Help…
Select the image and choose View > Document Display > Text Alternative.
Oh! I got it now. I should have read the manual more thoroughly.
I was hoping to get a handwritten PDF with a text layer that I can search. So I know where (on which page) the search phrase is used. Like macOS is doing it in the Fotos app.
Is there a way to accomplish that in DT4?
Development would have to weigh in on that, but there isn’t a text layer on top of the image. It’s still an image and remains unchanged when using the Searchable Text option.
Yeah, probably that‘s right. The photo is still a photo but when I search a phrase in Fotos, e.g. Test I get this:
That’s really neat!
But actually I‘m more interested in PDFs because they can have more than one page. PDFs can have an invisible text layer for search, which is neat for handwritten notes.
Imagine using DTTG to scribble down notes with an Apple Pencil, sync them to DT4, which then makes them searchable. Or scanning a handwritten note of paper into DT4 which then makes it searchable.
I‘d really love that feature.
Well… Fotos (or Photos here) ≠ DEVONthink 
Yeah, that’s why I’m keeping my photos in the Photos App 
I‘m trying to imagine, what I could do with the transcription function as implemented of today.
Audio notes come into my mind. I have never used them so far. But with a transcription they might come in handy. I got to try them out.
Concerning handwritten PDFs store them next to all my other files. A transcription I can now store in the DT Database as well, somewhere in the DB, in an separate annotation file or in the macOS comment field. A search will now find the PDF, which is great. It will not give me the exact page(s) the phrase is on. If I‘m looking for that, the annotation file option seems to be the best option I guess.
Does that make sense?