"stage output" -> mixed: picture zones layer

[Golddouble]

This is a question to the function "stage output" -> mixed

I do not understand, what the advantage (or sense) of the function "mixed" has compared with "Colour / Gray scale" mode?

Question 1: Is the sense, to save place? (I mean to make the tiff-file smaller compared with "Colour / Gray scale" mode)?

When I choose mixed-mode, then I can use the "picture zones to automatically detect pictures and separate them from text.

One problem of tesseract based OCR programmes is, that they can not proper separate text from picture. It looks like ScanTailor can this better. And in the tesseract based OCR programmes we have not the possibility to manually mark/select "text areas" to help tesseract only to apply OCR on areas that are really text.

So I ask me, if I can in any way use the mixed -> picture/text zones detected through ScanTailor in my OCR programme.

Question 2:
You speak about "auto layers" that can be seen in the tab "picture zones". Are this zones somehow saved in the resulting tiff? And if yes, can my OCR programme this zones use, to decide, if it should apply OCR to find text in a certain zone or not.

Would appreciate some answer. Thank you.

[mara004]

I think your guesses are largely correct. The mixed mode is used to separate image and text areas. Since the results produced by free algorithms often aren't perfect, it makes sense to have means of manual adjustment in a GUI like ScanTailor, as you say. The result may then be used for a variety of purposes, including the creation of mixed raster content DjVu or PDF files with particularly efficient compression, due to the use of different encoders for text and images. Or it could be passed to an OCR tool that needs this information as well. Whether that's possible with tesseract on the command-line I don't know, though.