PDF vs DJVU

[loveinallitsforms69]

OK. It looks like PDF is everywhere, and yet another very useful document format, DjVu, is almost nowhere to be found. It also looks like most younger onliners have not even heard of DjVu. Yet it has greater compression efficiency and renders images faster than PDF does. The format has been available cross-platform for almost 25 years.

In fact, three prominent magazines, The New Yorker, Playboy, and The Rolling Stone, each preferentially archived their decades of magazine scans in the DjVu format, as it was the only format that could efficiently compress those pages onto several DVDs instead of 40 or 50.

The PDF company FoxIt Technologies actually purchased the right to use DjVu-based compression technology to create its HC (high-compression) PDFs, which had file sizes similar to their DjVu counterparts.

So it is possible that some of our PDFs might be thought of as PDF-DjVu hybrids!

I have not heard of anyone purchasing the rights to use PDF-based technology to compress their files.

It is true that DjVu is not reflow-capable, but as an archival document format for ebooks, I think it is still a winner. It can create its own bitmap images from jpegs. Its compression efficiency is about 50% better than that of jpeg. So for instance a 400 KB DjVu image can look as good as an 800 KB jpeg image derived from the same source file.

DjVu is highly customizable in its OCR capabilities. You can add a separate OCR text file to your DjVu document such that if you have an image of an oak tree on page 3, you enter the coordinates of that oak tree (as a rectangular outline) into the text file, add in the word "oak" and incorporate the text file into the DjVu document. To do that, you use the free DjVu encoder called DjVuLibre, available for Windows, Mac, and Linux systems. And when you then type the word "oak" in the DjVu document, the oak tree on page 3 is highlighted!

I think one challenge is that most PDFs are not that lengthy, and the difference in performance between PDFs and DjVus is much more noticeable in longer, image-heavy documents. And of course computers are faster than they were 25 years ago.

Another challenge is that the free DjVuLibre encoder program is command-line only!

That's a lot more involved than "Export as PDF."

A text page in DjVu format may look like it has movable type, but it is in fact a bitmap image, as all DjVu pages are. The pages are essentially written in stone, which is a good thing for archiving. The advent of many free PDF editors that can edit characters adds a security risk to PDF that a DjVu document does not face. Oddly though, the DjVu format as it stands does not offer password protection, although a DjVu document can be "zip-locked" into a secured zip file.

One other thing about DjVu is that it does not require any "font help" to have its pages rendered. All of the text pages are actually snapshots of text pages. Simplified, think of a DjVu document as looking like a Word document, but actually being like crisp pngs of the Word document.

I asked the Library of Congress to include the specifications of the DjVu format in their document format records, a section that archives the various available document types for the public record. They politely refused to do so, with the strange excuse that the library did not have many DjVu documents, but had many PDF documents. Your tax dollars at work.

There is a free site called any2djvu that lets you convert PDFs, jpegs, etc to DjVu.
It has a vintage Y2K look to it!

There is a new program called minidjvu-mod that is being used to get compression up to 10x greater than even the commercial encoder. I am a contributor to the new program. It is free, and now has a GUI version, but is limited to text-related pages.

Ideally all of the free programs need to be integrated into one fine GUI DjVu encoder program.

Another thing is that you can revert DjVu to PDF and other formats using DjVuLibre.

It is remarkable how under-the-radar this very useful format truly is.

Thanks.

[candyflip]

[baddog]

Testing AI?

[loveinallitsforms69]

Originally Posted by baddog

Testing AI?

Thanks. At 30 posts I can send a link to show some AI material, but I do not have permission here yet.
This is according to GFY.

I actually used the IrfanView slideshow on a sequence of about 1000 DjVu images derived sequentially from about 1000 sequential jpegs in turn derived from an older version of Avidemux to try to make the first ever DjVu image-based video. The images were from the first 35 seconds of an adult movie. IrfanView can display DjVu images at a fast enough speed to make them look like a movie. But then I realized there was a roadblock to turning that into an upscalable DjVu movie as there were no viewers available to watch a DjVu movie of any sort. I mean with audio and better options than just a high-speed slideshow. So obviously I did not have a good pipeline set up!

But there was the idea that DjVu images really could be upscaled, but only indirectly. I would have to start ideally with lossless images created in Topaz Video Enhance. It does offer the option to upscale the video and save it as sequential lossless images. Then those images could be rendered as what would be considered upscaled DjVu images. So yes, AI was involved but I guess I was trying to be too much of a pioneer too quickly!

[Farang]

Originally Posted by baddog

Testing AI?

hehe

[AmeliaG]

I, for one, welcome Skynet.

[CaptainHowdy]

Cliffnotes ? ?

[DVTimes]

Never heard of DJVU.