Color Magazine Comparision using Segmented DjVu & PDF
a Comparision by PlanetDjVu, November 3, 2003
DjVu Version - 30.2 MB (requires DjVu plugin - any version)
PDF Version - 50.8 MB (requires Acrobat 6 viewer)
In this comparison we use a 260-page color magazine that was scanned and saved as 300-dpi JPEG page files. The JPEG file size for this magazine issue totals 1.31 GB, an average of 5.03 MB per page.
Conversion to DjVu
We used JRAPublish to convert the JPEG page files to Segmented (2-layer) DjVu using the default Scan300 profile. We then used the ABBYY FineReader engine in JRAPublish to OCR the document. We saved the DjVu file in the INDIRECT format and then embedded this in an HTML file using object tags, for web presentation. We also embedded thumbnail images in the document.
The final size of the magazine in DjVu format is 30.2 MB, an average of 11.6 KB per page, and just 2% of the original JPEG size.
Conversion to PDF
We used the latest version of LuraDocument PDFCompressor to convert the JPEG page files to Segmented (2-layer) PDF. We could not OCR this PDF file, because PDFCompressor does not have an OCR engine compiled with it, and because Acrobat cannot OCR it. Unfortunately, Acrobat recognizes only the 100-dpi color layer of the file, and not the 300-dpi bitonal layer of the file, so it presents an error message saying "the page image must be between 200 - 600 dpi".
The current version of the PDFCompressor product only outputs single page PDF files, but we had not problem binding them into one document in Acrobat, and then generating embedded thumbnail images.
The final size of the magazine in PDF format is 50.8 MB, an average of 19.5 KB per page, and just 3.9% of the original JPEG size.
For comparison, we also make a PDF version without using Segmentation. The resulting file is 624 MB, or 2.6 MB per page, a reduction in file size of 53%. This file is far to large for web delivery, a problem that has plagued PDF containing color scans, until the introduction of Segmented PDF by Algo Vision - LuraTech in September of this year.