IMSLP Forums

I found a post on another forum about optimizing scores from IMSLP for reading on tablets. There was a link to a website that would optimize a score for you. The website requires an email address, and the post looked just sketchy enough that I didn't want to enter my email address.

However, the website does provide two links to PDFs of Lucia di Lammermoor, one from IMSLP, one optimized for tablets. The optimized score scrolls much better on my tablet, but is significantly larger. Can anyone give me a hint as to how the optimized score is different? Is it just a lower compression ratio of the images in the PDF? The optimization shouldn't be too hard to accomplish on my own once I know what it is.

Here's a link to the website where you can download the optimized PDF.

http://fastpdfscores.com/

I extracted out the first page of music from each of these and saved them as tif files. The page from the IMSLP score (courtesy of Google) has dimensions of 5168 by 7063 pixels and is 697dpi. That is a high-quality image. The page from Fastpdfscores is 721 by 1024 pixels and is detected as being 72 dpi. That is garbage. By significantly reducing the image resolutions they get faster page loads. The images probably look OK on a tablet but you would never be able to print them and the quality difference is easily noticeable on a desktop. The reason for the file from Fastpdfscores being twice the size of the IMSLP score is because every page of their score is in grayscale (256 shades of gray) while the IMSLP file consists of (with a few exceptions thanks to Google's weird way of combining grayscale and black and white on certain pages) pages that are true black and white (2 colors). A true black and white image at the low resolution provided by Fastpdfscores would look horrible so they've made the image a little less worse by converting it to grayscale.

I would be interested to know how page turns on a tablet for the black and white versions of Ritter's Zur Geschichte des Orgelspiels compare to the IMSLP/Google PDF offered as an example by Fastpdfscores.

Cypressdome

I see. He is just reducing the resolution.

I downloaded the black and white music pdf from the page you linked to my Android tablet, and it had reasonable performance. I see that you uploaded that score. Did you do anything in particular to make it easier to load?

For example, here's a song that I've looked at that has terrible performance on my tablet.

http://imslp.org/wiki/The_Red_Rose_Whis ... e,_Arthur)

What is different about this one?

For the Ritter score I followed the same procedure that I've been following for a couple of years now. All the images within the PDF file are 600dpi, black and white tif files. I use Adobe Acrobat 9.2 (PDF version 1.6) to create the PDF and use CCITT G4 compression on the images. I did this score because another user was finding it impossible to use the color PDF files. Those might choke your tablet since those PDF files are based on color jpg2000 images which can be problematic for computers to render.

The images in the Foote score from Sibley are 400dpi black and white which seems to be the standard for Sibley. It was created using PDFScanLib v.1.2.2 in Adobe Acrobat 9.2 (PDF version 1.6). I can't tell what type of compression they used but the default for black and white tif files is JBIG2 (lossless) which appears to compress the file a bit more than CCITT G4 does. Don't know how, or if, any of this would affect its performance on a tablet. Perhaps, it would be instructive to extract the images from the Foote score and re-compress then using CCITT G4 in a new PDF to compare.

Cypressdome

I use the command line tools tiffcp and tiff2pdf for my own scans. At some point I looked up the compression scheme that these tools use, and they use the scheme recommended by IMSLP. It would not be hard to write a script to automatically convert the Sibley files.

I would like to test these files on iOS devices to see if they have similar issues.

Edit: I checked out that color score, and it is unuseable on my tablet.