Please help me reduce file size

[nfn]

I've submitted a number of items edited by Vladimir de Pachmann, including for instance the Chopin Polonaise op.53. My files are much larger, by a factor of about 5, than those of other submitters of the same piece and using the same 600 dpi. I've tried many methods to reduce my file size, without success. I'm using Adobe Acrobat X Pro. Can someone take one of those files, maybe the Polonaise, and reduce it, and then tell me how to do it myself? Otherwise, I'll have to continue as at present. Thanks, Nigel.

[coulonnus]

Do you mean #429513 here? http://imslp.org/wiki/Polonaise_in_A-fl ... %C3%A9ric)
Try monochrome instead of grayscale!

[nfn]

Thanks very much! That does seem to be the explanation. However, with monochrome, the quality is inadequate in this case, so the larger file size with grayscale seems warranted. But I will certainly experiment further with monochrome scanning... Nigel.

[coulonnus]

Are you aware this Chopin scan is only 150 dpi? Of course it will be ugly converted to monochrome!

[nfn]

No, all mine were scanned at 600 dpi Black and White (with paint.net). (I had misunderstood your first message, because it was never grayscale.)

I'm not sure whether you're looking at the right one: it is #429513 (edited by Vladimir de Pachmann, size 11.62 MB, scanned and uploaded by Nigel Nettheim). The quality is very good, as you would expect at 600 dpi, but I'm just wondering whether the file size can somehow be reduced without losing quality.

Thanks again, Nigel.

[coulonnus]

Yes I'm looking at this very scan. If you magnify a note head a lot you will see it is grayscale and about 10-pixels high. It's about 40 pixels for a good 600-dpi scan. Am I the only one seeing this? Is it an artefact of my browser here?

[nfn]

Good -- that is similar to what I see in the pdf (about 12 pixels per note-head grayscale). I scanned at 600 dpi B/W and still have the jpeg files, which also look the way you described them (about 40 pixels per note-head B/W).

Then the jpegs have to be converted to pdf for IMSLP. I used Adobe Acrobat Pro X, with "Create" -> "Combine files into a single pdf", and saved the file as pdf. That produces a very large file, so I then saved it as "Reduced Size pdf". I'm not sure exactly what that does, but it probably converts the images to grayscale at reduced dpi. It can reduce the file size to about 10% of what it was, and it still looks good to me, and can then be uploaded to IMSLP within their file-size limits. But I wonder whether I could do better than Adobe's "Reduced Size pdf". Somehow other people seem to get good quality with smaller file sizes than mine. Do they use a different method to reduce the file size? Any comments on that would be welcome!

[coulonnus]

Can't your scanning application provide tif files instead of jpg files? Even with a monochrome scan jpg files are coded with grayscale information.

[horndude77]

I'm not very familiar with how paint.net and adobe acrobat creates their outputs, but the suggestion to try monochrome tiff is correct. The trick is that tiff can encode its image information using different algorithms. One of the primary algorithms used to get good compression with bilevel images is CCITT Group 4 compression (see https://en.wikipedia.org/wiki/Group_4_compression). JBIG2 is another which has much better compression ratios, but it is somewhat encumbered by patents (see https://en.wikipedia.org/wiki/JBIG2).

[coulonnus]

Perhaps Procedures Nos.1-4 here http://imslp.org/wiki/IMSLP_talk:Scanning_music_scores can be of service to you.

[coulonnus]

I obtain about the same size as Procedure No.4 like this: convert -threshold 50% -compress group4 p*.jpg yourscore.pdf
See the freeware links in the URL of my previous message.

[coulonnus]

Forgive the off topic, but if you crop the introduction of Marguerite Labori (1864–1952) your score will be PD EU.

[coulonnus]

I obtain about the same size as Procedure No.4 like this: convert -threshold 50% -compress group4 p*.jpg yourscore.pdf

Oh, I just learnt something: if you have more than 9 pages, number them with leading zeroes. The page order in the pdf is that obtained with dir/on p*.tif

[nfn]

Thanks for all the suggestions.

Re the incidental suggestion about deleting the introductory note by Marguerite to overcome copyright, it's an interesting idea, but I'm not sure it's correct. The title page says "Transcribed and with notes by [her]". Some of those notes appear within the scores, and her transcription work might already make it PD EU; also I'm not at all sure that I should delete that introduction. But thanks for the idea!

Re the methods to reduce file size: I had looked into that question some years ago before starting this project and had found the various sophisticated methods a bit off-putting (certain software needed just for this purpose, even though freeware; results that would be uncertain in advance; and quite a lot of special study and work required -- I thought it was tempting but not quite worth the effort). I see from the present comments that there is unfortunately still no simple and easy way to get the desired results.

One fairly new feature I've noticed is this possibility in Adobe Acrobat X Pro: Edit -> Preferences -> Convert to PDF -> TIFF -> Monochrome Compression: CCITT G4 -> Edit settings. However, I have so far found no change in my resulting pdf file sizes by taking that approach.

As I cannot allocate the time required to delve into and implement sophisticated methods for this purpose (much as I would like to), I think I'll leave it for a while, and hope that a simple and easy way will eventually become available for non-specialists such as myself.

Nigel.

[coulonnus]

One fairly new feature I've noticed is this possibility in Adobe Acrobat X Pro: Edit -> Preferences -> Convert to PDF -> TIFF -> Monochrome Compression: CCITT G4 -> Edit settings. However, I have so far found no change in my resulting pdf file sizes by taking that approach.

If you're doing the tif's after the pdf it's a bit late! Can an Adobe Acrobat X Pro expert help please?