Hallo Nachus001, I noticed your recent uploads of Cambini quartets scanned by BNF, and you invest some time to crop and remove empty pages. However the pdf downloads offered by BNF are not very good (1000 pixel), and they seem recently to have reduced even this low resolution!
Very high resolution greyscale scans can be accessed on screen using the zoom (the icon on the far left): This is not tile-based but coordinate-based: Zoom to max resolution, then select the lower right corner for export. In the url, you can then change the coordinates to display the full page: From for example "5715,4205,41,37&save" to "0,0,5900,4500&save" (if your coordinates exceed the scanned image a little, the max scan will be displayed). You can save each high resolution image (on my mac by ctrl-click) as jpg file, leaf through the work by changing the page marker in the url, (e.g. f1 to f54). Not as convenient as an automatic download, but pretty fast, and something where imslp can easily improve on the direct download option from Gallica. And then I figured out how to use the command-line program cURL with the utility "Terminal" on my Mac (Thanks to Boccaccio for the command) - cURL is also available for Windows: In "Terminal" or a similar windows utility type (or copy/paste) curl 'http://gallica.bnf.fr/proxy?method=R&ark=btv1b9058527j.f[1-58]&l=6&r=0,0,5900,4500&save' -o 'Cambini_6_Quatuors_Op_3_#1.jpg', and soon you will find all 54 high resolution jpg's on your hard disk. You can then remove the empty pages, crop and batchconvert to pdf. Since the original scans already have some jpg-compression, it is possible to further halve size by saving as pdf with a jpg-compression of 4 or 5 without decreasing quality noticably. See my Devienne Sonata or Telemann Der getreue Musikmeister uploads.
If you have enough spare time, you may investigate this - but it is difficult to keep up with the speed new material is put online by libraries like BnF, SBB, SLUB... Just to quickly upload the offered download file may be much more time efficient. --Kalliwoda 19:46, 12 July 2011 (UTC)
Hi Kalliwoda. Great suggestion the " curl " didn't knew it. I usually do as you say. I go to the document page, select the "view" and download dragging page by page to a folder in my machine. Then I cut the image borders (that usually have paper ripped off ) with the MS Photo editor, and join all with the Solid Converter PDF. I do this way, because I judge the quality of individual pages are pretty fine and readable to upload to IMSLP. Contrary to the awful looking ~2Mb download for a document that BNF offers (which at this day are plain crap).
I'll investigate on this
Hi Kalliwoda: I noticed that you have to detect where are the extremes of the pic with this last methods. What I did with success was to enter the document page, go to first page, ask for the highest resolution (cheek against sheet), then use the navigator at left and go to bottom right corner of the document, then ask (with firefox) the "image properties" of the extreme bottom right jpg tile. Replaced the data fields as you pointed, but take as limit the information of that bottom right tile as the max x-y (y-x in fact) pixel data which would be ........&l=6&r=6144<<y>>,4096<<x>>,256,256 . So, In the string the you provided me I replaced &r=0,0,5900,4500&save with &r=0,0,6144,4096&save, and Voilá. That's it (I downloaded a 5Mb per page file, which is enormous, but can be downscaled to a decent definition and size. Apparently the size of pages doesn't change between pages. I tested it with a full page and a contigous blank page, with no cropping at all.