Slow RenderPageToFile for TIFF

Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic

   Hello QuickPDF community.

I use QuickPDF 1811 x64 to extract TIFF pages from PDF, then process them with tesseract
and merge the PDF FullText Only Document again with the orignal PDF.

That works well, my Problem is just that the RenderPageToFile and RenderDocToFile seems quite slow for big Documents and TIFF

Here is my Sample Document:
https://www.file-upload.net/download-14402446/LoreIpsum106Pages.pdf.html

With both TiffLzw and TiffG4 and with RenderPageToFile to individual pages and RenderDocToFile to a single file they both take arround 44 Seconds when converted with 300 dpi.

I would prefer to render every page to an indivdual file (RenderPageToFile).
If I use Ghostscript 9.53.3 with -r300x300 -sDEVICE=tiffg4 (so the same) it takes 5,8 seconds.

Can I accerate that?
For example is RenderPageToFile Multi Threading Capable? Could I load the document and call RenderPageToFile in different worker threads?

I am using C# + net5.0 and the x64 DLL Version DebenuPDFLibrary64DLL1811.dll

Thank you for your help and Merry Christmas

Edited by Michi91 - 20 Dec 20 at 3:05PM

Author	Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic
Michi91 Members Profile Find Members Posts Beginner Joined: 20 Dec 20 Location: Munich Status: Offline Points: 3	Post Options Post Reply Quote Michi91 Report Post Thanks(0) Quote Reply Topic: Slow RenderPageToFile for TIFF Posted: 20 Dec 20 at 2:59PM
	Hello QuickPDF community. I use QuickPDF 1811 x64 to extract TIFF pages from PDF, then process them with tesseract and merge the PDF FullText Only Document again with the orignal PDF. That works well, my Problem is just that the RenderPageToFile and RenderDocToFile seems quite slow for big Documents and TIFF Here is my Sample Document: https://www.file-upload.net/download-14402446/LoreIpsum106Pages.pdf.html With both TiffLzw and TiffG4 and with RenderPageToFile to individual pages and RenderDocToFile to a single file they both take arround 44 Seconds when converted with 300 dpi. I would prefer to render every page to an indivdual file (RenderPageToFile). If I use Ghostscript 9.53.3 with -r300x300 -sDEVICE=tiffg4 (so the same) it takes 5,8 seconds. Can I accerate that? For example is RenderPageToFile Multi Threading Capable? Could I load the document and call RenderPageToFile in different worker threads? I am using C# + net5.0 and the x64 DLL Version DebenuPDFLibrary64DLL1811.dll Thank you for your help and Merry Christmas Edited by Michi91 - 20 Dec 20 at 3:05PM

Ingo Members Profile Find Members Posts Moderator Group Joined: 29 Oct 05 Status: Offline Points: 3524	Post Options Post Reply Quote Ingo Report Post Thanks(0) Quote Reply Posted: 20 Dec 20 at 11:11PM
	Hi Michi :) is it necessary to render a page which has originally 72 dpi with 300 dpi? Quality won't be better i think ;-) Making 106 image files out of 1 document needs time. If the results will be still okay for you i would try to render with 150 or less dpi... Another option is to select another render engine like cairo for example, AGG or PDFium for better results -> https://www.debenu.com/docs/pdf_library_reference/SelectRenderer.php Here you'll find the developer guide and much more: http://www.quickpdf.org/ Cheers and welcome here, Ingo
	Cheers, Ingo

tfrost Members Profile Find Members Posts Senior Member Joined: 06 Sep 10 Location: UK Status: Offline Points: 437	Post Options Post Reply Quote tfrost Report Post Thanks(0) Quote Reply Posted: 21 Dec 20 at 8:48AM
	I tried rendering your file with our application which converts to monochrome 200dpi multi-page TIF, and it took about 30 seconds. It was not much different with PDFIUM. Of course a proportion of this time was spent dithering to bi-level; a color TIF should be faster, although there is more data to write. And there are doubtless many other variables which differ between your environment and mine. The only reliable way to determine whether it is better to render pages in different threads is to try it.

Michi91 Members Profile Find Members Posts Beginner Joined: 20 Dec 20 Location: Munich Status: Offline Points: 3	Post Options Post Reply Quote Michi91 Report Post Thanks(0) Quote Reply Posted: 21 Dec 20 at 6:05PM
	Hello Ingo and tfrost, thank you for your quick answers. Ingo: I guess it is not required ;) But can I quickly check what is the max dpi used in an document? Checking all images in the file won't make it quicker? http://www.quickpdf.org/forum/renderpage-dpi-from-current-pdf_topic2893.html I was just suprised that Quickpdf takes 44 seconds and ghostscript COMMANDLINE 5,8 seconds for 300 dpi and the same document. @tfrost: I think I will try multithreading, if that doesn't work I think about a way with Ghostscript. Merry Christmas, I would come back to you both in January. :) Edited by Michi91 - 21 Dec 20 at 6:05PM

Ingo Members Profile Find Members Posts Moderator Group Joined: 29 Oct 05 Status: Offline Points: 3524	Post Options Post Reply Quote Ingo Report Post Thanks(0) Quote Reply Posted: 22 Dec 20 at 11:53AM
	Hi Michi, render page to file doesn't mean image extraction - due to this a din a4-page you can render with 72 dpi. You can use more dpi but this won't raise the quality of the image. Rendering only paint the pdf-page onto an image file - that's all. QuickPDF offers real functions to extract embedded image files out of a pdf but that's another thing. Real multi threading won't work so i think you should start trying Ghostscript.
	Cheers, Ingo