Further genpdf testing

[Jonathan Hung]

Hi Hasan,

I have been testing genpdf and discovering some bugs.

1. Type 2 PDF generation does not produce good results for non-computer generated documents.
Test data:
http://source.fluidproject.org/svn/design/decapod/testing-images/zip/decapod-testing-set2.zip

2. The following image generates a blank Type 1 and Type 2 PDF.
http://dl.dropbox.com/u/11522887/Newspaper-corrected.png

3. Newspaper layout causes poor text generation. Is this expected for complex layouts or should we expect better results?
http://dl.dropbox.com/u/11522887/Newspaper1.png

Question:
Is the resolution / DPI for this test data sufficient or should we be using higher resolution?

- Jon.

[E^3]

Hi Brethren,

Actually, If book scanning is our concerns:
generating pdf ,image resolutions , conversion , I am wondering
why we need this to be more complicated instead of more practicability

I don't know how the issues making it stagnant;but we can do it all
using open source /software
1)image image magick
2)pdftk
3)pdfconcat
4) cuneiform,tesseract,hocr2pdf
why coding genpdf if you can call via script in python?

My suggestion , why not the 3D dewarping it could be more promising
and useful
in our scanning purpose using Decapod.

Thanks

e^3

On Oct 27, 1:41 am, Jonathan Hung <jh...@ocad.ca> wrote:
> Hi Hasan,
>
> I have been testing genpdf and discovering some bugs.
>
> 1. Type 2 PDF generation does not produce good results for non-computer
> generated documents.

> Test data:http://source.fluidproject.org/svn/design/decapod/testing-images/zip/...
>
> 2. The following image generates a blank Type 1 and Type 2 PDF.http://dl.dropbox.com/u/11522887/Newspaper-corrected.png

>
> 3. Newspaper layout causes poor text generation. Is this expected for

> complex layouts or should we expect better results?http://dl.dropbox.com/u/11522887/Newspaper1.png