Page 1 of 1

"eps3:" creates 32× larger file than "eps2:"

Posted: 2015-04-25T06:45:59-07:00
by feklee
Yes, for 2015-04-25+02_input.pdf the difference is a factor of 32:

Code: Select all

$ convert 2015-04-25+02_input.pdf eps2:output2.eps
$ convert 2015-04-25+02_input.pdf eps3:output3.eps
$ ls -sh1 *
128K 2015-04-25+02_input.pdf
 60K output2.eps
1.9M output3.eps
$ identify *
2015-04-25+02_input.pdf PDF 595x841 595x841+0+0 16-bit Bilevel Gray 63.1KB 0.000u 0:00.000
output2.eps[1] PS 595x841 595x841+0+0 16-bit sRGB 18.8KB 0.000u 0:00.000
output3.eps[2] PS 595x841 595x841+0+0 16-bit sRGB 19KB 0.000u 0:00.000
Is that expected? Shouldn't eps3 be superior to eps2? Why does "identify" report a different size than "ls"?

Re: "eps3:" creates 32× larger file than "eps2:"

Posted: 2015-04-27T06:56:34-07:00
by feklee
Should I report this as a bug?

Re: "eps3:" creates 32× larger file than "eps2:"

Posted: 2015-04-27T08:06:00-07:00
by feklee
Update: What identify shows is not the resolution of the embedded image. It's probably the resolution of the respective document rendered at a certain DPI level.

Resolution of the embedded image can be checked with pdfimages. For that to work, convert the EPS files first with ps2pdf. The result: Both prefixes, eps2 and eps3, reduce the resolution considerably, namely to the (rendered) resolution that identify shows for the PDF.

It seems better to convert PDF to EPS using pdftops -eps, which leaves the resolution of the embedded image untouched.