GNOME Bugzilla – Bug 331781
low resolution graphic appears smoothed
Last modified: 2008-01-13 16:52:12 UTC
Please describe the problem: I have a graphic showing a low resolution image with graph axes superimposed, saved as both EPS and PDF. When I view either file with Evince, the graphic appears smoothed. When I view the PDF with Acroread, I can see the individual pixels. I have uploaded three files to demonstrate the problem (the PNG shows the graph as I want it to appear): http://www.magres.nottingham.ac.uk/~magill/pixels/pixels.png http://www.magres.nottingham.ac.uk/~magill/pixels/pixels.pdf http://www.magres.nottingham.ac.uk/~magill/pixels/pixels.eps I am using Gentoo Linux, fluxbox, evince 0.5.0, poppler-0.5.0-r5 (cairo) Steps to reproduce: 1. open pixels.eps 2. open pixels.pdf 3. compare to pixels.png Actual results: The image appears as a smooth gradient. The PDF shows a white border on the bottom and right edges of the image and colourbar. Expected results: see pixels.png - I expect to see individual pixels in the image, and no white border on the PDF. The PDF file displays as expected in Acroread. The EPS file appears correct if displayed with xv. Does this happen every time? yes Other information:
Created attachment 68221 [details] Additional test case... Hi, I can confirm similar behavior with evince 0.5.2 on ubuntu and evince 0.5.3 on debian experimental, but only with postscript files in evince and not with pdf (though that might be dependent upon my workflow in producing each). Example eps is attached to this posting (smoothing behavior apparent in evince 0.5.x and correct behavior in evince < 0.4.x and gv). Larger canonical source for graphic and paper (in which it is embedded) is available at: http://arxiv.org/abs/cond-mat/0606014 Please note that this bug is especially CRITICAL for scientific communication, where such smoothing obviously changes the underlying data. Thank you kindly, Daniel
This bug is still present in evince-0.5.3 using poppler 0.5.3 (cairo).
This bug is still present in Evince 0.6.0 I want to emphasize here that because of this bug, evince cannot be used for reading scientific papers. It changes the underlying data, sometimes quite significantly. No self respecting scientist can afford to read pdf papers in a viewer that significantly modifies the figures. Adobe Reader 7 does not have this bug, and that's what we are forced to use on Linux, to view pdf files. Arthur, please update the version of this bug report to 0.6 and raise the severity to "Major".
To demonstrate the graveness of this bug please look at Fig. 3 in this pdf paper: http://arxiv.org/pdf/cond-mat/0606014 in evince and in Acrobat Reader. You can see that the data looks very different.
I can confirm I'm still seeing the same problem with Evince 0.6.0 and Poppler 0.5.4 (Cairo), both using the standard Gentoo packages. I work in MRI so this bug is a showstopper for me too, but I don't think it really counts as a major bug for most users. It would be nice if one of you Gnome folks would confirm it though...
hmm, this looks like a poppler or cairo issue, no?
I'd say that's a fair guess, but I don't know which one, hence reported under Evince hoping someone with more knowledge would point the report in the right direction. The end result is still that Evince is unusable for me.
I've just forwarded this to poppler (https://bugs.freedesktop.org/show_bug.cgi?id=9860). Thanks Arthur, I hope it will be fixed soon.
Thanks for checking this out Carlos. Keep up the good work with Evince. Let's hope Poppler sort this out soon and I can go back to using Evince.
*** Bug 484340 has been marked as a duplicate of this bug. ***
*** Bug 509132 has been marked as a duplicate of this bug. ***
I can reproduce this bug in evince 2.20.2 using poppler 0.6.2 (cairo), please update the version field of this bug.
it is even in evince 2.21.1 and poppler 0.6.3 (cairo)