Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
On Mon, 7 May 2018 20:13:32 +0200 Andreas Bilke wrote: > Evgeny Stambulchik found a nicer solution for this problem. The annotations > from the PDF are removed from the document (at least in the in-memory > version). > > Therefore we can use render() again and have nice pixel graphics. > > We released v4.1.2 to reflect this change. Sounds nice: I am looking forward to giving it a try. If I manage to find the time, I will rebuild and test it. Otherwise, I will install the new version, as soon as it is packaged for Debian unstable. Thanks to you and to Evgeny for working on a solution! Bye. -- http://www.inventati.org/frx/ There's not a second to spare! To the laboratory! . Francesco Poli . GnuPG key fpr == CA01 1147 9CD2 EFDF FB82 3925 3E1C 27E1 1F69 BFFE pgpT33gzkRyMA.pgp Description: PGP signature
Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
Evgeny Stambulchik found a nicer solution for this problem. The annotations from the PDF are removed from the document (at least in the in-memory version). Therefore we can use render() again and have nice pixel graphics. We released v4.1.2 to reflect this change. signature.asc Description: PGP signature
Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
> Perhaps as an interrim measure, maybe a command line option and/or hot > key to switch between modes would make sense? I implemented https://github.com/pdfpc/pdfpc/pull/342. Could you have a look if it satisfies your needs? If a page has annotations, it is rendered without visible icons of the annotations, in all other cases it uses the old way. signature.asc Description: PGP signature
Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
So we have a choice right now: low resolution, or missing annotations. If I had to choose, I'd pick the missing annotations, because I use a lot of images in my talks and don't use annotations at all. That's just me, but following my compelling usage survey (n=1) I'll leave the package this way for now. Perhaps as an interrim measure, maybe a command line option and/or hot key to switch between modes would make sense? Pending a proper fix to the underlying PDF library, of course. Or an ugly hack involving rendering each page twice, once in each mode. --Barak.
Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
On Tue, 1 May 2018 10:21:55 +0200 Andreas Bilke wrote: > I think this patch opens a new bug (see my previous post) and should not > count as a bug fix for this issue. Dear Andreas, I fully acknowledge that my single line patch is not a proper fix for the regression. I mean: it eliminates the poor-raster-image-rendering regression, but introduces a regression in the support for notes from PDF annotations (which, as far as I can tell, is a more recent feature). I fully agree with you that a proper fix for the poor-raster-image-rendering regression should fix the issue, without introducing other regressions. Maybe this requires improving the Poppler library: have you considered getting in touch with the Poppler developers? However I think this temporary patch should be kept in the Debian package (and possibly even adopted in the upstream pdfpc project) until a proper fix is devised and implemented. A newer feature (notes from PDF annotations) should not be introduced at the cost of breaking an older one (good raster image rendering). Hence, I think it's better if pdfpc makes a little step back (temporarily losing support for notes from PDF annotations) in order to keep a good raster image rendering capability. I hope this reasoning makes sense to you. Thanks for your helpfulness! P.S. for Barak: you have probably already read this in my long reasoning above, but let me say it explicitly: I can confirm that pdf-presenter-console/4.1-4 eliminates the poor-raster-image-rendering regression; thanks for uploading a new Debian revision to unstable so promptly! -- http://www.inventati.org/frx/ There's not a second to spare! To the laboratory! . Francesco Poli . GnuPG key fpr == CA01 1147 9CD2 EFDF FB82 3925 3E1C 27E1 1F69 BFFE pgpK7pY5_NrcC.pgp Description: PGP signature
Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
I think this patch opens a new bug (see my previous post) and should not count as a bug fix for this issue. signature.asc Description: PGP signature
Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
Thanks for the patch. I uploaded a version with it incorporated; if you could confirm that this issue is fixed that would be great. Cheers, --Barak.
Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
I can confirm that we have degraded quality for pixel graphics so far. The call to `render_for_printing_with_options` is the "cause" of the problem. If I switch back to `render` everything is fine. But: pdfpc supports notes from PDF annotations, and if we render a PDF with annotations with `render` we see a symbol at the annotation position in pdfpc. Therefor we have `render_for_printing_with_options` where we disable the icon for the annotations. With a quick look at the poppler API I was not able to find another function call to hide annotations but keep the good rendering. signature.asc Description: PGP signature
Bug#888677: pdf-presenter-console: [regression] raster image rendering is really poor
Control: tags -1 + patch On Sun, 28 Jan 2018 16:43:39 +0100 Francesco Poli (wintermute) wrote: [...] > When the PDF presentation includes raster images, their rendering > on screen (at the final resolution, requiring a scaling in almost > all cases) looks very poor, as if some aliasing failed to be applied > or something like that... Hello, after talking to the upstream developer (who reads us in Cc) and being asked to find the regression point with git-bisect, I managed to pinpoint the [upstream commit](https://github.com/pdfpc/pdfpc/commit/430d38e06cea152e74d26b9bc5fe6aa3a5f86169) that introduced the regression. Moreover, I found out the single line change that introduced the regression. The attached patch eliminates the regression for me. I have already informed Andreas about this finding. I am not sure I understand the precise effects of reverting this single line change (perhaps some capability related to notes could be lost). I hope Andreas may prepare a proper fix with no collateral damage. In the meanwhile, I suggest to apply the simple patch to the Debian package in order to close this bug report. Thanks for your time. -- http://www.inventati.org/frx/ There's not a second to spare! To the laboratory! . Francesco Poli . GnuPG key fpr == CA01 1147 9CD2 EFDF FB82 3925 3E1C 27E1 1F69 BFFE revert_one_line_from_commit_430d38e06cea152e74d26b9b Description: Binary data pgphgVYFdV6vh.pgp Description: PGP signature