| | <<<Back 1 day (to 2012/07/21) | 2012/07/22 |
| cryptopsy | can i offset pages in a pdf file with ghostscript, one at a time? | 08:11.54 |
| flnev | Hi, I am trying to improve the pdf files generated by a LaTeX package I wrote. If someone is willing to give me some hints, let me describe the problem. | 19:43.17 |
| | I want to draw some lines with arrow tips that consist of (pieces of) arrow characters from a font. The line itself is a graphic element, and not made from juxtaposing dashes. Now, pdf viewers and printers apparently render graphics and text using slightly different methods (for antialiasing, I suppose), so the transition between the line and the arrow tip is visible sometimes. This png http://i.imgur.com/7TEO1.png shows what I | 19:49.27 |
| | mupdf renders the pdf as I expected, but evince renders the vertical arrows strangely, depending on the zoom level, and Adobe Reader seems to do an even | 19:49.55 |
| | an even worse job. | 19:50.14 |
| | When printing, what I get in some cases is lines (graphical elements) that seem much thicker than corresponding lines in font glyphs, even though they have nominally the same line width. This can be fixed by switching to "high quality" printing mode. | 19:50.28 |
| | So my question is: Is there anything I could try to improve the display of this kind of graphic? | 19:50.46 |
| | To see the graphics I wanto to produce in a pdf file, see the bottom of page 10 in http://mirrors.ctan.org/graphics/pgf/contrib/tikz-cd/tikz-cd-doc.pdf | 20:02.13 |
| | If you have the time to look into it, I'll check the channel log in a couple of days; or send me an email. Thanks! | 20:03.12 |
| ray_laptop | cryptopsy: your question from the log... you can use Ghostscript to offset pages by setting up an "Install" procedure using the 'setpagedevice' operator. This modifies the matrix when the device does an 'initgraphics' | 20:45.28 |
| | cryptopsy: For example: from th command line: -c "<< /Install { 20 30 translate } >> setpagedevice" | 20:46.51 |
| | cryptopsy: will offset the page by 20/72 inch in X and 30/72 inch in Y | 20:47.24 |
| | flnev: For the logs. Relying on a symbol to match up with a line will be problematic since font rendering does things like 'snap to grid' and the line weight in the font probably won't match the line. | 20:49.30 |
| | flnev: for something as simple as an arrow head, just doing it with line graphics stroke and/or fill will be more pleasing, IMHO (and probably more efficient) | 20:50.48 |
| flnev | ray_laptop: I agree that doing line strokes for the arrowhead is better in many regards, but I want to have arrows matching the document's font... The line weight of the font is a parameter TeX is aware of, so that's easy to obtain | 21:21.48 |
| | is it somehow possible to disable 'snap to grid' and similar text optimizations? | 21:22.21 |
| | I can get better results if, instead of clipping and placing the glyph, I use the glyph as clipping path for a big black box. This seems to trick pdf viewers into treating the glyph as graphics and not text... | 21:24.41 |
| ray_laptop | I need to check with HEAD code (before my patch), but it's hard to imagine how NumRenderingThreads > 1 didn't SEGV often for multi-page input files. | 21:37.24 |
| | should be easy enough to check now that I know the conditions that (should) cause the failure | 21:37.58 |
| | but at least the garbled output cases and SEGV's are fixed now for a sampling of BandHeight values and files and NumRenderingThreads | 21:38.47 |
| | with multipage files, even devices other than psdcmyk should have failed (depending on NumRenderingThreads and number of bands) | 21:39.46 |
| | bbiaw | 21:40.06 |
| henrys | the meeting dates are sept 8/9 (Saturday and Sunday) ??? | 22:43.06 |
| cryptopsy | do pages must be splitted in a file before offsetting said page? | 23:22.32 |
| | the command has the form gs -starpage -endpage ... | 23:22.41 |
| | am wondering if i have to burst it before offsetting pages, then put it back together | 23:23.11 |
| | it seems that would be a better idea in the end anwyways because it would decrease the number of output files written by 'n-1' | 23:23.37 |
| | burst->offset->write all in one shot | 23:23.53 |
| | instead of (offset->write) times 'n' | 23:24.06 |
| | Forward 1 day (to 2012/07/23)>>> | |
IRC Logs
