| <<<Back 1 day (to 2016/02/15) | 20160216 |
Asuran | kens: you were maybe right yesterday, it seems the difference is due ghostscript 9.14 (on the better looking pdf) and 9.18 on the worse looking | 10:25.31 |
| in the worse looking the font is really blurry were on the 9.14 one the font is much sharper | 10:26.11 |
kens | This is not an effect I see on the PDF files you posted | 10:26.47 |
Asuran | kens: hmm than its adobe reader thats bad | 10:42.25 |
| the renderer you use uses gs i gues? | 10:42.36 |
| well with the 9.14 it looks better then with the 9.18 on adobe reader | 10:42.57 |
| idk is this normal or bug | 10:43.08 |
chrisl | Is Reader "smoothing" the display? | 10:44.53 |
Asuran | yep and without it looks terrible | 10:45.14 |
kens | I was looking at the PDF files using Acrobat Professional and Acrobat Reader | 10:45.15 |
Asuran | kens: please tell me youre kidding, please... | 10:45.31 |
kens | But *both* the files you posted were created using PDF24 and Ghostscript 9.14 | 10:45.45 |
Asuran | kens: i running standard windows 10 setup, no fancy stuff | 10:45.54 |
| i check it | 10:46.30 |
| right my mistake: https://www.docdroid.net/file/download/pe6Vv4Z/gs914.pdf && https://www.docdroid.net/file/download/phTEF3j/gs918.pdf | 10:51.16 |
kens | It would be helpful to use the same source for each file, comparing different documents is harder | 10:52.00 |
| Ah no, that's a bug in pdf.js :-) | 10:52.16 |
| I still do not see any difference, where on the document are you looking ? | 10:52.53 |
chrisl | There's no visible different between them in Evince..... | 10:52.53 |
Asuran | well on the 9.14 the font is sharper, even in webview of docdroid | 10:53.26 |
kens | Not that I see | 10:53.32 |
Asuran | at the text | 10:53.33 |
| hmm | 10:53.38 |
| but i see it clearly, i dont complain for fun ^^ | 10:53.49 |
kens | No, the text is identical at all zoom levels in Acrobat Professional | 10:53.57 |
Asuran | honestly | 10:53.59 |
kens | Perhaps you should take a screenshot of the two files open and tkiled | 10:54.12 |
| I see no difference using Acrobat and nor does Chris using Evince. That would suggest there is no difference | 10:54.32 |
chrisl | I didn't even see a difference in the docdroid display, but I didn't look that closely at it | 10:54.49 |
| And now it goes straight to download....... | 10:55.10 |
kens | For me pdf.js (which is what Firefox is set to use) displays the 9.18 incorrectly, but not bl;urred | 10:55.49 |
Asuran | http://www2.pic-upload.de/img/29757960/gs.png | 10:56.42 |
kens | Well looks to me like these files were not creted the same way | 10:56.44 |
Asuran | its just a bit | 10:56.49 |
kens | ANd its not visible here like that | 10:57.11 |
Asuran | and if my eyes are not bad, the right one (9.18) looks bit darker | 10:57.34 |
kens | I don't thnk you used the same source for these 2 documents, or the same command line | 10:57.42 |
Asuran | the left one looks fine to me | 10:57.48 |
| the gs9.18 is based on the command you suggest mostly | 10:58.09 |
| and the 9.14 is the pdf24 one | 10:58.16 |
kens | These files are using different fonts (or at least different names) and have a different encoding applied as well as being at a different resolution | 10:58.20 |
Asuran | same font | 10:58.28 |
| hmm | 10:58.37 |
| the 9.18 uses what? some font from gs? | 10:58.46 |
| because since the 9.14 looks nicer | 10:58.56 |
kens | One is called TimesNewRoman,Bold the other is called TT54o00 | 10:58.59 |
Asuran | the TT54o is on 9.18? | 10:59.13 |
kens | It doesn't matter | 10:59.24 |
| The point is still that I cannot reproduce your result using Acrobat | 10:59.37 |
| Your files look identical to me on Acrobat | 10:59.53 |
Asuran | if you zoom everything looks nicer... i reference at 100% view | 11:00.21 |
| and there the 9.14 looks bit sharper | 11:00.36 |
kens | Well given that Acrobat 'smooths' the text using anti-aliasing it would not be too surprising that there's a differemce. Not least the fact that you have not used the same scaling on each file. | 11:01.19 |
Asuran | scaling? i didn't specified any scaling i guess | 11:01.40 |
kens | But at 107% (which is what Acrobat uses when I tile these files) the display is identical | 11:01.48 |
| You must have specified a different resolution to one of these cases. | 11:02.00 |
| Whch means the CTM is different | 11:02.06 |
| And therefore the scaling is different | 11:02.13 |
Asuran | in the docs i didn't saw anything to specify this in ps2pdf doc | 11:02.47 |
| and the output is pretty much same, libreoffice, print and then print as ps or with pdf24 | 11:03.10 |
kens | You were previously specyfing the resolution with -r | 11:03.11 |
Asuran | not more | 11:03.16 |
kens | That isn't the point | 11:03.23 |
| When you alter the resolution you change the scale matrix. Now overall the result is no different. 10*72 is the same as 720 for example | 11:03.53 |
chrisl | Results of Acrobat's "Export to PNG": http://ghostscript.com/~chrisl/gs9.14_Page_1.png and http://ghostscript.com/~chrisl/gs9.18_Page_1.png | 11:04.00 |
kens | However, when you apply further scaling then you can get 'dropouts' because the maths works out differently | 11:04.40 |
Asuran | zoom to 100% you see there a difference which maybe indicate what i see on 100% | 11:04.55 |
| @chri | 11:04.57 |
| its tiny difference but its there | 11:05.09 |
kens | No I do not see such a difference | 11:05.27 |
Asuran | okay i understand | 11:05.46 |
| well then thanks guys | 11:05.52 |
chrisl | Can you be *very* specific - that's an entire page | 11:05.59 |
Asuran | chrisl: its over, believe what you want, not my problem ^^ sorry for bothering you guys | 11:06.17 |
kens | If you want to compare output then you must be certain that the two outputs have been prepared in the same way. Yours haven't | 11:06.31 |
Asuran | as i said believe what you want sir, its your choice | 11:06.48 |
kens | Its not a belief its a standard fact, you can't compare two thnkgs which were created differently. I do not believe that the version of Ghostscript makes any difference. The difference is that the two inputs, or the characteristics of the conversion are different. Since I can point to filenames being different and matrix values being different, I can clearly show that you did not follow the same process when comparing '9.14' against '9.18' | 11:09.26 |
| I mena font names, not filenames.... | 11:09.46 |
| Oh, and again ths is nothing to do with ICC profiles either | 11:10.11 |
Asuran | kens: yea i can believe this with ICC profile now, but in scribus once i changed color profile and my fonts did looked much thicker just of that so... | 11:10.48 |
kens | If you do the two conversions (using the two versions of Ghostscript) in the same way then I believe you will get the same result. | 11:11.22 |
| If you don't then I would be prepared to look into it. But not when the two executions differ by more than just the version of Ghostscript being used. | 11:12.03 |
| And frankly, the anti-aliasing applied by Acrobat is what is causing your 'blurriness' | 11:12.32 |
Asuran | i cant see what pdf24 gives to ghostscript | 11:15.38 |
| but they get same input from libreoffice | 11:15.46 |
| since the way is the same | 11:16.07 |
| if so, why looks the 9.14 one from pdf24 better then the one from 9.18 with nearly youre recommandation or command params | 11:17.00 |
kens | You can, however, use different versions of Ghostscript yourself, and show that the result is the same, regardless of the version. I believe you can also use different versions of Ghostscript with pdf24 (they offer several version for download on their web site) and again you could show that there was a difference (if tghere was) | 11:17.05 |
| 'nearly'..... | 11:17.15 |
Asuran | i tried this too kens | 11:17.22 |
| but it sometimes fails then and idk, i better avoid this | 11:17.34 |
| i prefer using gs from command line, like this you actually control things | 11:17.51 |
kens | What do you mean 'fails' ? Have you reported a bug ? | 11:18.05 |
Asuran | why bug? its different gs version | 11:18.16 |
| they dont have to work together if not explicit made by the author | 11:18.31 |
kens | And again, while it may be that Acrobat's anti-aliasing happens to work out better for a specific font at a specific size at a specific location on the page, and specific Acrobat zoom factor, ths is not going to be tha case for a different combination | 11:18.39 |
| Anti-aliasing makes fonts blurry there's no way to avoid that, in its in the nature of the process. | 11:19.28 |
| jiggle the position of the glyphs slightly and you'll get a slightly different result | 11:19.44 |
Asuran | yep but still the 9.14 looked better | 11:19.50 |
kens | 'for ths file' | 11:19.55 |
Asuran | why i screenshoted it for you | 11:19.56 |
| as you requested hehe | 11:20.07 |
kens | It doesn't on my setup | 11:20.11 |
| THey look identical. | 11:20.16 |
Asuran | nope on the screenshot you see: no | 11:20.31 |
kens | Though I will believe that if I played about with Acrobat zoom factors enough I could get *either* to look better than the other | 11:20.36 |
Asuran | the right one got some slight blurryness | 11:20.40 |
| its not extreme but enough to annoy and make reading hard | 11:20.56 |
| especially long reading | 11:21.03 |
kens | Like I sad, its the anti-aliasing from Acrobat. There is no point in worrying about ths, it will vary from file to file and even from glyph to glyph in the same file | 11:21.09 |
Asuran | yes but as i said its looks better in the 9.14 | 11:21.39 |
| even with santi aliasing | 11:21.42 |
kens | And teh difference is due to the way either the input or the command line has been constructed | 11:21.43 |
chrisl | They are both blurry..... | 11:21.49 |
Asuran | chrisl: yea but the left one is atleast acceptable compared to right one | 11:22.07 |
kens | OK we're going in circels here. IMO there is no bug, except that you are using anti-aliasing. I don't propose to carry on investigating this, its taken way too much of my time already | 11:22.13 |
Asuran | i hope youre kidding kens | 11:22.31 |
kens | No | 11:22.36 |
Asuran | because i like jokes | 11:22.37 |
| but not this ones | 11:22.41 |
chrisl | The left one has pixel dropouts in the glyphs | 11:22.44 |
Asuran | i accept you not responsible for creating for me a pdf creator | 11:22.53 |
| but the screenshot is clear | 11:22.59 |
| the commandline is nearly the same | 11:23.05 |
kens | The 'problem' is entirely due to the anti-aliasing | 11:23.15 |
Asuran | and cant be much other, except you didn't told me something | 11:23.17 |
kens | ANd the command line, or input, is absolutely not the same. YOu said yourself you cannot tell me the command line form pdf24, so you cannot say that the command line you used is the same | 11:23.49 |
Asuran | well that could only be if the left one would look as the right one | 11:23.51 |
kens | Forget pdf24. | 11:24.03 |
| Use two different versions of GS. Ssame inptu file, same command line. | 11:24.17 |
| THen show me a difference | 11:24.22 |
Asuran | alright | 11:24.26 |
| okay | 11:31.13 |
| youre right | 11:31.18 |
kens | So the difference is either in the input or, more likely, the command line. | 11:31.34 |
Asuran | command line fine, input was it | 11:31.44 |
| i used now pdf24 to create the ps file | 11:31.52 |
| works better | 11:31.55 |
kens | OK well that's slightly surprising to me, but as long as its the case | 11:32.01 |
Asuran | and i get now complainings about gs couldn't find fonts | 11:32.18 |
| and using instead nimbus | 11:32.24 |
| or so | 11:32.25 |
kens | Well that's not surprising | 11:32.38 |
| Hmm, actually it slightly is | 11:32.50 |
| What does the first few lines of the PostScript look like ? | 11:33.05 |
| Especially the line beginning %%Creator: | 11:33.18 |
Robin_Watts | Asuran: Can I ask what your interest in gs is? Are you a private user? Or are you a developer looking to integrate it with a product? | 11:34.04 |
Asuran | private | 11:34.09 |
| just want create pdf documents | 11:35.48 |
| pdf/a ones | 11:35.52 |
| i dont want pirate for it any software | 11:35.59 |
| feels better | 11:36.01 |
kens | If you like hte pdf24 result, it is free, you don't need to pirate it | 11:36.58 |
Asuran | it uses old gs version | 11:37.10 |
| i prefer latest stable for software | 11:37.17 |
kens | I'm reasonably sure it can be told to use aq newer vgersion, since they offer newer versions for download | 11:37.31 |
| But if you want to use GS yourself you are of course welcome. | 11:37.48 |
Asuran | in their log they use 9.14 because they got problems with 9.16 | 11:37.56 |
| look on pdf24.org | 11:38.00 |
kens | What I can't tell you is how pdf24 is getting the input, or the command line they are using | 11:38.03 |
Asuran | its better i can see command line | 11:38.18 |
kens | THey do offer 9.18 for download, so I assume they are happy with that | 11:38.20 |
Asuran | and i get the most best result hopefully when using latest stable | 11:38.30 |
| well pdf24 comes with 9.14 | 11:39.01 |
| and on ubuntu its still 9.16 i hate ubuntu | 11:39.15 |
| i hate all distros like ubuntu | 11:39.19 |
| i prefer on it the gui but the software versions are... | 11:39.37 |
| even on debian maybe better | 11:39.40 |
kens | Somewhat surprising, Ubuntu is usually up to date with our releases | 11:39.40 |
Asuran | nope its 9.16 and it wont get changed i guess | 11:39.54 |
| even in 16.04 its still 9.16 | 11:40.03 |
| look in software repo online browsing | 11:40.15 |
| archlinux is kinda nice but much to manage yourself | 11:40.29 |
| wow nice sharp fonts now atleast better then before | 11:46.40 |
kens2 | slaps kens around a bit with a large trout | 13:38.28 |
Asuran | double the kens | 13:38.34 |
| doubles the price | 13:38.37 |
| ^^ | 13:38.39 |
| but intresting the one who got slapped actually quits, weird | 13:39.08 |
malc_ | double KO | 13:39.33 |
Asuran | lol | 13:39.56 |
| nvm | 13:40.05 |
remi_ | hello, i'm tryning to find a way to perfom a rotation with mupdf in an android app, and i can't find a way do to it. is there a function for that ? | 16:40.23 |
Robin_Watts | remi_: A rotation of what ? | 16:43.16 |
remi_ | a rotation of a pdf page | 16:47.51 |
Robin_Watts | remi_: So you want to load a PDF and output the same PDF but with a page rotated? | 16:50.23 |
| Or you want to open a PDF and render it rotated? By an arbitrary angle, or my a multiple of 90 degrees? | 16:50.48 |
remi_ | i want to open a pdf and render it rotated wirh an angle of 90 degrees or 180degrees | 16:54.31 |
Robin_Watts | remi_: Ok, so the core certainly supports that. | 16:57.09 |
| whether the android classes that wrap the core do that is another question. | 16:57.28 |
| Are you basically taking our example android viewer and tweaking it? | 16:57.46 |
| Or are you writing code from scratch ? | 16:57.55 |
remi_ | We tweak (twoke?) the example Android viewer, just modifying the main activity for now | 16:58.48 |
Robin_Watts | The Android app uses some JNI code to expose the MuPDF core through the MuPDFCore class. | 16:59.08 |
| We don't expose the whole API, just a few pickled functions to let us do enough. | 16:59.26 |
| We have more general code in git now that offers a better JNI experience, but the app has not yet been updated to use it. | 16:59.53 |
| So you're going to have to make changes yourself. | 17:00.01 |
remi_ | ok, we'll dig in that | 17:00.43 |
Robin_Watts | This is not something we can spare time to help with at the moment (unless you're a commercial or supported customer?) | 17:00.54 |
| remi_: You are aware of the license terms of MuPDF, right? | 17:01.12 |
remi_ | yes, our app is under an Affero V3 | 17:01.32 |
Robin_Watts | Cool. | 17:01.41 |
| I like to check up front before people invest time and effort and then discover it's all wasted. | 17:02.06 |
remi_ | i-Parapheur on Google Play, MuPdf is even quoted in the licences | 17:03.08 |
| but we don't have any support contract | 17:03.27 |
kens | Goodnight all | 17:10.09 |
remi_ | bye | 17:24.55 |
Robin_Watts | sebras: Did tor8 go over your commits? | 18:19.48 |
sebras | Robin_Watts: we did do this face to face, yes. | 21:35.09 |
| Robin_Watts: we also looked into your explanation for newFragCb(), so now tor8 knows more about this one as well (unless he did already). | 21:35.47 |
| tor8: Robin_Watts: can you close http://bugs.ghostscript.com/show_bug.cgi?id=696004 as its fix was merged here: http://git.ghostscript.com/?p=mupdf.git;a=commitdiff;h=93699403812c87a8d96d0029f37134cf86e86205 | 22:19.27 |
Robin_Watts | sebras: What about the last 3 comments? | 22:45.36 |
sebras | Robin_Watts: oh! I didn't notice that those comments were recent. | 22:47.27 |
| Robin_Watts: hm. so it seems that something is still awry. will take a look. :/ | 22:47.53 |
| Robin_Watts: unfortunately I can confirm Pablos findings. not only that but I checked all possible vertical alignment alternatives in CSS and well. mupdf is all of the place. | 23:07.54 |
| Forward 1 day (to 2016/02/17)>>> | |