| | <<<Back 1 day (to 2019/03/10) | 20190311 |
| chrisl | mrd0ll4r: My first thought is to try removing the " -dTextAlphaBits=4 -dGraphicsAlphaBits=4" - I suspect those can have strange interactions with the transparency stuff. | 08:00.17 |
| kens | i3v, you can't do that with pdfwrite, and I can't see any reason you would want to. The -dDviceGrayToK and the vast majority of the colour management described in the PSDF you mentionm only affect rendering devices. The high level devices (pdfwrite etc) go to considerable effort to retain colours in the space they were originally specified. You can change all colour uage by setting the ColorConversionStrategy, but there's no way to modify j | 08:03.41 |
| mrd0ll4r | chrisl, just tried it without those two, but still the same :( | 08:32.19 |
| chrisl | mrd0ll4r: Well, my other thought was to try it with -dNOTRANSPARENCY and also try it with png16m - neither is a solution, but could narrow down where the problem lies. Also, I suspect you'd guess from the last time: there's no way we're going to debug a file that large and complex, so you'll have to find a smaller example to show the problem before one of us can debug it | 08:38.53 |
| mrd0ll4r | yee | 08:43.17 |
| | I'll try those, thx! | 08:43.20 |
| | alright so, preliminary results: still shows with png16m, doesn't show -dNOTRANSPARENCY, but I truly don't see a whole lot in that one because, well, everything is full of non-transparent curves. I'll whip up a minimal example later today :) | 10:14.09 |
| kens | Then nits likely an artefact of transparency in some sense. Whether its correct or not is always hard to tell with transparency. You should probably make a bug report once you have a small example | 10:14.58 |
| chrisl | mrd0ll4r: Unless I'm misconstruing the problem, it looks to me like Acrobat has the same artifacts that gs does: https://ghostscript.com/~chrisl/output-1550927627105.png | 10:21.32 |
| mrd0ll4r | yup, those are the ones I meant | 10:24.16 |
| kens | If our rendering is the same as Acrobat, then its unlikely we'll regard this as a bug | 10:25.19 |
| mrd0ll4r | as i said, I'll make a small example later, but for now: totally poking in the darkness here - is it possible that the curves, which kind of define a shape, declare that area of intersection as outside of the shape, thus not drawing the stroke? | 10:25.56 |
| | hmm, I see | 10:25.59 |
| chrisl | Well, Acrobat is the de fact standard to which everyone works (unfortunately) | 10:26.46 |
| kens | If you can make a simple example we can ask our transparency expert whether he thinks its correct. | 10:30.52 |
| mrd0ll4r | i will, thanks! :) | 10:32.27 |
| | huh, turns out that's not actually as easy as I thought. I'll try again tomorrow | 20:12.46 |
| | Forward 1 day (to 2019/03/12)>>> | |
Log of #ghostscript at irc.freenode.net.
