lots of changes done.. * one diff against original vdrrip-0.3.0 (previous patches merged) * modified roundValue (does not do simple flooring anymore, but nearest) - this is done to get Scale Width's/Height's which are closer to CalcAspect * in r7 -vop was fixed with -vf which traverses the filterlist left to right, so crop needs to precede scale -> fixed * cropdetection works now, previous version did use sort -r at the end of CROPCMD which did not suffice (the ordering would not work, since uniq -c inserts non-zeropadded counts) -> fixed by using sort -nr * cropdetection and preview is done starting at the middle of 001.vdr for sure -> fixed using -sb option of mplayer, -ss fails on most dvb recordings * CropWidth is taken into account for CalcAspect calculation * restore saved Settings actually brings back values you left the movie with (bitrate was discarded, since it was pulled in from TemplateValues, each time settings were loaded from save.vdrrip * CropPosX, CropPosY are not set by vdrrip anymore -> fix, it relies fully on mplayer suggested values (background: vdrrip puts the crop mask at the real center of the original video, but the cropped image is not alway at the center - mplayer's detection algorithm does better, it really detects where the image is, then rounds cropwidth/height and afterwards buts the down-rounded cropped image to the center of that image, not of the original video size!!) * the logic to round values to multiples of 16 is not necessary after calling the CROPCMD, since the default of this mplayer funtion already is to round to 16's -> removed * made CropPosX, CropPosY adjustable in the Expert Section, to manually adjust or have a look at those values (complete with idiot's checking ;-) -> osd translations wanted * made CropWidth, CropHeight steppable by 8, not by 16 -> as cropping in vdrrip is only used in combination with scaling, it's stupid, since you loose aspect accuracy twice (rounds to 16's at crop time and at scale time again) -> makes it possible for the automatic scaler to find a really good width/height that is close to the original aspect (since more aspect's are possible due to more cropvalues possibilities) * fixed bug with interlacing (leading colon handled properly now if useropts are not set), also interlacing at codec level is only done, if you choose so with deinterlacing: codec_does_il -> previous behavior was strange, since the codec was always forced to do interlacing, you didn't have the chance to not use it, except if you turned on a deinterlace postprocessing filter (which is at least unneccessary if not unwise if your material is not interlaced) * added l5 deinterlacing option.. * init script fixed - restart was problematic since the daemon is a shell script and has sleeps in it, start-stop-daemon is done before the script finishes (or so), adding a sleep for the same time in the init script seems to help (background: bash waits until sleep is finished before interrupts are handled) * maybe some things i forgot.. * please test it (~x86)
additional items were fixed, therefore no files yet: * better performance with recordings that include more than one .vdr file (piping is used) * init script fix (now really runs with vdr:vdr! - did with root:root previously since --user was used instead of --chuid) * vdr recordings are sorted in menu * [container=avi?] -> vdrrip takes vdr cut marks into account, so you don't have to cut prior to encoding (first two marks are taken as start and end only) -> this is shown in osd if applicable * [container=ogm,etc] -> looks like vdrsync.pl did cut material already (also complex, more than 2 marks) on demuxing Regards, cmuelle8
Created attachment 131988 [details] vdr-vdrrip-0.3.0-r8.ebuild
Comment on attachment 131988 [details] vdr-vdrrip-0.3.0-r8.ebuild ebuild, depends on one patch file and the original sources
Created attachment 131989 [details] files/vdrrip-qh.conf
Comment on attachment 131989 [details] files/vdrrip-qh.conf minor change to run also under dropped group privileges
Created attachment 131991 [details] files/vdrrip-qh use --chuid instead of --user
Created attachment 131992 [details, diff] files/vdrrip-0.3.0-20070925-vdr1.3.7-mplayer-merged.diff needs at least vdr-1.3.7, enhancements see above..
are bziped diff files allowed in the portage tree? .. might be a possibility to shrink the patch size? The changes are released under the GPL.. Regards.. cmuelle8
Uiui, thanks for the lot of work you have done. I will take a look on this in the next few days.
To get a better understanding of what was changed since r7 you might want to ebuild-unpack r7 and r8 (or simply start an emerge/paludis run and break after unpack phase), then diff those directories. Else it is harder to see what is new in r8 since I merged all patches and diffed against original 0.3.0 release.. Also note that r8 depends on >=vdr-1.3.7 .. r7 could still be used with older versions (though I did not test this, in r7 all patches in files/ were only applied if you built against >=vdr-1.3.7 anyway, so except for the gentoo specific adjustments bare 0.3.0 was used vor <1.3.7) There is still lots of room for improvement.. a lot of things could be handled more elegantly if you weave vdrrip code closer to the VDRAPI (e.g. getting the recordings is fully stand-alone directory based).. but then I wonder if that's really a good thing, after all the more independet you are from VDRAPI the more independent (though ugly) the code is from changes in VDRAPI.. Also the queuehandler should be implemented in C for better signaling response - shell scripts don't do too well on that.. (but hey, it works ;-) Regards, cmuelle8
Created attachment 132042 [details, diff] files/vdrrip-0.3.0-20070925-vdr1.3.7-mplayer-merged.diff - updated updated Length computation, when marks are set (was wrong, Frames divide by Fps missing..)
there are still some issues I'm working on right now..
Created attachment 132147 [details, diff] files/vdrrip-0.3.0-20070925-vdr1.3.7-mplayer-merged.diff - updated2 happy testing.. * rewrote setMarksVDR code, piping should work now.. * code cleanups, fixes piping is still problematic, i'll probably revert to the old code of cating the files together (although this is ugly and space demanding): * seems like the generated avi files are missing an index when you pipe the stuff to mplayer - can anyone confirm that?
it dosnt work on my system, i have nothing changed in configs after a fresh install if i try to add a movie to convert, vdr-1.5.13 crashs without any warnings or logs ... btw. review your .diff for what is this needed? --> - VCodecs[c] = tmp; + VCodecs[c] = tmp; c1++; - NumVCodecs++; + NumVCodecs++; and the patch is full of lines like that, if you will you make the source more Beautiful add a seperated patch for this ;) this blow up only the size of the patch..
Yes, they are actually just changes in whitespace (the original had tabs at random, if I recall right) Please don't consider putting this in portage yet, it's just not ready, please stay at r7. I still find the handling of >1 source files ugly (and the need to concatenate the files before) - I'm experimenting with a kernel module, since just piping doesn't work well (as in this patch), but a) it's not ready for production yet b) it's not elegant Do you have ideas how to gracefully handle the issue of representing more than one file as one seekable file (this is what mencoder needs to do it's job/write an indexed file)? The solution now is to copy together all vdr source files (so you need double the space and you're wasting time). After all I think the solution of vdrrip, to wrap around command line tools, is not so good after all.. I do not know if vdr and mencoder allow to get and put frame data by library calls, but this might be a much nicer solution.. Please also note that I'm using vdr 1.4.x .. I have not tried this at all with 1.5.x .. If I get some time, I may look at it, but at the moment there are exams to be learned for.. regards, cmuelle8
vdr-vdrrip is removed from the tree, depends on a lot of more issues: not converted to gettext dead on upstream .... ...
