Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
19th October 2016, 10:35 | #2981 | Link | |
AVStoDVD Dev
Join Date: Apr 2006
Location: Italy
Posts: 1,302
|
Quote:
Bye |
|
19th October 2016, 14:56 | #2982 | Link | ||
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Quote:
http://forum.doom9.org/showthread.ph...60#post1705060 For muxing with Mplex, the -f 8 option for MPEG2 muxing seems to be the universal method which is used by just everyone. It does matter if for example you take the muxed MPEG2 file and import it into another software like DVDStyler which uses DVDAuthor to create the DVD. DVDAuthor will fail if the source MPEG2 does not have DVD NavPacks. And for the Mplex version, this is a long story, I have to go back a long way... My first encounter with Mplex was with good old DVD2SVCD, and the (modified) Mplex version which came with it had a strong tendency to cause stuttering. (Which eventually led to my collaboration with Sir Didymus to create BatchMux for DVD2SVCD). But later versions of Mplex got rid of the stuttering, and the version I am using comes from DVDForger 0.61 Alpha by Alvarez (the man who created WAVI). You can still find it at Videohelp. The next version from DVDForger 0.62 Alpha did not work any more on my machine. Looks like you are shipping this version with the current Alpha, it shows the same version info, but it is a little bigger. For the IVTC feature: Quote:
//EDIT// I believe that pulldown via DGPulldown can indeed only be used for progressive sources, but PAL speedup should work no matter if the source is progressive or interlaced. Cheers manolito Last edited by manolito; 20th October 2016 at 06:24. |
||
20th October 2016, 06:30 | #2983 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
FFMS2 C-Plugin
Contrary to what qyot27 posted a little while ago he did just publish a new version of his ffms2 C-Plugin:
Get it here: http://forum.doom9.org/showthread.ph...12#post1783312 Tested it extensively under AVStoDVD, works perfectly even with my old WinXP Non-SSE2 computer. Just replace the file "ffms2.dll" (use the x32 version) in the "AVStoDVD\Lib" folder with the new version. Cheers manolito |
21st October 2016, 15:43 | #2984 | Link |
AVStoDVD Dev
Join Date: Apr 2006
Location: Italy
Posts: 1,302
|
@manolito
bitrate caps for NTSC and PAL: my bad, it was documented in this thread... thanks anyway... Mplex for non SSE2: I would like to include the non SSE2 executable, but switch to it only if needed. What is the error message you get on your P3 system when you use the newer mplex.exe? NTSC to PAL conversion when NTSC is telecined: you are right, the pulldown/speedup approach would be much more appropriate. Will implement it. New FFMS2 C plugin: great! thanks to you and qyot27 Bye |
21st October 2016, 17:27 | #2985 | Link | ||
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Quote:
there is no error message when using the newer version of Mplex.exe (size 165 KB), it just does not create a muxed file. The older version (size 151 KB) has no problems. Both versions display exactly the same version info: Quote:
It should be easy to test for you if I am right (I obviously cannot test it on my non-SSE2 Coppermine CPU): Take a longer source file and mux it using both Mplex versions and note the time for the multiplexing. If the muxing times are the same or close then using the newer version would be useless. Cheers manolito |
||
22nd October 2016, 09:24 | #2986 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
I bit the bullet and did my own muxing tests using 3 different versions of Mplex. And I was quite surprised by the results...
Test bed: ThinkPad T410, Core i5 CPU, Win7 64-bit Latest AVStoDVD 2.85 Alpha Mplex versions: 1. Version 1.9.0 (2.2.7), size 151 KB, the version which works without SSE2 2. Version 1.9.0 (2.2.7), size 165 KB, the version which comes with the latest A2D Alpha and does not work without SSE2 3. Version 2.1.0 (2.2.7), the version which comes with the latest 32-bit version of DVDStyler. It also needs SSE2, and it is a shared build which needs 2 external libraries. Download it here: http://www99.zippyshare.com/v/I4vwyd2r/file.html Source file: A full 3 and a half hour DVD with 2 audio tracks (Woodstock Director's cut) I rebooted the computer before every muxing test to exclude any possible caching effects. The muxing times were as follows: 1. 3 min 45 sec 2. 4 min 08 sec 3. 4 min 45 sec I repeated the tests several times because I had a hard time believing this, but the muxing times were constant in every test. I also compared the resulting MPEG2 files, they had exactly the same size for every version of Mplex.exe. I believe (without testing this) that the results were bit identical. Conclusion: The oldest Mplex version (size 151 KB) was by far the fastest one. Absolutely no need to upgrade to a newer Mplex version to achieve better speed. Enabling SSE2 in the compiler settings has nothing but adverse effects in this case. This is a clear example that the attitude of some people in this forum (to look down on others who still use older hardware and older Operating Systems) is not justified. In many cases if a newer software version does not work any longer under WinXP or on a CPU without SSE2 support the reasons are not technical at all. Mostly it is just the laziness of the developer who cannot bring himself to toggle a couple of compiler switches. And of course there's also a couple of developers who seem to be on a mission to eradicate older OSs or hardware. Evangelists... Alright, enough ranting for today... Cheers manolito Last edited by manolito; 22nd October 2016 at 09:28. |
31st October 2016, 13:49 | #2987 | Link |
AVStoDVD Dev
Join Date: Apr 2006
Location: Italy
Posts: 1,302
|
Thanks manolito for the extensive testing and reporting. That's very useful.
I have prepared a new 2.8.5 Alpha, receiving some of the inputs. Cumulative change log from 2.8.4: - Some bugs fixed - Added 'IVTC' option (Inverse Video Telecine routine for hard-telecined NTSC source) to 'Edit Title'/'AviSynth' section - Added TIVTC plugins package to implement IVTC (TFM.TDecimate) - Added tooltips to 'Edit Title'/'AviSynth' options - Added video avg bitrate cap (7100 for PAL, 6300 for NTSC) when VCD resolution and FFmpeg are selected to avoid padding - Substituted FFmpeg with Mplex for MPEG2 muxing - Refactored wave PCM audio stream retaining - FFMpegSource plugin updated to release 2.22 (r1103 C plugin) Thanks Bye |
31st October 2016, 21:25 | #2988 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Thanks for the new build, just gave it a short test run, looks good...
The bitrate cap for FFmpeg VCD frame sizes works well (even though it is not really necessary for the FFmpeg version which comes with A2D). The latest ffms2 r1140 plugin did not make it into this release yet. And pasting the updated German language part into the existing file caused a missing line break, here is the complete file: http://www98.zippyshare.com/v/ja9RWenF/file.html Thanks again and Cheers manolito Last edited by manolito; 24th December 2016 at 01:14. |
2nd November 2016, 09:59 | #2989 | Link | |
Registered User
Join Date: Jan 2011
Posts: 1
|
MrC, thank you for the new alpha-version of your software.
Quote:
I have tested MeGUI with a couple of hard-telecined DVDs, and the detection worked fine. The script should also be able to detect frame-repetition, but I haven’t tested that. JuMe |
|
3rd November 2016, 15:52 | #2990 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Hi MrC,
just found an issue which only happens in FFmpeg Safe Mode when using a source file with 4:2:2 chroma subsampling. Please have a look here: https://sourceforge.net/p/dvdstyler/...19525d5b/#9da1 FFmpeg will create an m2v file with 4:2:2 subsampling which is not DVD compliant. To avoid this the command line for FFmpeg Safe Mode must contain the parameter "-pix_fmt yuv420p". In normal modes (using AviSynth) there is no problem because the source is always converted to YV12. Cheers manolito |
13th November 2016, 17:04 | #2992 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Yes, it should be possible - if you have a supported CPU...
HCenc does NOT support GPU accelerated encoding, but current FFmpeg versions do support Intel QSV (Quick Sync Video) encoding. If you have a newer Intel CPU and a current version of FFmpeg then you can modify the FFmpeg command line to enable the QSV MPEG2 encoder. (You have to specify "-vcodec mpeg2_qsv") Here is a White Paper from Intel about QSV and FFmpeg: http://www.intel.de/content/dam/www/...tall-valid.pdf Keep in mind that the encoding quality will generally be worse when using QSV compared to software encoding. Cheers manolito |
16th November 2016, 22:41 | #2993 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
A heads up for folks who like to use the newest versions of FFmpeg with AVStoDVD:
Starting around end of July 2016, shortly before version 3.1.2 was published, the FFmpeg devs changed the way FFmpeg interacts with AviSynth (again). In AVStoDVD this results in a crash as soon as FFmpeg starts its thing. Luckily the workaround is pretty easy: All you need to do is add the path to the "AVStoDVD\Lib" folder to your path variable. The file "A2DSource.avsi" in this folder loads a couple of AviSynth plugins, but with current FFmpeg versions this plugin loading fails. Unless you add this Lib folder to your path. Of course there is also the question how much sense it makes under AVStoDVD to always use the latest FFmpeg versions. I have no idea if anybody still works on the MPEG2 encoder, for the last year or so I could not see any differences in encoding quality using different FFmpeg versions. Cheers manolito |
18th November 2016, 18:48 | #2994 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
New version of my AVStoDVD plugins, had to modify the FFmpeg_VBR plugin for the latest AVStoDVD_285_Alpha161027.
Download link in this post: http://forum.doom9.org/showthread.ph...93#post1773293 Cheers manolito |
20th November 2016, 18:23 | #2995 | Link |
AVStoDVD Dev
Join Date: Apr 2006
Location: Italy
Posts: 1,302
|
@Mister XY
in addition to manolito's hints, I would add that LAV Filters lets you to decode using your GPU capability. Usually it does not lead to improve encoding time, but you will have your CPU free for other tasks, and your CPU fan much more silent... @manolito thanks for highlighting the FFmpeg 4:2:2 chroma subsampling issue. I will add the "-pix_fmt yuv420p" flag to FFmpeg safe mode in the next release. Thanks also for the suggestion how to use the more recent FFmpeg releases with A2D. And thanks also for the updated FFmpeg VBR plugin! 2.8.5 Final should be released before Xmas. Bye |
10th December 2016, 15:08 | #2996 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
This is for the folks who would like to use motion protected FPS conversion, but feel that MotionProtectedFPS causes too many artifacts:
Forum member johnmeyer just published a script which is based on MVTools V2, and IMO this script produces results which are vastly superior to MotionProtectedFPS. If you want to check it out, download it here: http://www13.zippyshare.com/v/SprGa0oX/file.html After copying the files into your "AviSynth\plugins" folder, all you have to do is edit your AviSynth script within AVStoDVD and replace your FPS conversion command with "jm_fps(fps)". Example: The script generated by AVStoDVD has the command "ChangeFPS(25)" in it. Uncheck the "automatic" option and replace "ChangeFPS(25)" with "jm_fps(25)". There is one catch: This script requests frames in a non-linear fashion. DirectShowSource is not too good at this, so you should force FFmpegSource for video if you want to use this script. (or use DSS2Mod - ask me how to do this if you are interested) //Edit// Actually there is another catch: It is sloooow... Speed is about 25% of MotionProtectedFPS. Cheers manolito Last edited by manolito; 12th December 2016 at 10:22. |
15th December 2016, 21:30 | #2997 | Link | |
Registered User
Join Date: Jun 2014
Posts: 82
|
Quote:
|
|
24th December 2016, 01:28 | #2999 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Thanks MrC for this nice Xmas present...
Everything works perfectly here, maybe users should update ffms2 with the latest version from here: http://forum.doom9.org/showthread.ph...12#post1783312 And the German language file still has a missing line break, the correct file is here: http://www98.zippyshare.com/v/ja9RWenF/file.html I also updated my plugins for the new version, I removed the chroma subsampling workaround for ffmpeg safe mode. Get it here: http://www103.zippyshare.com/v/zEBA4BcR/file.html Thanks again, Merry Xmas and a Happy New Year... manolito |
24th December 2016, 10:45 | #3000 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Well, I did find a bug with the new audio speedup command for FFmpeg safe mode:
1. For a 24fps -> 25fps speedup the command -filter:a "atempo=1,0427" is added. This crashes FFmpeg, you need to use a decimal point instead of a comma. (My system locale is set to standard German) Also, shouldn't the factor be 1.0417? 2. The speedup command must not be used if FFmpeg Safe Mode is also used for video. In this case (without AviSynth) the video does not get sped up, DGPulldown is used instead. So the audio should not have a speedup command, too. Cheers manolito Last edited by manolito; 24th December 2016 at 11:44. |
Thread Tools | Search this Thread |
Display Modes | |
|
|