Mencoder and DVD compliance
can you try this command line with mencoder and report if you have compliance problem with your standalone player ... thx
mencoder.exe HDTV.avs -o MPEG2-HQ1.m2v -ovc lavc -lavcopts vcodec=mpeg2video:vpass=1:vqmin=1:lmin=0.01:vqblur =0:vqscale=4 :vrc_maxrate=9000:vrc_buf_size=1835:keyint=15:vqco mp=0.75 :vmax_b_frames=2:vb_qfactor=1.25:vb_qoffset=0.0:vi _qfactor=1.0 :vi_qoffset=0.0:me_range=128:preme=2:dia=6:predia= 6:mbd=2 :mbcmp=262:precmp=262:cmp=256:subcmp=262:mv0:trell :cbp :dc=8:qprd:psnr:aspect=16/9:inter_matrix=20,21,22,23,24,25,26,27,21, 21,22,23,24,25,26,28,22,22,23,24,25,26,28,30,23,23 ,24,25,26,28,29,32, 24,24,25,26,28,31,33,35,25,25,26,28,31,34,36,39,26 ,26,28,29,33,36,41, 43,27,28,30,32,35,39,43,48:intra_matrix=8,16,16,16 ,17,17,18,19,16,16, 16,16,17,18,19,20,16,16,16,17,18,19,20,22,16,16,17 ,18,19,21,23,26,17, 17,18,19,21,24,27,31,17,18,19,21,24,28,33,40,18,19 ,20,23,27,33,42,51, 19,20,22,26,31,40,51,64 -of rawvideo -ffourcc MPG2 mencoder.exe HDTV.avs -o MPEG2-HQ2.m2v -ovc lavc -lavcopts vcodec=mpeg2video:vpass=3:vqmin=1:lmin=0.01:vqblur =0 :vbitrate=3000:vrc_maxrate=9000:vrc_buf_size=1835: keyint=15 :vqcomp=0.75:vmax_b_frames=2:vb_qfactor=1.25:vb_qo ffset=0.0 :vi_qfactor=1.0:vi_qoffset=0.0:me_range=128:preme= 2:dia=6 :predia=6:mbd=2:mbcmp=262:precmp=262:cmp=256:subcm p=262 :mv0:trell:cbp:dc=8:psnr:aspect=16/9:inter_matrix=20,21,22,23,24,25, 26,27,21,21,22,23,24,25,26,28,22,22,23,24,25,26,28 ,30,23,23,24,25,26, 28,29,32,24,24,25,26,28,31,33,35,25,25,26,28,31,34 ,36,39,26,26,28,29, 33,36,41,43,27,28,30,32,35,39,43,48:intra_matrix=8 ,16,16,16,17,17,18, 19,16,16,16,16,17,18,19,20,16,16,16,17,18,19,20,22 ,16,16,17,18,19,21, 23,26,17,17,18,19,21,24,27,31,17,18,19,21,24,28,33 ,40,18,19,20,23,27, 33,42,51,19,20,22,26,31,40,51,64 -of rawvideo -ffourcc MPG2 mencoder.exe HDTV.avs -o MPEG2-HQ3.m2v -ovc lavc -lavcopts vcodec=mpeg2video:vpass=3:vqmin=1:lmin=0.01:vqblur =0 :vbitrate=3000:vrc_maxrate=9000:vrc_buf_size=1835: keyint=15 :vqcomp=0.75:vmax_b_frames=2:vb_qfactor=1.25:vb_qo ffset=0.0 :vi_qfactor=1.0:vi_qoffset=0.0:me_range=128:preme= 2:dia=6 :predia=6:mbd=2:mbcmp=262:precmp=262:cmp=256:subcm p=262 :mv0:trell:cbp:dc=8:psnr:aspect=16/9:inter_matrix=20,21,22,23,24, 25,26,27,21,21,22,23,24,25,26,28,22,22,23,24,25,26 ,28,30,23,23,24,25, 26,28,29,32,24,24,25,26,28,31,33,35,25,25,26,28,31 ,34,36,39,26,26,28, 29,33,36,41,43,27,28,30,32,35,39,43,48:intra_matri x=8,16,16,16,17,17, 18,19,16,16,16,16,17,18,19,20,16,16,16,17,18,19,20 ,22,16,16,17,18,19, 21,23,26,17,17,18,19,21,24,27,31,17,18,19,21,24,28 ,33,40,18,19,20,23, 27,33,42,51,19,20,22,26,31,40,51,64 -of rawvideo -ffourcc MPG2 |
Re: Mencoder and DVD compliance
Quote:
Let me laught... My two players don't have the slighest choke with things as far from the standard than 544*576 resolution and a GOP of 25. So what can you deduce from that ? The encoder that output in 544*576 / 25 GOP is DVD compliant ? I also mixed PAL / NTSC DVD with no problems. And even PAL resolution with NTSC framerate. I guess my player is more compliant than the standard itself :cid: |
I do think the post above is especially adressed to people which had problems using the CLI lines we used 1 year ago.
We treat our encoders like Girlfreinds So every encoder gets a second chance ;) As our Girlfriends too :flowers: @Sagittaire I think the second line was supposed to be a 2nd pass, right? As it still shows the 3d pass parameter. |
Re: Mencoder and DVD compliance
Quote:
It's still basicallly the same code base as it was one year ago (I've looked at the code). Still with the same rate control bugs. So no matter what command line arguments are used, the final result will still be non DVD compliant data. Or, did the Mencoder group add or clean up code which I'm not aware of :?: I hope I'm wrong :) Edit: Also, you're using dc=8 on a high bitrate encode, which should really be 10 bits for those settings ;) -kwag |
Quote:
Can 10 testimonies make a whole truth ? And as I told, for sure my pioneer won't have any problem with none of these three lines. It didn't have any with the mencoder of last year :!:. So okay, that means that the old mencoder is "as good as" the new one can be. I let you put what you want in this sentence ;-) |
Quote:
In fact I read completely your mencoder sub-forum and all your CLI for mencoder are always not MPEG2 MP@ML DVD compliant ... :?: Mencoder and 1 pass constant quantizer (vqscale mode) It's impossible with mencoder to make CQ mode like for TMPGEnc because RC and VBV specification are desactived in this mode and can produce MPEG2 stream with buffer underflowin Mencoder and maximum vector motion range You must use me_range=128 (unknow with search on your forum ... ???) for make compliant MPEG2 stream Mencoder and Custum Matrix You use for example Notch Matrix for your test. Notch Matrix is compliant with MPEG2 but not with libavcodec: you must use value superior at 16 for all coef (high energy intra coef is an exception) HC (HCEnc developper) analyse my MPEG2 stream produce with my CLI and say: it's compliant with vbv and mpeg2 DVD specifictions ... ??? |
Quote:
Well at least in that time we where (imho) the only Community where Mencoder testings have been done on mpeg2 outputs in such an extra-try and try and trying way ;) |
Quote:
Quote:
IMO high bitrate encoding (more than 4 or 5 Mbps) can be a problem with mencoder (multipass encoding are necessary for vbv compliance) but certainely not for low/medium bitrate ... At this time Mencoder is the best MPEG2 encoder and by far for low/medium bitrate: - Rate control tweak like quant variability, ratio/offset for I,Bframe ... - Rate Distortion Optimisation - Adaptative Quantisation with multiple masking |
Sorry but now this is enough !
Quote:
Quote:
(I'm sure that if you read the mencoder sub-forum you read that I already wrote that a year ago. Nothing has changed). |
Well DivX6 don't use CM like XviD but DivX6 is an MPEG4 ASP compliant codec ... DivX6 don't use 3WP GMC but only 1WP GMC but DivX6 is an MPEG4 ASP compliant codec ... NDAVC Main Profil don't use "reference bframe" but NDAVC MP is an H264 compliant codec, NDAVC in Nero Recode can not use more than 8 Reference frame but NDAVC in nero recode is H264 compliant codec ... etc etc etc
You must not use coef < 16 (libavcodec source and my experience with mencoder) because low coef for intra/inter can make buffer underflow and corrupt vbv specification. It's just only a little specifical limitation for intra/inter coef and you can use all the other matrix for respect this particular limitation (like HVS matrix serie for example) Quote:
Quote:
|
just a word about vector limiting : this parameter is probably new (I looked for something like that one year ago and I don't think I missed it), and this surely a fix for that problem.
But others are still there. The respect of the bitrage is one of them. And if you can't understand that NOTHING in the code can prevent some under or over flow, whatever the option you use I just ask you to stop speaking about what you can't understand. We tried to explain to you more than once in your previous post that this is not because you never faced the problem that it does not exist. And I'm sorry to tell you that if for you, limiting the coeff into the matrix to "16" to have a respect in the bitrate is just "little limitation", then you still don't understand what is and is not "compliant". In a last word I want to focus on your own word : "You must not use coef < 16 (libavcodec source and my experience with mencoder) because low coef for intra/inter can make buffer underflow and corrupt vbv" Your experience... in other word you admit that the tool doen't give any insurrance that there won't be any underflow. You just experienced (and this experience is precious, I do not tell the opposite) that under given conditions you did not have the problem. Finally, I can't do as others suggested to me : "KVCD Matrix is not usable with mencoder ? Okay. So mencoder has no interrest for us", ans just close the thread. But I find that too easy. |
Excuse me I post a second message about your fist lines because I did not understood them before, so I simply drop them in my first post.
Quote:
Do you see the difference between "not implementing all the features of a codec spec" and "not implementing control in the code that will insure that we will stay within the spec of the codec" ? I fear you not... An instance taht you didn't give but it is closer to the problem with mencoder than Divx6 : CCE does not accept values above 99 for matrix coef. And this does not make CCE not compliant. Why ? Because the limitation is not in the code that makes the engine of the encore, it is in the GUI that does not allow you to put value upper 99. But in an ECL file, you can use the value you want : there won't be any flow problem ! They just decide to limit the possibility offered to the user to shorter range within the codec limit. Just like Divx6 designers decided to limit the usage of GMC to 1way (note: are you sure of that ? I think I see test with 3way GMC on the Divxtest v2 disc - but I may be wrong). To come back to mencoder : this is not something the designers decided to that limit you to 16. This is something that users, like you, experienced as a safe limit. That makes all the difference |
Quote:
KVCD's Notch matrix was integrated a LONG time ago into mencoder. Read the manual :arrow: http://64.233.161.104/search?q=cache...n&client=opera ( Link is cache, because the damn site is down ) Search for: "The KVCD "Notch" Quantization Matrix:" there. Please get your facts straight before adding confusion to the unaquainted people. -kwag |
Well...
what was the problem related to mpeg2 MP@ML with our encodings? 1- We dropped Mencoder due the fact that Peaks on encodings couldnt be handled 2- We dropped Mencoder cause many members got Bitrateallocation problems 3- We dropped Mencoder cause some members did report that libavcodec encodings wherent able to be played back on their SAP. Regarding 1- What about peaks?? Still there? I dont want to do risky encodes anymore but being able to tweak the Bitrates as Im used to like in CCE or TmpgEnc. Means IF I want to end up in a higher Bitrate due an 1movie2oneDVD-5 was my choice, then I want that bitrate to be kept in avg and max. also min bitrate settings would make it more professional. BUT thats even NOT fully supported at CCE as it also likes sometimes to drop even down to 100kb. SO WAHT ARE THE REAL FACTS regarding this IN THE STATE OF MENCODER/LIBAVCODEC NOW? Regarding 2- Based on my practical experiences and subjective eye-checking. I was able to do a CQ like encoding in mencoder by tweaking the Lmin parameter. As in that case on DVB-Reencodings Mencoder was my first choice. So (if) u can explain me what are the risks IF they are present by doing this. I see you do use "trellis" in a combination with Lmin lower than 2 .. PeterCheat did proof on a simple example that garbage on some given frames "will" be the result. Regarding 3- I already mentioned it in the Thread before you started in here about mencoder MP@ML compilance: For personal backups I do stick on what gives me the best subjective/objective result and if MY SAP can play it back without problems. SAME to HCencoder which is *now* the new free-encoder STAR in the sky --- but some Users also do have problems when playing back its result on "their" SAP. So ... do look for a person who reported problems when playing back mencoder generated streams on his SAP and do offer "HIM" the Lines above, so HE can test it and report if something has changed. That "proof/fact" can't base on only YOUR-ONLY experience using only YOUR SAP. If compilancy within the encoding community is proofed like in the case of CCE and TmpgEnc then we can make suggestions on using mencoder as its based on practical proofing/facts. And about Notch .... On the one hand it is even offered at the Mencoder Dokumentation, on the other hand you dont suggest it due to your explanations. - where is the logic? IF all MP@ML compilant encoders dont have problems when using Notch, why does Mencoder suffer from that theory? (not a question but something to think about). Resumee: You can contribute testing results, but you can't insist on 100% compilancy when only using YOUR Testing Environment for doing proofs. |
Quote:
Warning, QMAT_SHIFT is larger then 21, overflows possible Code:
========================================================================== Quote:
First Pass: constant quant encoding with no RC limitation Very bad target bitrate : 6184.768 Kbps Overall PSNR = 47.22 dB vbv are not compliant with MP@ML Code:
========================================================================== Second Pass: RC limitations are actived with first pass stat file Very good target bitrate : 5014.041 Kbps Overall PSNR = 46.47 dB -> vbv are not compliant with MP@ML Code:
========================================================================== Third Pass: RC limitations are actived with 2 pass stat file update Very good target bitrate : 5012.150 Kbps Overall PSNR = 46.33 dB vbv is compliant with MP@ML Code:
========================================================================== |
Quote:
Quote:
Quote:
|
Quote:
I change 4CC in all stream for simulate DivX stream encoding with 3WP GMC ... :idea: |
Quote:
Do you mean that the stream that are supposed to test SAP compliance with Divx GMC are in fact Xvids ? I don't see the benefit of theses tracks then ? |
Quote:
In fact GMC sample test is split in 3 part - first part is MPEG4 ASP without GMC (400 frames) 4CC DX50 - second part is MPEG4 ASP with GMC 1WP (400 frames) 4CC DX50 - third part is MPEG4 ASP with GMC 3WP (400 frames) 4CC DX50 Finally I test 1WP and 3WP SAP compatibility in only one files test. I make that too for bframe (Packed Bitstream or not) and bitrate crash test ... If make too all the DivX6 format compatibilty test (here a menu demonstration) or here MPEG2 MP@HL 1280*720*60 compatibilité test |
Quote:
If changing the FOURCC is enought, I don't think there were so much posts on the "surdvd" forum about that (even if I never tested to just change the FOURCC and see what happen). But MPEG4 is not my cup of tea, so I don't know... Quote:
|
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.