DGAVCDec 1.0.9 - Page 31

Guest · 20th November 2007, 14:41

Quote:

Originally Posted by drmpeg

It's a Transport Stream. Didn't the 0x47 at the beginning tip you off?

Yes, it did, but the one at location 0x50 made me think it was spurious. I should have looked at more sync bytes to find the 188-byte sync. But I was busy with the info dialog and only looked at it briefly.

After demuxing, it plays fine with CoreAVC but it crashes VLC, so libavcodec doesn't like it. Therefore, it doesn't work in DGAVCDec either. Most likely the PAFF encoding is the problem.

tjmitchem · 20th November 2007, 14:57

Quote:

Originally Posted by neuron2

Yes, it did, but the one at location 0x50 made me think it was spurious. I should have looked at more sync bytes to find the 188-byte sync. But I was busy with the info dialog and only looked at it briefly.

After demuxing, it plays fine with CoreAVC but it crashes VLC, so libavcodec doesn't like it. Therefore, it doesn't work in DGAVCDec either. Most likely the PAFF encoding is the problem.

This is weird. If that was a transport stream, then it means that the demuxer I used is not really demuxing the stream.

I just redid the file with a different demuxer. Can someone take a look at it now and tell me if it's actually a raw H.264 stream? I cut the file down to 5MB

Thanks,
Terry

Guest · 20th November 2007, 15:15

Quote:

Originally Posted by tjmitchem

I just redid the file with a different demuxer. Can someone take a look at it now and tell me if it's actually a raw H.264 stream?

Yes, that is a raw stream.

tjmitchem · 20th November 2007, 15:20

Quote:

Originally Posted by neuron2

Yes, that is a raw stream.

Thanks.

D-Train · 20th November 2007, 16:33

Quote:

Originally Posted by tjmitchem

I installed 1.0.0 Alpha 8 and tried to index a file. The stream came from DishNetwork and contains PAFF. I know it's not supported, but decided to try it just for the heck of it.

I also have a bunch of Dish Network HD files I've been wanting to re-encode. You can download a small sample here (hopefully downloading is not too slow).

I know these are PAFF, and I'm pretty sure they have no IDR frames. I think this combination causes issues for libavcodec. For the larger files (1 hour), seek times are really bad. Avisynth basically hangs for 20 minutes per seek. Anyways, if someone wants to experiment with this then have at it. The uploaded file indexes fine but only decodes the first frame.

Thanks neuron2, I'm mostly a lurker but have found a lot of use for DGMPGDec, and now DGAVCDec.

Guest · 20th November 2007, 16:52

Thank you for the test stream. Fairly soon, I will be turning my attention to this libavcodec problem with some PAFF streams.

Guest · 22nd November 2007, 00:28

* Lots of GUI bugs and crashing bugs fixed.

* New info window field: max bitrate seen.

* Major code cleanup and purging of DGIndex leftovers. Conversion to exclusive C++ compilation.

http://neuron2.net/dgavcdec/dgavcdec100a10.zip

Please delete your INI file before running this, or replace it with the one in the zip file.

Note that the previous message box popup for forcing the IVTC'ed frame rate is now replaced by Force Film in the Video menu. I have not yet implemented decimation of non-film sections, so for now this will not work on hybrid video.

With the disappearance of a lot of menu options, some explanation is in order. It is really very hard to integrate the container parsers with the stream buffering, NALU parsing, and parallel parsing and keep everything working right. For DGMPGDec, it is a nightmare that I still lose sleep over. I'm not sure I even fully understand how it works and if it is provably correct. I do know of several low-probability cases that can cause explosions. Yes, it's low probability but with enough uses, it's going to happen to someone. The code base is also very hard to modify and enhance.

So I am taking a new tack with DGAVCDec. I want to keep it simple and provably correct. So my plan is that cutting, demuxing, container verification and repair, etc., will be done by separate tools. DGAVCDec will be for serving video only. There are existing tools for these things but I plan to make my own and make them available via the Tools menu. This approach will allow me to keep my sanity as well as to continue moving forward despite my limited time for coding these days. I don't need another massive nightmare on my hands!

Probably, then, the [ and ] arrow keys will simply allow for cutting out portions of your AVC raw stream. Yes, you can use Trim() in your script, but it's more convenient this way and it's useful sometimes to generate a cut AVS raw stream as well.

canTsTop · 22nd November 2007, 01:03

thank you neuron2 for your great tool.
in my video, i noticed in information windows, aspect ration is 12:11 but i think it should be 15:11

Guest · 22nd November 2007, 04:00

It prints what is in the stream. Why do you think it should be something else? Can you post a link to the stream?

LigH · 22nd November 2007, 08:00

It is about pixel aspect ratio (stored pixels vs. straightened-for-display pixels)

12:11 ~ 785:720 (the width of 785 was calculated as 720*12:11)

canTsTop · 22nd November 2007, 11:32

sorry for my mistake, i thought it was display aspect ratio.
576*15:11~785

this is my sample http://rapidshare.com/files/71446036/raw.h264.html

p.s. i would be grateful if some one advise how to deal with this video, demuxed audio is 3min 17s but video is 4min 11s

Guest · 22nd November 2007, 12:07

Quote:

Originally Posted by canTsTop

p.s. i would be grateful if some one advise how to deal with this video, demuxed audio is 3min 17s but video is 4min 11s

I can't help you with just the raw stream. You'll have to supply the un-demuxed source container sample.

canTsTop · 22nd November 2007, 12:43

thank you for answer, this is un-demuxed TS sample http://rapidshare.com/files/71459314/ttt.ts.html

g0blez · 22nd November 2007, 16:05

I have a question concerning the latest alpha 10.
I've demuxed an EVO file. evodemux claimed the source was 60fps. However, looking at the number of calculated frames and taking into consideration the total length of the stream, I calculated that the frame rate was 23.976fps.

When I opened the demuxed mpv (h.264) file in DGAVCDec, the frame rate listed was 29.970030 fps.
How should I treat this stream? Should I force film it?

bob0r · 22nd November 2007, 21:54

Quote:

Originally Posted by canTsTop

thank you for answer, this is un-demuxed TS sample http://rapidshare.com/files/71459314/ttt.ts.html

Heh, interesting, the .ts file contain no errors, but when you demux/remux the h.264 file, it is indeed: 04:10, while the .ts plays as 03:17

I remuxed the video into .mkv and it plays uhm, variable FPS, i never seen this before.

drmpeg · 22nd November 2007, 23:27

Quote:

Originally Posted by bob0r

Heh, interesting, the .ts file contain no errors, but when you demux/remux the h.264 file, it is indeed: 04:10, while the .ts plays as 03:17

I remuxed the video into .mkv and it plays uhm, variable FPS, i never seen this before.

It's a PAFF bitstream that actually switches between field and frame pictures. Here's the output from the reference decoder that shows if a frame is a field pair.

Code:

----------------------------- JM 13.0 (FRExt) -----------------------------
 Decoder config file                    : (null) 
--------------------------------------------------------------------------
 Input H.264 bitstream                  : bits0001.mpv 
 Output decoded YUV                     : test_dec.yuv 
 Output status file                     : log.dec 
 Input reference file                   : test_rec.yuv does not exist 
                                          SNR values are not available
--------------------------------------------------------------------------
POC must = frame# or field# for SNRs to be correct
--------------------------------------------------------------------------
  Frame          POC  Pic#   QP    SnrY     SnrU     SnrV   Y:U:V Time(ms)
--------------------------------------------------------------------------
Warning: Decoding does not start with an IDR picture.
00023( I )       46    96    30                             4:2:0     220
00020( b )       40    97    30                             4:2:0     150
00021( b | b )   43    97    51                             4:2:0     130
00022( b )       44    97    48                             4:2:0     111
00027( P )       54    97    33                             4:2:0     160
00024( b )       48    98    29                             4:2:0     100
00025( b )       50    98    33                             4:2:0     110
00026( b )       52    98    32                             4:2:0     110
00031( P )       62    98    31                             4:2:0     181
00028( b )       56    99    28                             4:2:0     150
00029( b )       58    99    31                             4:2:0     120
00030( b )       60    99    31                             4:2:0     140
00035( P )       70    99    30                             4:2:0     180
00032( b )       64   100    27                             4:2:0     121
00033( b )       66   100    30                             4:2:0     150
00034( b )       68   100    30                             4:2:0     130
00039( P )       78   100    29                             4:2:0     190
00036( b )       72   101    26                             4:2:0     130
00037( b )       74   101    30                             4:2:0     121
00038( b )       76   101    30                             4:2:0     120
00043( P )       86   101    29                             4:2:0     200
00040( b )       80   102    26                             4:2:0     130
00041( b )       82   102    30                             4:2:0     130
00042( b )       84   102    30                             4:2:0     141
00047( P )       94   102    30                             4:2:0     160
00044( b )       88   103    27                             4:2:0     130
00045( b )       90   103    30                             4:2:0     130
00046( b )       92   103    30                             4:2:0     120
00051( P )      102   103    30                             4:2:0     141
00048( b )       96   104    27                             4:2:0     170
00049( b )       98   104    30                             4:2:0     220
00050( b )      100   104    30                             4:2:0     120
00055( P )      110   104    30                             4:2:0     151
00052( b )      104   105    27                             4:2:0     120
00053( b )      106   105    30                             4:2:0     130
00054( b )      108   105    30                             4:2:0     120
00059( P )      118   105    30                             4:2:0     140
00056( b )      112   106    27                             4:2:0     241
00057( b )      114   106    30                             4:2:0     130
00058( b )      116   106    30                             4:2:0     110
00063( P )      126   106    30                             4:2:0     130
00060( b )      120   107    27                             4:2:0     130
00061( b )      122   107    30                             4:2:0     131
00062( b )      124   107    30                             4:2:0     120
00067( P )      134   107    30                             4:2:0     220
00064( b )      128   108    27                             4:2:0     130
00065( b )      130   108    30                             4:2:0     130
00066( b )      132   108    30                             4:2:0     131
00068( P )      136   108    30                             4:2:0     140
00069( P )      138   109    27                             4:2:0     140
00000(IDR| P )    1     0    24                             4:2:0     341
00001( I | P )    3     1    31                             4:2:0     250
00002( P | P )    5     2    37                             4:2:0     190
00003( P | P )    7     3    41                             4:2:0     170
00004( P | P )    9     4    42                             4:2:0     171
00005( P )       10     5    39                             4:2:0     210
00006( P )       12     6    40                             4:2:0     190
00007( P )       14     7    40                             4:2:0     181
00008( P )       16     8    39                             4:2:0     220
00009( P )       18     9    37                             4:2:0     200
00010( P | P )   21    10    38                             4:2:0     381
00014( P )       28    11    32                             4:2:0     260
00011( b )       22    12    35                             4:2:0     210
00012( b )       24    12    33                             4:2:0     221
00013( b )       26    12    33                             4:2:0     230
00018( P )       36    12    32                             4:2:0     280
00015( b )       30    13    29                             4:2:0     261
00016( b )       32    13    32                             4:2:0     240
00017( b )       34    13    31                             4:2:0     240
00019( P )       38    13    31                             4:2:0     281
00020( P )       40    14    28                             4:2:0     240
00024( P )       48    15    28                             4:2:0     271
00021( b )       42    16    29                             4:2:0     210
00022( b )       44    16    31                             4:2:0     220
00023( b )       46    16    30                             4:2:0     211
00028( P )       56    16    29                             4:2:0     280
00025( b )       50    17    26                             4:2:0     220
00026( b )       52    17    29                             4:2:0     221
00027( b )       54    17    28                             4:2:0     230
00029( P )       58    17    28                             4:2:0     280
00030( P )       60    18    25                             4:2:0     261
00031( P )       62    19    27                             4:2:0     240
00032( P | P )   65    20    32                             4:2:0     240
00033( P | P )   67    21    32                             4:2:0     261
00034( P | P )   69    22    33                             4:2:0     400
00035( P | P )   71    23    33                             4:2:0     251
00036( P | P )   73    24    32                             4:2:0     240
00037( P )       74    25    29                             4:2:0     230
00038( P | P )   77    26    33                             4:2:0     591
00039( P | P )   79    27    32                             4:2:0     241
00040( P | P )   81    28    32                             4:2:0     260
00041( P | P )   83    29    34                             4:2:0     260
00042( P | P )   85    30    34                             4:2:0     241
00046( P | P )   93    31    35                             4:2:0     240
00043( b | b )   87    32    36                             4:2:0     230
00044( b | b )   89    32    37                             4:2:0     231
00045( b | b )   91    32    35                             4:2:0     230
00047( P | P )   95    32    30                             4:2:0     250
00051( P | P )  103    33    31                             4:2:0     271
00048( b | b )   97    34    36                             4:2:0     230
00049( b | b )   99    34    34                             4:2:0     250
00050( b | b )  101    34    35                             4:2:0     221
00000(IDR| P )    1     0    30                             4:2:0     320
00001( P | P )    3     1    32                             4:2:0     231
00002( P | P )    5     2    32                             4:2:0     220
00003( P | P )    7     3    32                             4:2:0     230
00007( P | P )   15     4    33                             4:2:0     241
00004( b | b )    9     5    37                             4:2:0     300
00005( b | b )   11     5    35                             4:2:0     270
00006( b | b )   13     5    34                             4:2:0     251
00008( P | P )   17     5    31                             4:2:0     280
00012( P | P )   25     6    25                             4:2:0     260
00009( b | b )   19     7    29                             4:2:0     281
00010( b | b )   21     7    28                             4:2:0     230

The reference decoder produces 4940 frames from this bitstream. 4940 / 25 = 197.6 seconds, or 3 minutes 17.6 seconds.

Ron

G_M_C · 23rd November 2007, 11:24

I'll tabe a stab at guessing why this is;

My guess it that it is a TS from a digital TV-station that has used this method to "speed-up" and/or "speed-down" parts of the movie, so that it'll fit nicely bewtween commercial breaks.

I've seen movies encoded in MPEG2 with variable pull-up/pull-down rhythms, for the same reasons (Finding Nemo, ABC, Xmas last year springs to mind)

LigH · 23rd November 2007, 11:51

Thread Topic Reminder:

DGMPGDec 1.0.0a10 was already released.

Guest · 23rd November 2007, 17:48

Quote:

Originally Posted by g0blez

I have a question concerning the latest alpha 10.
I've demuxed an EVO file. evodemux claimed the source was 60fps. However, looking at the number of calculated frames and taking into consideration the total length of the stream, I calculated that the frame rate was 23.976fps.

When I opened the demuxed mpv (h.264) file in DGAVCDec, the frame rate listed was 29.970030 fps.
How should I treat this stream? Should I force film it?

Do you see frame repeats in the Info dialog when you preview it? If so, then Force Film may be correct. If not, can I please see a sample of the stream?

g0blez · 24th November 2007, 00:14

Quote:

Originally Posted by neuron2

Do you see frame repeats in the Info dialog when you preview it? If so, then Force Film may be correct. If not, can I please see a sample of the stream?

No Frame Rpts.
I get Field Rpts however; do you still need a sample?

20th November 2007, 16:52	#606 \| Link
Guest Guest Join Date: Jan 2002 Posts: 21,901	Thank you for the test stream. Fairly soon, I will be turning my attention to this libavcodec problem with some PAFF streams. Last edited by Guest; 22nd November 2007 at 00:09.

22nd November 2007, 00:28	#607 \| Link
Guest Guest Join Date: Jan 2002 Posts: 21,901	Version 1.0.0 alpha 10 * Lots of GUI bugs and crashing bugs fixed. * New info window field: max bitrate seen. * Major code cleanup and purging of DGIndex leftovers. Conversion to exclusive C++ compilation. http://neuron2.net/dgavcdec/dgavcdec100a10.zip Please delete your INI file before running this, or replace it with the one in the zip file. Note that the previous message box popup for forcing the IVTC'ed frame rate is now replaced by Force Film in the Video menu. I have not yet implemented decimation of non-film sections, so for now this will not work on hybrid video. With the disappearance of a lot of menu options, some explanation is in order. It is really very hard to integrate the container parsers with the stream buffering, NALU parsing, and parallel parsing and keep everything working right. For DGMPGDec, it is a nightmare that I still lose sleep over. I'm not sure I even fully understand how it works and if it is provably correct. I do know of several low-probability cases that can cause explosions. Yes, it's low probability but with enough uses, it's going to happen to someone. The code base is also very hard to modify and enhance. So I am taking a new tack with DGAVCDec. I want to keep it simple and provably correct. So my plan is that cutting, demuxing, container verification and repair, etc., will be done by separate tools. DGAVCDec will be for serving video only. There are existing tools for these things but I plan to make my own and make them available via the Tools menu. This approach will allow me to keep my sanity as well as to continue moving forward despite my limited time for coding these days. I don't need another massive nightmare on my hands! Probably, then, the [ and ] arrow keys will simply allow for cutting out portions of your AVC raw stream. Yes, you can use Trim() in your script, but it's more convenient this way and it's useful sometimes to generate a cut AVS raw stream as well.

23rd November 2007, 11:24	#617 \| Link
G_M_C Registered User Join Date: Feb 2006 Posts: 1,076	I'll tabe a stab at guessing why this is; My guess it that it is a TS from a digital TV-station that has used this method to "speed-up" and/or "speed-down" parts of the movie, so that it'll fit nicely bewtween commercial breaks. I've seen movies encoded in MPEG2 with variable pull-up/pull-down rhythms, for the same reasons (Finding Nemo, ABC, Xmas last year springs to mind) Last edited by G_M_C; 23rd November 2007 at 12:03.

22nd November 2007, 01:03	#608 \| Link
canTsTop Registered User Join Date: Apr 2007 Location: Lithuania Posts: 120	thank you neuron2 for your great tool. in my video, i noticed in information windows, aspect ration is 12:11 but i think it should be 15:11

22nd November 2007, 04:00	#609 \| Link
Guest Guest Join Date: Jan 2002 Posts: 21,901	It prints what is in the stream. Why do you think it should be something else? Can you post a link to the stream?

22nd November 2007, 08:00	#610 \| Link
LigH German doom9/Gleitz SuMo Join Date: Oct 2001 Location: Germany, rural Altmark Posts: 6,783	It is about pixel aspect ratio (stored pixels vs. straightened-for-display pixels) 12:11 ~ 785:720 (the width of 785 was calculated as 720*12:11)

22nd November 2007, 11:32	#611 \| Link
canTsTop Registered User Join Date: Apr 2007 Location: Lithuania Posts: 120	sorry for my mistake, i thought it was display aspect ratio. 576*15:11~785 this is my sample http://rapidshare.com/files/71446036/raw.h264.html p.s. i would be grateful if some one advise how to deal with this video, demuxed audio is 3min 17s but video is 4min 11s

22nd November 2007, 12:43	#613 \| Link
canTsTop Registered User Join Date: Apr 2007 Location: Lithuania Posts: 120	thank you for answer, this is un-demuxed TS sample http://rapidshare.com/files/71459314/ttt.ts.html

22nd November 2007, 16:05	#614 \| Link
g0blez Registered User Join Date: Jul 2003 Posts: 31	I have a question concerning the latest alpha 10. I've demuxed an EVO file. evodemux claimed the source was 60fps. However, looking at the number of calculated frames and taking into consideration the total length of the stream, I calculated that the frame rate was 23.976fps. When I opened the demuxed mpv (h.264) file in DGAVCDec, the frame rate listed was 29.970030 fps. How should I treat this stream? Should I force film it?

23rd November 2007, 11:51	#618 \| Link
LigH German doom9/Gleitz SuMo Join Date: Oct 2001 Location: Germany, rural Altmark Posts: 6,783	Thread Topic Reminder: DGMPGDec 1.0.0a10 was already released.