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. |
4th August 2015, 04:23 | #21 | Link |
Registered User
Join Date: Sep 2005
Location: Vancouver
Posts: 600
|
Hi hank315. I've experienced a little bug in HCgui 0.28 where changing the Output filename resets my Interlacing options from interlaced TFF to interlaced BFF. Searching for the issue, I saw that Blue_MiSfit had encountered the same oddity with 0.25.
I created a short video showing this easily-repeatable nuisance: https://youtu.be/3x64gG5VmWU |
16th August 2015, 21:07 | #22 | Link |
HCenc author
Join Date: Nov 2003
Location: Netherlands
Posts: 570
|
__________________
HCenc at: http://hank315.nl |
16th August 2015, 22:30 | #23 | Link |
47.952fps@71.928Hz
Join Date: Mar 2011
Posts: 940
|
Great stuff! Thanks for another update. I'd hate to see the day this ends.
__________________
Win10 (x64) build 19041 NVIDIA GeForce GTX 1060 3GB (GP106) 3071MB/GDDR5 | (r435_95-4) NTSC | DVD: R1 | BD: A AMD Ryzen 5 2600 @3.4GHz (6c/12th, I'm on AVX2 now!)
|
19th August 2015, 21:27 | #24 | Link |
Registered User
Join Date: Jan 2011
Location: Southern California
Posts: 13
|
I'm encoding progressive content for PAL and have clients complaining that the quality is terrible when played back on a TV.
Settings 1: Interlacing options is set to 'progressive'. Chroma downsampling is 'progressive', BFF Settings 2: progressive sequence is checked Hank, I just read in another thread that you recommend disabling 'progressive sequence' for PAL. So I will try this next. However, using MediaInfo I analyzed a progressive PAL encode from Apple Compressor and Adobe Encore that look great on TV for the client. The only difference between the HC Encoder file and the ones from Apple and Adobe is that the HC Encoder version specifies the 'Scan Order: Bottom Field First'. HC Encoder seems to force the scan order as TFF or BFF. There does not seem to be a way to have it not be set like the Adobe and Apple progressive PAL encodes. Any way we can disable setting the scan order?
__________________
Jon Geddes |
20th August 2015, 00:04 | #25 | Link | |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Hi Jon,
looks like you have some reading about the MPEG2 flags ahead of you... If you encode progressive content then technically there is no field order. There is only one field order flag, this is the TFF flag. It can either be set (for TFF) or not set (for BFF). HCenc behaves like any other encoder, it just sets the flag or not. Now comes the confusing part: A lot of vieving devices do observe the field order flag even for progressive content. And it is allowed to set the TFF flag for progressive content, but ONLY if the "Progressive Sequence" flag is NOT set. (The TFF flag in this case has a different meaning, it specifies frame pulldown) From an older thread: Quote:
*PROGRESSIVE *TFF The *PROG_SEQ parameter should NOT be set. The chroma downsampling parameter must only be set if you are converting color space like 4:2:2 to 4:2:0. BTW there is no need to reencode your existing source files. You can use a tool called "ReStream" to repair the flagging of your encodes. Check "Frametype progressive" and "TFF", uncheck "Progressive sequence". Cheers manolito |
|
20th August 2015, 00:32 | #26 | Link | ||
Registered User
Join Date: Jan 2011
Location: Southern California
Posts: 13
|
Quote:
Here is the MediaInfo print out for the Adobe encode: Video Format : MPEG Video Format version : Version 2 Format profile : Main@Main Format settings, BVOP : Yes Format settings, Matrix : Default Format settings, GOP : M=3, N=12 Duration : 20s 0ms Bit rate mode : Variable Bit rate : 6 483 Kbps Maximum bit rate : 7 500 Kbps Width : 720 pixels Height : 576 pixels Display aspect ratio : 16:9 Frame rate : 25.000 fps Standard : PAL Color space : YUV Chroma subsampling : 4:2:0 Bit depth : 8 bits Scan type : Progressive Compression mode : Lossy Bits/(Pixel*Frame) : 0.625 Time code of first frame : 00:00:00:00 GOP, Open/Closed : Open GOP, Open/Closed of first frame : Closed Stream size : 15.5 MiB (100%) Here is HC Encoder for a similar PAL progressive clip: Video Format : MPEG Video Format version : Version 2 Format profile : Main@Main Format settings, BVOP : Yes Format settings, Matrix : Custom Format settings, GOP : Variable Duration : 1mn 0s Bit rate mode : Variable Bit rate : 6 750 Kbps Maximum bit rate : 7 500 Kbps Width : 720 pixels Height : 576 pixels Display aspect ratio : 16:9 Frame rate : 25.000 fps Standard : PAL Color space : YUV Chroma subsampling : 4:2:0 Bit depth : 8 bits Scan type : Progressive Scan order : Bottom Field First Compression mode : Lossy Bits/(Pixel*Frame) : 0.651 Time code of first frame : 00:00:00:00 GOP, Open/Closed : Open GOP, Open/Closed of first frame : Closed Stream size : 48.3 MiB (100%) What would explain the omission of the scan order in the Adobe encode? Quote:
Here is the MediaInfo for the TFF encoded video: Video Format : MPEG Video Format version : Version 2 Format profile : Main@Main Format settings, BVOP : Yes Format settings, Matrix : Custom Format settings, GOP : Variable Duration : 1mn 0s Bit rate mode : Variable Bit rate : 6 750 Kbps Maximum bit rate : 7 500 Kbps Width : 720 pixels Height : 576 pixels Display aspect ratio : 16:9 Frame rate : 25.000 fps Standard : PAL Color space : YUV Chroma subsampling : 4:2:0 Bit depth : 8 bits Scan type : Progressive Scan order : Top Field First Compression mode : Lossy Bits/(Pixel*Frame) : 0.651 Time code of first frame : 00:00:00:00 GOP, Open/Closed : Open GOP, Open/Closed of first frame : Closed Stream size : 48.3 MiB (100%) |
||
20th August 2015, 01:34 | #27 | Link |
Registered User
Join Date: Jan 2011
Location: Southern California
Posts: 13
|
I think I found the problem. After opening my original HC encode in ReStream, it is showing the "Progressive Sequence" flag not being set. It appears I tried to specify -prog_seq in the command line instead of in the ini file.
Both the Adobe and Apple PAL progressive files have 'Frametype progressive' and 'progressive sequence' enabled. Top field first is disabled. Thanks for your help manolito. |
20th August 2015, 02:11 | #28 | Link |
47.952fps@71.928Hz
Join Date: Mar 2011
Posts: 940
|
You could also set zones for parts that need less bit rate (end credits, etc) and boost parts that would need it the most.
If you're using 2 passes, it will look better.
__________________
Win10 (x64) build 19041 NVIDIA GeForce GTX 1060 3GB (GP106) 3071MB/GDDR5 | (r435_95-4) NTSC | DVD: R1 | BD: A AMD Ryzen 5 2600 @3.4GHz (6c/12th, I'm on AVX2 now!)
|
20th August 2015, 02:27 | #29 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Glad you got it working...
Specifying Frametype progressive and TFF and not using Progressive Sequence has always worked best for me. CCE SP does it that way by default. Other encoders (all libavcodec based encoders) do it differently. They check Frametype progressive AND Progressive Sequence and do not set the TFF flag. Both ways are technically correct, and in theory any TV should play those encodes properly. But in real life things are often a little different. Cheers manolito |
11th November 2015, 12:54 | #30 | Link |
Registered User
Join Date: Nov 2004
Posts: 106
|
There is a long standing issue in HCenc. Sometimes, a single GOP happens to be of much lower quality then it should be. I found an example clip which can be used to reliably reproduce this issue: http://gratissaugen.de/test/dttest.m2v
With that certain clip, it encodes as it should with "maximum (kb/s)" set to 9000, "dc prec" to 10, "autogop" to 15, "closed gops" off, "interlaced tff" "check max bitrate" on and "Constant Quantization" set to 6.0, and B and P factors to 1.0. However, if I set quantization to 5.0 or lower, I get massive artifacting at the fast zoom out. A lower quantization setting should never lead to a worse quality, right? In the worse case, it should be the same quality, if the bitrate limit is reached. With "dc prec" set to 8, "closed gops" on, or "maximum (kb/s)" set to 9800, quantization 5.0 seems to work fine, but the issue appears again at quantization 4.0 or lower. I also encountered this issue in 2-pass encodings before, though it doesn't seem happen with this clip. I tested this with HCenc 0.28 and also with the older 0.26, 0.25, 0.23, which I still have on my disk. The behaviors in each are very similar. This is what the 4th GOP (and only that) looks like, despite high settings:
__________________
AMD Ryzen 5, 16 GM RAM, Nvidia GTX 1060, Linux Mint |
12th November 2015, 19:24 | #31 | Link |
Big Bit Savings Now !
Join Date: Feb 2007
Location: close to the wall
Posts: 1,545
|
A thought: The demanded low quantisation factor seems to fail with a bit-greedy zooming-in scene
because the quantisation factor wants to spend more bits then the given bitrate limit is willing to budget for. So starving of bits because the bitrate limit of 9Mbps prevails ?
__________________
"To bypass shortcuts and find suffering...is called QUALity" (Die toten Augen von Friedrichshain) "Data reduction ? Yep, Sir. We're that issue working on. Synce invntoin uf lingöage..." |
12th November 2015, 23:42 | #32 | Link | |
Registered User
Join Date: Nov 2004
Posts: 106
|
Quote:
I checked with Bitrate viewer: With q=6, it reports 6617 kbps on the 4th GOP, which is the highest rated GOP in this snipped, and it looks just like you would expect with that quantization factor. Also the GUI shows q=6.0 throughout the encoding, indicating that it isn't running out of bits so far. With q=5 however, I only got 2618 kbps for the same GOP while all other GOPs are encoded at a slightly higher bitrate. It also sais "underflow, fixing VBV" inbetween, which doesn't explain such a low bitrate in this GOP, though. It shouldn't be less then with q=6, right? Here is the frame log for q=6: Code:
frame type length enc.type % scantype intraVLC ------------------------------------------------------ 0 I 31543 I 0 alt 14 1 P 4736 I 0 alt 14 2 B 4291 I 1 alt 14 3 P 4113 I 0 alt 14 4 B 4239 I 1 alt 14 5 B 4370 I 0 alt 14 6 P 5432 I 0 alt 14 7 B 5675 I 0 alt 14 8 B 5199 I 0 alt 14 9 P 4194 I 0 alt 14 10 B 3922 I 1 alt 14 11 B 3456 I 0 alt 14 12 P 1960 I 0 alt 14 13 B 2956 I 0 alt 14 14 B 2882 I 0 alt 14 15 I 31507 I 0 alt 14 16 B 2776 I 0 alt 14 17 B 2655 I 0 alt 14 18 P 1967 I 0 alt 14 19 B 2942 I 0 alt 14 20 B 2738 I 0 alt 14 21 P 4020 I 0 alt 14 22 B 3166 I 0 alt 14 23 B 4524 I 1 alt 14 24 P 4405 I 1 alt 14 25 B 4864 I 1 alt 14 26 B 4821 I 0 alt 14 27 P 5307 I 0 alt 14 28 B 5244 I 0 alt 14 29 B 5370 I 0 alt 14 30 I 31627 I 0 alt 14 31 B 2789 I 0 alt 14 32 B 2695 I 0 alt 14 33 P 63167 I 100 alt 14 34 B 2593 I 0 alt 14 35 B 2065 I 0 alt 14 36 P 35611 I 1 alt 14 37 B 5914 I 1 alt 14 38 B 4114 I 1 alt 14 39 P 4676 I 1 alt 14 40 B 4738 I 1 alt 14 41 B 4579 I 1 alt 14 42 P 70276 I 58 alt 14 43 B 60786 I 41 alt 14 44 B 57494 I 57 alt 14 45 I 73639 I 46 alt 15 46 B 61944 I 55 alt 15 47 B 61766 I 53 alt 15 48 P 63069 I 1 alt 15 49 B 61942 I 28 alt 15 50 B 34353 I 8 alt 15 51 P 24723 I 1 alt 15 52 B 13314 I 2 alt 15 53 B 13796 I 1 alt 15 54 P 23062 I 1 alt 15 55 B 5529 I 1 alt 15 56 B 7968 I 1 alt 15 57 P 27498 I 1 alt 15 58 B 9710 I 1 alt 15 59 B 13866 I 2 alt 15 60 I 70448 I 3 alt 15 61 B 17358 I 5 alt 15 62 B 10420 I 1 alt 15 63 P 10852 I 1 alt 15 64 B 7119 I 1 alt 15 65 B 7109 I 2 alt 15 66 P 12819 I 4 alt 15 67 B 8020 I 2 alt 15 68 B 7384 I 1 alt 15 69 P 15433 I 1 alt 15 70 B 12978 I 4 alt 15 71 B 5563 I 1 alt 15 72 P 15384 I 1 alt 15 73 B 10523 I 4 alt 15 74 B 12876 I 2 alt 15 75 I 72902 I 2 alt 15 76 B 5229 I 1 alt 15 77 B 4219 I 1 alt 15 78 P 9271 I 1 alt 15 79 B 4538 I 1 alt 15 80 B 4003 I 1 alt 15 81 P 8422 I 2 alt 15 82 B 8330 I 1 alt 15 83 B 8346 I 1 alt 15 84 P 36146 I 9 alt 15 85 B 5656 I 1 alt 15 86 B 25560 I 7 alt 15 87 P 39102 I 15 alt 15 88 B 21480 I 1 alt 15 89 B 22497 I 1 alt 15 90 I 66711 I 1 alt 15 91 B 28659 I 17 alt 15 92 B 2525 I 1 alt 15 93 P 22115 I 6 alt 15 94 B 17177 I 7 alt 15 95 B 13263 I 1 alt 15 96 P 18307 I 2 alt 15 97 B 7179 I 1 alt 15 98 P 15145 I 4 alt 15 99 B 7603 I 1 alt 15 Code:
frame type length enc.type % scantype intraVLC ------------------------------------------------------ 0 I 35078 I 0 alt 15 1 P 5530 I 0 alt 15 2 B 4565 I 1 alt 15 3 P 5163 I 0 alt 15 4 B 4804 I 1 alt 15 5 B 4728 I 0 alt 15 6 P 6278 I 0 alt 15 7 B 6426 I 0 alt 15 8 B 5991 I 0 alt 15 9 P 5489 I 0 alt 15 10 B 4394 I 1 alt 15 11 B 3950 I 0 alt 15 12 P 3011 I 0 alt 15 13 B 3629 I 0 alt 15 14 B 3698 I 0 alt 15 15 I 35073 I 0 alt 15 16 B 2988 I 0 alt 15 17 B 2723 I 0 alt 15 18 P 2380 I 0 alt 15 19 B 3127 I 0 alt 15 20 B 2732 I 0 alt 15 21 P 4620 I 0 alt 15 22 B 3352 I 0 alt 15 23 B 5019 I 1 alt 15 24 P 5220 I 1 alt 15 25 B 5644 I 1 alt 15 26 B 5211 I 0 alt 15 27 P 6411 I 0 alt 15 28 B 6036 I 0 alt 15 29 B 6479 I 0 alt 15 30 I 35313 I 0 alt 15 31 B 2923 I 0 alt 15 32 B 2771 I 0 alt 15 33 P 68088 I 100 alt 15 34 B 3234 I 0 alt 15 35 B 2357 I 0 alt 15 36 P 42276 I 1 alt 15 37 B 7433 I 1 alt 15 38 B 5191 I 1 alt 15 39 P 5439 I 1 alt 15 40 B 4733 I 1 alt 15 41 B 4409 I 1 alt 15 42 P 76088 I 58 alt 15 43 B 67082 I 41 alt 15 44 B 60374 I 57 alt 15 45 I 16402 I 46 alt 14 46 B 16204 I 55 alt 14 47 B 15336 I 53 alt 14 48 P 14975 I 1 alt 14 49 B 12925 I 28 alt 14 50 B 16298 I 8 alt 14 51 P 16282 I 1 alt 14 52 B 12864 I 2 alt 14 53 B 13530 I 1 alt 14 54 P 15414 I 1 alt 14 55 B 4150 I 1 alt 14 56 B 6985 I 1 alt 14 57 P 12326 I 1 alt 14 58 B 9038 I 1 alt 14 59 B 13551 I 2 alt 14 60 I 79342 I 3 alt 15 61 B 21389 I 5 alt 15 62 B 12953 I 1 alt 15 63 P 13276 I 1 alt 15 64 B 8044 I 1 alt 15 65 B 8150 I 2 alt 15 66 P 15733 I 4 alt 15 67 B 9032 I 2 alt 15 68 B 8108 I 1 alt 15 69 P 20599 I 1 alt 15 70 B 15593 I 4 alt 15 71 B 7030 I 1 alt 15 72 P 22190 I 1 alt 15 73 B 12333 I 4 alt 15 74 B 18847 I 2 alt 15 75 I 82296 I 2 alt 15 76 B 6587 I 1 alt 15 77 B 5243 I 1 alt 15 78 P 12925 I 1 alt 15 79 B 5169 I 1 alt 15 80 B 4609 I 1 alt 15 81 P 11136 I 2 alt 15 82 B 9384 I 1 alt 15 83 B 9136 I 1 alt 15 84 P 43568 I 9 alt 15 85 B 6870 I 1 alt 15 86 B 30623 I 7 alt 15 87 P 47576 I 15 alt 15 88 B 26456 I 1 alt 15 89 B 28669 I 1 alt 15 90 I 73605 I 1 alt 15 91 B 34809 I 17 alt 15 92 B 3696 I 1 alt 15 93 P 27362 I 6 alt 15 94 B 20690 I 7 alt 15 95 B 15941 I 1 alt 15 96 P 22243 I 2 alt 15 97 B 8461 I 1 alt 15 98 P 19342 I 4 alt 15 99 B 9080 I 1 alt 15
__________________
AMD Ryzen 5, 16 GM RAM, Nvidia GTX 1060, Linux Mint |
|
22nd November 2015, 17:11 | #34 | Link |
HCenc author
Join Date: Nov 2003
Location: Netherlands
Posts: 570
|
@LogicDeLuxe
This is a bug, it is easily reproducible with this clip. Seems the VBV code screws up and for some reason raises the quantizers sky high. If VBV checking is turned off it runs OK but that's not the proper way to solve it Thanks for the bug report. Will investigate further. EDIT: This issue should be solved now. Can be downloaded here
__________________
HCenc at: http://hank315.nl Last edited by hank315; 21st December 2015 at 22:36. Reason: issue solved |
5th January 2016, 00:10 | #37 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
Version 0.28 much slower than 0.26?
An AVStoDVD user over at VideoHelp just reported a much longer encoding time after switching from 0.26beta to 0.28. He was using 1-pass VBR for a 2h 30min source file. The only difference in the encoding parameters was that he used Spline16Resize for HC 0.26 and Lanczos4Resize for HC 0.28. For HC 0.26 his encoding time was 2h 22min, for 0.28 the encoding took 3h 20min. This seems like a huge difference. Is there any explanation for this? Cheers manolito Last edited by manolito; 5th January 2016 at 21:53. |
5th January 2016, 20:16 | #38 | Link | |
Registered sausage
Join Date: May 2012
Posts: 78
|
Quote:
The logs clearly indicates a pirated file. |
|
5th January 2016, 21:55 | #39 | Link |
Registered User
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,079
|
@ Andouille
Alright, I wasn't aware of the source... (and I also refuse to be the copyright police, and since you are not a moderator, you should report such posts to a mod instead of reminding me to read the rules) Deleted the link in my last post. But the original question is still valid: Why is HCenc 0.28 so much slower than version 0.26? Cheers manolito Last edited by manolito; 6th January 2016 at 00:33. |
5th January 2016, 23:18 | #40 | Link | |
Join Date: Mar 2006
Location: Barcelona
Posts: 5,034
|
Quote:
I just tried to reproduce this with a similar source file and the same settings, 0.28 is slightly faster in my case. The idle priority setting may influence the encoding speed when something else is running in the background or the user is doing something else during the encode. |
|
|
|