HC encoder

[ukb008]

Thanks, artoor.

Regards.

[Prodater64]

Can CQ_MAXBITRATE option be selected from command line?

[hank315]

Can CQ_MAXBITRATE option be selected from command line?No, that's not possible ATM.

[Rumbah]

Hi!

It would be great if you could set the VBV size because it is needed as I'm writing a plugin for Cuttermaran (www.cuttermaran.de) to support HCEnc. For that project I also wanted to know if HCEnc sets the SequenceDispayExtension (Because Cuttermaran only encodes small pieces of the original Stream, the author said it is important to set it if it was used in the original stream, and a SEC would be MPEG compliant, but not DVD compliant).

[ljpp]

@hank
Uh... always hard to answer this, let's say when it's finished

Dammit Hank - now I can't do any backups because I know that a better version is lurking around the corner :scared:.

I have used HC a lot recently, and it seems constantly to deliver very good results with normal natural source DVD backups. Can't wait for the next version.

p.s. Check your PM Hank!

[artoor]

I'm waiting for new version as well... it looks as if Hank teaches us patience ;)
But seriously - can You tell us (more or less) when the newest version will be release??

Regards

[dragongodz]

can You tell us (more or less) when the newest version will be release??
considering some of the things hank315 is working on it wont be probably released tommorow... or the day after that. :D
seriously though i have talked to him and know some of the things that he is working on and they take time, pure and simple.

and before anyone askes, NO I CANT TELL YOU WHAT HE IS WORKING ON. sorry but he said he will come around and put me through a grinder if i did. ;)

[artoor]

I didn't speak with Hank but I think that I know what He is working on :D (have a guess) It can be 2 important things ;) If I predict well I'll tell You after the newest version will be relesed :)
I'm waiting impatiently

[hank315]

but he said he will come around and put me through a grinder if i did.:) Did I really say that ?
Would never do such a thing, you're just living too far away :D

@ljpp
Already read it, will answer it ASAP.

Next version will still take some time to finish, for the things I want to implement the old database structure wasn't flexible enough.
That part is now completely rewritten, also the database is now compressed, size will be about 30% of the old one.
That was just alot of stupid programming and testing...

Now just see if the ideas I have really work and that again is alot of programming and testing.

[Easy123]

Just tested the Encoder together with DVD Rebuilder and I was stunned.. @Hank: great job done! Thx for this Software... But if I want to use this software as "standalone" is there a possibility to tell the Encoder which frame has to be coded as an I-Frame per commandline (for Chapters)????

[artoor]

Ok... I'll tell You something... As far as I'm concerned I'm waiting for some day when I will see the newest HC version and on the changelog.txt there will be write "- support for Hyperthreading/multiple processor PC's" :D I dream of this ;) I hope that I can have dream like this at last it is a part of "ToDo list"... Am I right?? :dreamer:

Regards

[dragongodz]

:) Did I really say that ?
Would never do such a thing, you're just living too far away :D
well you asked me not to say anything about the things being worked on and planned etc. to me thats close enough as people who know me will know that i dont divulge things when specifically asked not to. :D
as for distance, ye just a little far away. ;)

[ilpippo80]

Originally posted by Easy123
...But if I want to use this software as "standalone" is there a possibility to tell the Encoder which frame has to be coded as an I-Frame per commandline (for Chapters)????

Me too waiting for the possibility to force I-frames, I think that RB is a very good program but I'd prefer to use directly the gui version of hc...
If I remember well, some time ago Hank said that it wasn't urgent and it will be done sooner or later... but I hope that it will come sooner!:p

BTW, thanks for your great work, Hank!:D

[jeanl]

I have started to use HC extensively for my interlaced DV -> DVD conversions, and so far, I'm really happy with the results, both in terms of quality and encoding speed.
I found however, that it was very tedious to have to write avi scripts for each avi file I needed to encode (I usually break my DV footage into many small avi files which are easier to edit and manipulate, and I want to encode them independently to make DVD chaptering easier).

To help with that task, I wrote a tiny application that can scan a folder for all the avi files, and generates a bunch of avi scripts in a "AVIScripts" subdirectory, along with a .bat file that contains the HCBatch commands.

Typically, you would start the app, drag/drop one of the avi files into it, and press "Create Batch". The app assumes there's a HC.ini file that contains your encoding setup. There's a button to edit that file too.

Here's a screenshot.
http://img281.echo.cx/img281/5259/test5kb.png (http://www.imageshack.us)
Anybody who's interested can download the application here (http://jean.laroche.free.fr/Binaries/HCBatchMaker1.0.zip).

If there's enough interest, I'll start a new thread to discuss improvements, but at this point I'm just offering this as a freebie, just in case somebody might find it useful.

jeanl

[jdobbs]

I like to use segmented AVI files with my DV material... I do a lot of encoding of DV to DVD.

[jeanl]

Originally posted by jdobbs
I like to use segmented AVI files with my DV material... I do a lot of encoding of DV to DVD.
Out of curiosity jdobbs, which encoder do you use? I'm assuming your DV material is interlaced, right? What bitrate do you consider appropriate?

I'm still struggling to decide how low a bitrate I should go for, considering this is for archiving but also for viewing. I currently use at least 5500Mb/s.
jeanl

[jdobbs]

I usually encode DV using OPV (CCE) so I can aim for a Q factor as opposed to a bitrate. The only time I wouldn't would be when I'm worried about getting it to fit. I normally set the maximum bitrate to 9000.

Contrary to what I've heard from some others, I am very satisfied with the interlaced output I get from CCE, especially the newer versions. I am considering switching to HC, though... as it seems to handle interlaced sources very well -- and Hank315 has added a (maximum) bitrate controlled OPV mode now.

[Encoder Master]

But the interlaced quality of HCE isn't very well at the moment. Only if you deinterlace your material before giving it to HCE.

How good are your interlaced encodes with HCE?

[jdobbs]

What's wrong with the output from HC -- I haven't noticed any problems?

[Encoder Master]

I never tested it before because my sources are always progressive and so I've only very good results. But in the last time I've see some test pics of QuEnc and HCE with interlaced sources and the QuEnc gives in this test much more quality. The pic of HCE looks not so good like progressive. Does anybody noticed such problems with interlaced sources?
Many people say it that HCE have problems with interlaced sources.:(

[Archimedes]

I’ve made a lot of tests with different encoders and interlaced sources. Subjective and objective tests. The objective tests was done with Fritz Framalyzer.

One of the testclip, i used, is located at
http://www.tecoltd.com/enctest/enctest.htm (test.avi, interlaced).

As a result, the software calculate an average euclidean distance in RGB. Euclidean distance in RGB means the average distance between an encoded pixel and an original pixel in the RGB color space. Lower Values means better quality.


Here are some results:

Canopus Procoder Express
Highest Quality, VBR, 2 pass, 3500, 6500, 8500, DC 8
Euclidean distance in RGB: 7,3347

QuEnc
VBR, 2 pass, 6500, 8500, DC 10
Euclidean distance in RGB: 7,5422

HC
VBR, 2 pass, 6500, 8500, DC 8
Euclidean distance in RGB: 7,8581

FreeEnc
VBR, 2 pass, 3500, 6500, 8500
Euclidean distance in RGB: 8,0893

Cinema Craft Encoder Basic
VBR, 2 pass, 3500, 6500, 8500, DC 8
Euclidean distance in RGB: 8,2326

TMPGEnc
High Quality, VBR, 2 pass, 3500, 6500, 8000, DC 10
Euclidean distance in RGB: 8,7692


The same order comes out with subjective tests. In all my tests, the Procoder produces the best results with interlaced sources. QuEnc produces very good results on interlaced sources also (but it's also the slowest encoder of all).

Josef

[onesoul]

Why did you use dc 10 on quenc and tmpeg?

[Archimedes]

The best result with QuEnc was the setting with DC 10. I didn’t tried many other settings with TMPGEnc, because i saw no chance to get much better results.

Another setting was:

TMPGEnc
Without B frames, High Quality, VBR, 2 pass, 3500, 6500, 8000, DC 9
Euclidean distance in RGB: 8,3593

Josef

[zilexa]

I am creating 3 dvds with each 8 episodes of Lost (source is Xvid/mp3, 350MB,42 minutes, have chosen to do 8 episodes per dvd since I didn't see quality difference with 6 episodes on one dvd with CCE SP 2.70, 4 pass).

Now I am considering switching to a free encoder... since I am using DIKO I was very interested in FreeEnc, but after reading lots of tests and forums... I cannot decide between FreeEnc and HC.

It seems HC and QuEnc are at the same level, QuEnc is able to produce better quality sometimes, but so is HC.... Unfortunately there arent any tests between HC and FreeEnc! That's why I cannot decide.

since these encodings require low bitrates.. and VMesquita has optimized his FreeEnc for low bitrates.. I suppose FreeEnc is the better choice for now..
any comments?

[Encoder Master]

Without B frames

This settings aren't good. The best quality you get with 1 B-Frames in TMPGEnc and also the best compression but this isn't DVD-compliant.

[Mug Funky]

1 b-frame is compliant, isn't it? otherwise scene-detection and autogopping wouldn't work...

[JohnG]

Originally posted by zilexa
I am creating 3 dvds with each 8 episodes of Lost (source is Xvid/mp3, 350MB,42 minutes, have chosen to do 8 episodes per dvd since I didn't see quality difference with 6 episodes on one dvd with CCE SP 2.70, 4 pass).

Now I am considering switching to a free encoder... since I am using DIKO I was very interested in FreeEnc, but after reading lots of tests and forums... I cannot decide between FreeEnc and HC.

It seems HC and QuEnc are at the same level, QuEnc is able to produce better quality sometimes, but so is HC.... Unfortunately there arent any tests between HC and FreeEnc! That's why I cannot decide.

since these encodings require low bitrates.. and VMesquita has optimized his FreeEnc for low bitrates.. I suppose FreeEnc is the better choice for now..
any comments?

HC is better quality than FreeEnc. I use those two encoders almost exclusively. For anything other than a full backup, I use the BDVD matrix in both, and HC is consistently better (kbps range 1500-2500 is where I usually fall for multiple movie, captures, etc. on one DVD) It's simple, when I don't care about the quality difference and want pure speed, FreeEnc is the choice. The quality for HC however, is as good or better than ANY encoder out there (free or otherwise). I have used Procoder, TMPGEnc, CCE etc. When it comes to reliabililty and quality it can't be beat. It's just nice to have all the available choices :D

[Encoder Master]

@Mug Funky

1 b-frame is compliant, isn't it? otherwise scene-detection and autogopping wouldn't work...

Oh, really. I only used 1-B Frame for my ONE-CD-Encode and a very low bitrate to improve quality and save some bits. I don't know how good is the quality on other encoders but with TMPGEnc I got more quality.

I have used Procoder

This make me happy because this is the only encoder which I don't have and so I can't test the quality.:)

[Fl0ppy]

I'm new using HC, but the quality and speed are amazing!! I have a cpu with HT.
It's very good with low bitrate like 1500 etc...in CVD :D

Can somebody try this new matrix for HC?

http://rapidshare.de/files/1880244/ivcd_1_1_0.zip.html

cu.:D

[LigH]

@ Fl0ppy:

Accordiny to my "Custom Quantization Matrix Cross-Compare" (CQMXComp), your iVCD matrix is very similar to "Auto-Q2" by Avalon. (The used similarity metric is a combined Root Mean Square over both intra and inter matrices.)

CQMXComp is a CLI tool, and included in

http://www.ligh.de/software/qmatrix.zip

Usage: ?:\*> CQMXComp *.cqm [...] [> output.txt]

Values near 0.0 mean "not very similar", values near 1.0 mean "very similar", exactly 1.0 means "the two matrices are identical".

*.cqm files are the same plain binary matrix files as used in XviD (*.xcm) or ffdshow/ffvfw (*.qmatrix).

[Encoder Master]

Oh, this is a nice tool. I will test it. Very useful. Thank you LigH.;)

[Fl0ppy]

@Ligh
Thanks for the tool, it's nice. And now i have more than 60 matrices to test it! :rolleyes: It' s lot of matrix.

Wich it's better? The original of mpeg? The tmpgenc..? I think depends of the person and the film.

cu.

[LigH]

Originally posted by Fl0ppy
Wich it's better? ... I think depends of the person and the film.
Of course! It always depends heavily on the subjective preferences, the material, and the target. Can you guess why one matrix was called "28 Days After"? - It was optimized for personal needs on this special movie. Several matrices even had a whole paragraph in a readme which described the exact circumstance they were made for.

There are matrices which are useful for high quality, but require a lot more bitrate to achieve it - in my CQME, you can see that they are rather flat/low. Other matrices can save bitrate, but will reduce details - they raise more quickly, look like mountains. Around half the matrices were made for MPEG-1/2 encoders, their authors may not even know if they can be recommended for MPEG-4 encoders at all!

[zilog jones]

HC (using the GUI) locks up on loading AVS files on my PC (Win2k), unless I turn off the "resident shield" on my anti-virus program (Grisoft AVG Free).

Anyone else having problems with the GUI freezing should try turning off their anti-virus too. Does anyone know why this is happening?

[jeanl]

I think this was mentioned before on this thread or a previous one, the fact that HC froze on win2k, but I don't remember reading about the anti-virus being responsible for that!
jeanl

[doxville]

It was mentioned some time ago by me, but I don't think anyone took notice (scroll down a little in the thread): http://forum.doom9.org/showthread.php?s=&threadid=88888&postid=615630
Sometimes, even when I switched off the AV-guard, the GUI version locks up in the FIRST try, but when I close HC and open it a SECOND time, it usually works. Though i see it more like a 'cosmetical' problem, because it IS possible to use the GUI with Win2K, it would be really nice, if there was a solution to this bug...

[zilog jones]

Yes, actually, it doesn't seem to work every time first time for either when I turn off my anti-virus program. I find it works after a while of me randomly turning off a few other programs (like my Epson printer taskbar dealy), but I don't know if that's actually helping.

[jeanl]

The easiest way to get to the bottom of that would be for Hank315 to replicate the problem which apparently he can't. Next way is to create a version that outputs debugging info in a file so you can narrow the point where bad things happen. It's a bit time consuming because you have to make several tests back and forth before you can pinpoint exactly the offending command. Kind of a binary search type of approach!
I would be willing to help (I have win2K on one machine here, and it HC crashes no matter what) by testing "debug" apps... But Hank315 would have to create that first...
jeanl

[zilog jones]

I seem to be having problems with it making videos correctly too. I'm trying to make it encode an AVS file using ImageSource to make a sequence of test cards, but I'm just getting a diagonal mess. I'll do some more investigating some other time - I really should be going to bed now!

[peteag]

which is the best matrix for content of 1500-2000 at PAL resolution (high motion, sometimes heavy film-damage)?????

[LigH]

2 Mbps, heavy motion, and full resolution? You must be joking... :rolleyes:

Which matrix... In this case: Maybe the one of the Wachowski Bros.! ;)

[peteag]

No! :)

my thread (http://forum.doom9.org/showthread.php?s=&threadid=94298)

[LigH]

Then, I'd almost call it "crossposting", because your topic does not cover the HC encoder development too much...

Let's talk that question there, I'd suggest. :cool:

[peteag]

i heard hc seems to be the best in very low bitrates - so i ask you whats the best matrix for low bitrates at hc? this could be a good question because of including matrices into a new build ... isn't it?

[Guest]

Originally posted by peteag
whats the best matrix for low bitrates According to forum rule 12, there is no best and you should please not ask for it. Thank you.

[onesoul]

Originally posted by peteag
which is the best matrix for content of 1500-2000 at PAL resolution (high motion, sometimes heavy film-damage)????? Maybe you should try xvid or other mpeg4 encoder instead...

[LigH]

(@ onesoul: Your avatar looks familiar... The "killer rabbit" by Monty Python? :D)
__

@ peteag:

In general, for low bitrate encoding, you would use matrices which have rather high values (compared to the default MPEG matrix), and values that raise rather quickly from the top left corner towards the right and lover edge. Those matrices will reduce details to avoid too much block/ring artifacts. Already built-in are e.g.: QLB, BACH1, Jawor, AutoQ2.

In contrast, high quality matrices have values which raise slower than those of the MPEG matrix, or even stay below them. They will restore more detail, but will require more bitrate to achieve that goal. Already built-in are e.g.: the 3 HVS.

(HC and NOTCH are hard to decide for me ...)

Look for my thread about my "Custom Quantization Matrix Editor" CQME. You will find there an archive of many collected quantization matrices. Use them for your own experiments. And always be aware: The possible quality of the quantization result is a (multiplication) product of a per-block quantization factor, and the per-frequency-component matrix entry. Smaller products give better quality for those frequency components, but require more bitrate. As long as you don't understand this concept, you won't understand your own question about the "best matrix"!

[onesoul]

@ LigH

Good guess! Yep, "That rabbit's dynamite!" :D

[Amnon82]

@ Fl0ppy:

Accordiny to my "Custom Quantization Matrix Cross-Compare" (CQMXComp), your iVCD matrix is very similar to "Auto-Q2" by Avalon. (The used similarity metric is a combined Root Mean Square over both intra and inter matrices.)

CQMXComp is a CLI tool, and included in

http://www.ligh.de/software/qmatrix.zip

Usage: ?:\*> CQMXComp *.cqm [...] [> output.txt]

Values near 0.0 mean "not very similar", values near 1.0 mean "very similar", exactly 1.0 means "the two matrices are identical".

*.cqm files are the same plain binary matrix files as used in XviD (*.xcm) or ffdshow/ffvfw (*.qmatrix).

@all: Sorry for this (I think) offtopic post.

@Fl0ppy:

Old Times come back. Avalon was my nic at www.doom9.de. Here I used Amnon82. Greetings to LigH ;)

AUTO-Q1 (AVAMAT6/CYMAT 1.1) and AUTO-Q2 (CYMAT 1.2) are my matrices. You can find them in my programs QCCE-CYBEREDITION 1.12 (http://forum.doom9.org/showthread.php?s=&threadid=77825&highlight=QCCE) and in Auto(C)Q 1.5.3 (http://www.autoq.fx.to).

As I remember right, You wanted me once to add Your matrices called PVCD into QCCE-CE. Here the matrices to remember:

PVCD 17d

8 8 15 18 24 28 35 43
16 17 18 21 27 30 35 46
20 18 20 24 27 33 36 50
21 24 28 30 34 38 40 45
24 30 27 34 38 39 46 60
26 30 35 40 45 50 60 70
34 39 41 46 52 61 70 80
30 41 47 48 54 67 80 84

16 18 20 22 23 27 32 40
18 20 22 24 26 32 36 42
20 22 24 26 33 38 42 44
22 24 26 34 38 44 44 46
23 26 33 38 44 44 46 48
27 32 38 42 44 46 48 50
32 36 42 44 48 48 50 52
40 42 44 46 48 50 52 58

PVCD 18a

8 9 16 18 25 28 33 44
15 16 17 20 27 29 34 47
19 19 21 25 28 34 37 48
22 27 30 30 34 37 38 50
24 32 30 35 39 40 48 58
25 32 37 42 46 50 58 68
35 37 41 45 50 56 66 79
40 39 47 48 54 66 79 83

16 18 20 22 23 27 32 40
18 20 22 24 26 32 36 42
20 22 24 26 33 38 42 44
22 24 26 34 38 42 44 46
23 26 33 38 42 44 46 48
27 32 38 42 44 46 48 50
32 36 42 44 48 48 50 52
40 42 44 46 48 50 52 58

and Your newest I found today on Your Homepage (http://usuarios.lycos.es/pvcd/html/) (24e):

8 8 15 18 24 28 35 43
16 17 18 21 27 30 35 46
20 18 20 24 27 33 36 50
21 24 28 30 34 38 40 45
24 30 27 34 38 39 46 60
26 30 35 40 45 50 60 70
34 39 41 46 52 61 70 80
39 41 47 48 54 67 80 84

16 18 20 22 24 27 32 40
18 21 22 24 26 32 36 42
21 23 24 26 35 38 42 44
23 24 26 32 38 44 44 46
25 26 31 36 44 45 46 48
27 34 37 42 46 46 48 50
31 36 42 43 48 50 52 54
39 39 43 46 49 52 54 57

... are You realy shure what You're doing ... :confused:

[lbecque]

I tried 0.14 beta version on a MPEG2 file that I ran through avisynth, cropped adn resized before frameserving to HC.
Sometimes HC would just exit in the middle of its second pass with no warning or error message.
If I used the option to restart the 2nd pass it would finish.
What's that all about???
Am I doing something wrong?

The .avs script was running through DirectShowSource.
I haven't retried it with DGDecode.dll and mpeg2source yet.