HC encoder

[LigH]

This must be the one. The "Sequence End Code" is usually important for other uses than DVDs. And it can confuse the decoder when several videos get joined so that such a sequence end code appears in the middle of the stream.

But still, because a sequence end code appears only at the end of an encoded file, I wonder why it shall influence the quality of the beginning of the video... Thanks forward for anyone who can explain this in detail.

[urvieh]

I'll try and turn off scene detection first. Don't expect further results before tomorrow night MET :) .

[hank315]

But still, because a sequence end code appears only at the end of an encoded file, I wonder why it shall influence the quality of the beginning of the video... No, it doesn't.
The sequence end code just marks the end of a sequence and is in fact obsolete for these kind of encodings.
For DVD it's required that every GOP starts with a sequence header, this header contains info about aspect ratio, frame rate, horizontal and vertical size etc.
But you're not allowed to change this header info, only after a sequence end code these items can be changed in the next header.
This way you can concatenate two video streams with different aspect ratio's etc.
You may even use a sequence end code after each GOP but some players don't handle that very well...
So the normal way to do it is to put one sequence end code at the end of the stream.

@urvieh
Could you test it with the next settings:
*PROFILE NORMAL
*GOP 12 2
*NOSCD
*DC_PREC 9

[urvieh]

@urvieh
Could you test it with the next settings:
*PROFILE NORMAL
*GOP 12 2
*NOSCD
*DC_PREC 9

I will. My hc.ini will then look like

*PROFILE NORMAL
*GOP 12 2
*NOSCD
*DC_PREC 9
*maxbitrate 9500
*dc_prec 9
*logfile i:\filme\test.log
*wait 0

I'll use DVD Rebuilder Pro.

[cmsoliveira]

@urvieh

What version of predator do you have? I own the special edition, region 2, and i've just backup using rebuilder (0.93,free) and hc0.15 and the result was "perfect". Absolutly no problems like the ones you've described (just finished analysing the m2v files created by hc).

My ini for this movie is

*MATRIX MPEG
*PROFILE BEST
*NOSEQ_ENDCODE
*LOGFILE c:\predator.log
*WAIT 0

All the other parameters are defined by rebuilder (gop size, dc_pred...) or are default of hc0.15, with an average bitrate of 4220 KB/s.

[johnhamler1]

use profile best if you want a better quality!com'on !??!!

[urvieh]

@urvieh
What version of predator do you have? I own the special edition, region 2

I have the same as yours, I suppose. The movie looked fine on the display, but the pio 520 can't handle it.

[urvieh]

use profile best if you want a better quality!com'on !??!!

Normally -- yes. But we are still testing what's causing the problems.

[urvieh]

Now that's weird again: I don't seem to stand a chance of turning scene detection off, although this has been done in DVDR and the ini file.

I'll play around with it for a while and keep you posted.

[cmsoliveira]

urvieh, have you tested the backup on any other player besides your pio 520?

also, does the problem only occured on the parts you've mentioned, i.e, in the beginning of the movie? Perhaps it occurs due to low bitrate...

[urvieh]

I have no other stand-alone device. Maybe it is of importance: The TV set is a 100 Hz-device.

The blocking occurs in every area where there is rapid movement. I haven't had this with version 0.14 or any other encoder/transcoder.

I've just encoded MI1 again, but will burn it tomorrow.

[LigH]

Blocking on rapid movement ... my "guess of the day": Interlaced material was encoded in "progressive mode".

[Heathcliff!]

@Hank

Hank, could you please comment on my previous post about RoBa? Would it be possible to create a database during CQ (OPV) encoding and then do an additional VBR encoding pass using this database?

Thanks,
Heathcliff

[urvieh]

Blocking on rapid movement ... my "guess of the day": Interlaced material was encoded in "progressive mode".

Not sure about that. HCBatch showed clear "interlaced" and "tff" settings.

[urvieh]

@urvieh
Could you test it with the next settings:
*PROFILE NORMAL
*GOP 12 2
*NOSCD
*DC_PREC 9

Alright. I've just burnt the dvd and watched the beginning of the film. No change, blocking all the way. I'm sorry.

[johnhamler1]

why the sound and the image doesnt match?(no synchronisation)
the sound come 0.5 seconds after or before!
by chance it is only on 2-3 movies and only for a few minutes.

[urvieh]

Okay, latest results:

Encoded movie: Mission Impossible 1, main title only.

Profile NORMAL, no sequence end code, etc. => heavy blocking
Profile BEST, no hc.ini, just the parameters handed over by DVD-RB => heavy blocking to the same extend, at the same places at the same time.

I gave QuEnc a chance, and it has turned out fine, although a bit undersized.

[hank315]

@urvieh
I also did a test:

DVD-RB PRO 0.93.2 and HC015
movie: Matrix Revolutions, PAL
movie only, profile best, matrix QLB, interlace disabled
Encoding time: 167 min. (movie: 124 min)

There are some pretty heavy action scenes in it but it played OK on my PC (using PowerDVD) and on my HW player.

In your case the movie plays fine on a PC but not on your HW player.
Blocking appears in heavy action scenes where motion vectors can be long,
seems your player doesn't like that, however it is DVD compliant.
Until now I didn't receive any other messages about this blocking.
Will PM you...

@Heathcliff!
What you want isn't possible with HC.
Ofcourse I can write a database file during a CQ run and do a second pass,
this is also the normal way for HC to do a 2-pass encoding:
CQ run --> construct compression curve --> run VBR second pass.
But the second pass requires that the first pass is done with a fixed CQ and I really mean fixed (set by HC)
otherwise the bitrate control doesn't work properly in the second pass.

[Heathcliff!]

CQ run --> construct compression curve --> run VBR second pass.
But the second pass requires that the first pass is done with a fixed CQ and I really mean fixed (set by HC)
otherwise the bitrate control doesn't work properly in the second pass.

Thank you for this explanation. But this actually raises some more questions for me. You explain that in the first pass of VBR encoding a fixed CQ-factor is determined. I wonder how you can do this. Most encoders would need multiple passes in which the quantisation / bitrate database is fine-tuned everytime. Every pass the values are closer to a constant quantisation factor. The other way to do this is by sampling the video and estimate a CQ-factor. There are some CQ-calculators around to do this (http://forum.doom9.org/showthread.php?t=96205). But the problem is that they do not provide an acurate videosize. And this is what RoBa is all about. With the data of the first CQ-pass a second VBR-pass can be done to get an acurate filesize for the resulting video. You somehow do all this natively? I am really curious about the technique you use in the first pass to build the database for the constant quantisation database. Can you explain this a bit more?

Thank you,
Heathcliff

PS. uit welke plaats in Nederland kom je eigenlijk?

[Marius-the-Mad]

hank315, I wanted to say "thank you".

Your encoder is very, very good. Thank you for creating and maintaining it - and thank you even more for doing it for free. HC is my encoder of choice. :)


Best regards,
MTM


P.S. Will you accept a minor-bug report ? ;) In the log file, HC reports a usage of HVSGOOD when HVSBEST is selected - and vice versa. But judging by quality, HVSBEST is being used when I select it. Sorry I didn't have time to check it and make sure - I just discovered it.

<edit> Bitrate with CQ seems to confirm it's only a log issue.

[TFM_TheMask]

Hi,

Everytime I use HCBatch I got an read address error at the end when HCBatch is closing.

This is the commandline I use:

HCBatch.exe -i "F:\The FilmMachine\Programma\Output\TFM_Avisynth_Script_File.avs" -o "F:\The FilmMachine\Programma\Output\TFM_Enc_Video.m2v" -b 9608 -maxbitrate 9808 -aspectratio 16:9

I have tried to change the parameters but it doesn't change anything.

Does anyone has the same problem or knows how to solve it?

[Mug Funky]

a little info aboot your system? all the machines i've tried batching on have been fine (p3, a p4 and another p4 with hyperthreading turned off)

[PJones2101]

Hey.

Once again Hank does a brilliant job. Guys please remember, Hank does this in his free time so lets appreciate his work.

Suck up !!!!!!!!!!!! :-)

Hank: When encoding PAL movies with a low bitrate ( Ave:1600 Max:4000 ) I get a brilliant picture and a 90 min movie to +- 1.2 Gig. However I noticed night areas with a minimum difference in color cause blocky playback ( PC with WinDVD through a TV out ). Could it be the matrix ( HVSBest ) ??

[urvieh]

a little info aboot your system? all the machines i've tried batching on have been fine (p3, a p4 and another p4 with hyperthreading turned off)

A normal p4 system with standard components only. I think Hanks is tracking down things. As soon as he has found a solution, he'll inform you. I'm not sure if I'm allowed to give details here. :)

[TFM_TheMask]

Don't understand me wrong I really appreciate the work Hank is doing, but I think he like to know if somethings don't work on some computers.

My specs are:

AMD Athlon 64 3400+

[urvieh]

PIV3200 (Prescott), 1 GB DC-RAM

[PJones2101]

Ya of cource. I'm not bickering. I'm just reminding people about the fact. I also posted a problem/question ( proberly mine, not his ).

I was just concerned people forgot this is a part-time project and expect him to jump.
I'm full up for reporting bugs and asking help.

Just a reminder. That's all.

[hank315]

@TFM_TheMask
This error has been reported a couple of times but because i've never seen it happen on my system it's hard to find.
Will PM you.

@Marius-the Mad
You're right, HVSGOOD and HVSBEST are switched in the logfile of the GUI version, will correct that, thanks for reporting it.

@PJones2101
Those dark scenes can be hard to encode.
Try the bias command, it will give some more bits to these scenes which might reduce blocky playback.
Also you could try the NOTCH or HC matrix.
And ofcourse, an average of 1600 is pretty low ;)

[johnhamler1]

got an error 6, it says: there was a problem to rebuild, something like that.
what is that?

i did 10-20 encode with no problems...why i have some problem sudenly?
Ok, I have canceled some FBI warning with, and I have controlled if DVD shrink loads the movie to be sure the files are in a correct order.they are!

but what is error 006?

updated: well, my problem is with DVD rebuild, old version!

[skipper152]

Hi Everyone,

You can see my other post here...
http://forum.doom9.org/showthread.php?p=677711#post677711

@TFM_TheMask
I had the same problem, but I managed to fix it (not sure how). I can give you what I did, not in any particular order, since this is over several days of trial, error and several late nights. Use the information at your own risk!

My System: Athlon 64 3500+, 1GB Ram

- AOL was starting it's connectivity service. Changed this to manual and hc015 seemed to work. I then changed it back to Automatic, but I could not duplicate the problem of the memory error.
- Deleted everything in the c:\windows\prefetch directory
- Regedit. Deleted references to hcbatch_015.exe, hcgui_15.exe, hcbatch_014.exe, hcgui_14.exe after reviewing the entry.
- External commands were not working (like more, fc). Found when I did a set, my path was coming up with %SYSTEMROOT%, rather than c:\windows. Internal commands (dir) worked fine. Changed the environment path from %SYSTEMROOT% to c:\windows where necessary, and my external commands started working.
- Changed the comatibility of the HCGui to win 98 / me

It doesn't seem like anything should have made a difference. Both applications seem self contained (using only system dll's).

Now for the kicker....
In my previous post, I mentioned that my laptop Pentium M was working fine. Came in one day and it started having the memory problems with hcgui_015, 14 is still working fine. I tried all of the stuff above, but still nothing.
- This machine had avisynth 2.5.6 installed on it. Last I remember, this was still in development. I uninstalled and reinstalled 2.5.5, but still no success.

With both machines, I have been installing and uninstalling some of the other great programs people here have been working so hard on to try them out.

@Hank
I'm not sure if this will help, but I can get HCgui_015 with the memory error if I just load an avs script, not change any configuration on the screen or do any processing, but just exit the application. If I don't load a script, I can exit the application with no problem. This same script works fine in 14. I haven't tried a .d2v file.

Hope this helps someone...

[skipper152]

Sorry for posting again so soon, but I just tried something. Maybe it will give Hank some more firepower during the debugging process....

This script doesn't work....

AVISource("E:\Down\test\test.avi")
LanczosResize(352,128,0,0,640,256)
AddBorders(0,48,0,64)
KillAudio()

But this script with the 1 line does

AVISource("E:\Down\test\test.avi")

I've tested this several times on my pentium M just now...
I've loaded the bad first or the good first, but the bad keeps giving me the memory reference error. I've also tried different resizings in the script too.

I've attached the files (1st time I'm doing this), so hank can use exactly what I am using. Not that is should make any differnce, but the avi referenced is XviD-1.1.0-Beta2-04042005 _Beta Release_

Skip

[TFM_TheMask]

I confirm that it has to do with the Avisynth Script.

I think it has to do with the line AviSource, because when I change this line in DirectShowSource the error doesn't show (I only tried with the gui and not with the batch, will test that tomorrow)

The only thing I did is open the scriptfile and then close HCgui.

[TFM_TheMask]

Ok did test HCBatch with this script:

# 16:9 encoding
DirectShowSource("D:\JamesBondTest.avi", fps=23.976)
ConvertToYV12()
LanczosResize(720,356,0.0,0.6)
AddBorders(0,62,0,62)

and it worked without the error. When I use the old script:

# 16:9 encoding
AviSource("D:\JamesBondTest.avi", false)
ConvertToYV12()
LanczosResize(720,356,0.0,0.6)
AddBorders(0,62,0,62)

I got the read error. So I think something is wrong with reading and releasing the command AviSource.

[LigH]

Which AviSynth version - 2.55?

[hank315]

I also tried these scripts.

My sytem: P4 3.2 (Prescott), 1 GB memory.
Avisynth version: 2.56, build: jan 31 2005

DirectShowSource("test.avi")
ConvertToYV12()
LanczosResize(720,356,0.0,0.6)
AddBorders(0,62,0,62)

runs OK.

Avisource("test.avi",false)
ConvertToYV12()
LanczosResize(720,356,0.0,0.6)
AddBorders(0,62,0,62)

also runs OK.

Whatever I try it just always works OK...

[LigH]

Then there must be some garbage in the ICM / VfW system, maybe several colliding codecs. Either due to unusual registration (e.g. by an evil codec pack), or due to evil code (one example: AngelPotion), or due to misconfiguration (yes - you can ruin it through ffdshow with several RAW video settings)..

[TFM_TheMask]

I use Avisynth 2.56, build sep 28 2004 on AMD 64 3400+

Will try the new build.

[TFM_TheMask]

@Ligh

Why is it then that CCE and QuEnc and Procoder are working correctly with the same scripts?

[Inc]

I dont know if HC encoder uses a diff. way but usually serving an avi (avs) stream or frames within an applikation to an encoding procedure will be done via the vfw api.
Means AVIfileOpen(), AVIFileReadStream() etc.
And so the system internal vfw core does just use (imho) the framework of avisynth.dll to serve the bitstream out of the avs resulted textfile.

Means ..... I do still think your avs setup is messed up or something like that.

[LigH]

AviSynth 2.56 beta 3 messes the video (see here (http://forum.doom9.org/showthread.php?p=676120#post676120)) - better stay with AviSynth 2.55

[TFM_TheMask]

Just de-installed version 2.56 and installed version 2.55 but no difference, still the read error.

Maybe it only happens on AMD 64 processors?

[DrShrinker]

I read the posts by TFM_TheMask and wanted to try his scripts on my brand new AMD 64 3000+ in Windows XP. I also get a read error:

The instruction at "0x085d8bfc" referenced memory at "0x086036e8". The memory could not be "read".

Here's what I fed HCBatch:

HCBatch.exe -i "F:\Temp\Movies\Sample\test.avs" -o "F:\Temp\Movies\Sample\test.m2v" -b 9608 -maxbitrate 9808 -aspectratio 16:9

Here's the avisynth script:


#ASYNTHER AVISource
AVISource("F:\Temp\Movies\Sample\sample.avi")
ConvertToYV12()
LanczosResize(720,356,0.0,0.6)
AddBorders(0,62,0,62)


I have AviSynth version 2.55.

HCBatch works great with DVD Rebuilder, but then again that doesn't use AVISource.

Is AMD 64 the problem?

[DrShrinker]

I just tried the same script in HCGUI_015.exe (same version as the HCbatch I used) and I did not get an error message when the encoding finished. But when I tried to close the GUI, I got the error message. I used the same settings, too: average bitrate 9608, maxbitrate 9808, 16:9. Weird.

[doxville]

I read the posts by TFM_TheMask and wanted to try his scripts on my brand new AMD 64 3000+ in Windows XP. I also get a read error:

The instruction at "0x085d8bfc" referenced memory at "0x086036e8". The memory could not be "read".

Do you have ffdshow installed? I usually get this kind of error, when I installed ffdshow with the makeAVIS-part. When I choose to leave makeAVIS uninstalled, i never get this message. That's some mystery to me, why ffdshow crashes different programmes when closing them. :confused:

[LigH]

Just found out (I hope not too many people posted this without me noticing that):

There is a "provider" for Cuttermaran to support HC as "GOP recoder":

http://prdownloads.sourceforge.net/cuttprov/

[hank315]

Maybe it only happens on AMD 64 processors?Yes, what I've seen so far it only happens on AMD processors.

Maybe I should switch to AMD to see what actually is going wrong :eek:

[DrShrinker]

Do you have ffdshow installed?

Nope.

[Xeno86]

@TFM_TheMask: What OS do you use? Is is Windows XP 32 or 64 bit?

[TFM_TheMask]

Windows XP 32 bit with SP2.

[DrShrinker]

Windows XP 32 bit with SP2.

Ditto here. But my processor is the AMD Athlon 64 3000.