HC encoder

[Prodater64]

HCBatch_0.14 doesn't show accurate info in main window (neither in log file).
As example, source total frames (155731) doesn't match processed ones (175028). I see that in CQ processing mode.

http://img249.echo.cx/img249/1917/hc3wa.png

Thanks to ImageShack for Free Image Hosting (http://www.imageshack.us)

[feedback]

@JohnG
I use the BDVD matrix in both, and HC is consistently better (kbps range 1500-2500

I used HC and tried the BDVD matrix, however the resulting DVD would not play
on one of my standalone players. Do you have this problem and if not how
did you fix it?

Regards,:)

[hank315]

HCBatch_0.14 doesn't show accurate info in main window (neither in log file).
As example, source total frames (155731) doesn't match processed ones (175028). I see that in CQ processing mode.Just found the error and fixed it.
The source total frames should be correct.
Thanks for reporting it.

[Encoder Master]

I saw the correct total frames only in the log file. :)

I used HC and tried the BDVD matrix, however the resulting DVD would not play

I heard of this problem. But I can't believe it because I never hear of imcompatibility with a q- matrix.
Do you use another matrixes of HC which also deletes few details like HC or perhaps autoQ matrix?

[PJones2101]

Just wanted to say thanx to Nic for a GREAT compressor. Your hard work is well-appreciated. I can compress directly from VOB files to a m2v stream at 720x576 (PAL Of course) at a BitRate of 1600 Ave and Max of 3000 using HVXBest. Average file size for 90 min is a about +-1.1gig to 1.2gig. Add the ac3 and a final DVD of 1.4-1.5 Gig. Absolutely amazing work. And the quality is still good enough to watch on PC. I remember someone asking if you were going to make a DLL out of the encoder, but I can't remember seeing an answer.
Once again, thanks for the great work ....... Keep it uphttp://forum.doom9.org/newreply.php?do=newreply&noquote=1&p=661971#
http://forum.doom9.org/newreply.php?do=newreply&noquote=1&p=661971#

[onesoul]

@ PJones2101

You are a bit confused aren't you?

And what kind of link is that? lol

[jeanl]

I think PJones2101 meant to post in the Quenc thread... But go easy on him, that was his first post! :D
I'm sure nick appreciates just the same!
jeanl

[PJones2101]

Sorry Guys, I meant Hank, I has been awake working for 48 straight hours so please forgive.

[PJones2101]

Thanks JeanL - Appreciate the assist.

[jeanl]

Sorry Guys, I meant Hank, I has been awake working for 48 straight hours so please forgive.
48 straight hours? Are you out of your mind?! :D I hope it's fun at least!
jeanl

[Prodater64]

CQ OPV encoding with prediction.

HC-Qmatic (http://forum.doom9.org/showthread.php?t=95428)

[peteag]

what's OPV?

[onesoul]

One Pass Variable bitrate.

[EpheMeroN]

Is it normal to get some VBV Underflows during an encode using HC Encoder v0.14 Beta? I got 27 during my last 2-pass encode using 6470 avg and 8000 max bitrate. I didn't worry too much though because it said that all 27 were fixed. I did notice that I got more underflows though when I used LimitedSharpen in my script.

[ljpp]

Has anyone tried encoding the AC/DC - Family Jewels (PAL) DVD with HC?

I think it could be considered as a problem case for HC, as the quality reduction seems worse than I would have expected.

This is most visible in the last song of Disc 2 - "Are You Ready". Like most AC/DC videos, it consists of shaky camera, audience clapping hands, and some colorful light effects. Especially through the blue lighted sequences, shots of the band members and audience I got pretty bad/rough blockyness. This is clearly visible with the 29" 4:3 Sony TV. Now the source ain't perfect either - crap in crap out, as they say. But still this is the worst case I've seen so far with HC.

I applied MenuShrink and then encoded with DVD-RB .93 Free+HC .14. The source is interlaced, reduction is around 70% and overall bitrates above 5000kbps for both discs.

Any tune-up ideas?

[hank315]

Any tune-up ideas?Yes, the next release...
It should do a better job with these kind of sources.

Is it normal to get some VBV Underflows during an encode using HC Encoder v0.14 Beta? I got 27 during my last 2-pass encode using 6470 avg and 8000 max bitrate. The average bitrate is already pretty high so yes this can happen.
As long as the muxing program doesn't complain about buffer underruns it should be OK.

[genesabas]

Hello,

To hank315, When the next version of HC?


Thanks for your work.

[ilpippo80]

And with which improvements?
I think we're all so curious! ;)
Thanks again!
Pippo

[LigH]

One thing that may be whistle-blown:

Loading AviSynth scripts under Windows 2000 shall be fixed now.

[jeanl]

hi hi hi, LigH, couldn't resist huh! ;)
jeanl

[LigH]

Why not. It took a while, and several test builds (mainly for katjarella and me), but the last one worked reliably in several test cases.

[hank315]

When the next version of HC?Next release is almost finished.
And with which improvements?- compressed database, size will be 30% of the old one
- improved motion estimation, means better quality especially for high motion scenes
- biased compression curve
- shutdown command

And yes, the W2K Avisynth issue seems to be solved thanks to numerous tests done by Katjarella, LigH and Jeanl.

[dbloom]

- compressed database, size will be 30% of the old one
Speaking of which, what kind of format is the database in? I'm thinking of revising my mpeg-2 utility (http://davidbloom.home.mchsi.com/readme/readme.html) to use HC instead of TMPG (tmpg, being GUI only, requires a lot of complicated repetitive user interaction to get things to work right). I would need to read and change the encoding bitrates HC has for various sections of the video, based upon the framerate they will be slowed down/speeded up to.

If this would be practical with HC, that would be wonderful, because TMPGEnc is getting kind of messy for this.

[STR]

To hank315,

just wondering if there could be a 3 pass conversion

P.S.
Awesome tool ;-))

[ukb008]

I don't know. A tool that needs more than 2 passes has to have some drags in the code. Or that's what I always thought.

Regards.

[Trahald]

I dont know if its been mentioned, but are there plans to support chapter setting? ie setting a i-frame (kindof a force scene change) at given frames of an encode based on a list. i have written support for hcenc in an app (it uses trims to split by chapters) but generally being able to encode in one chunk would be desirable.

[Hostile_Monk]

Thanks for all the hard work Hank, I've been using this encoder a lot lately. Any word on v.15b? :)

[hank315]

About HC015, actually it's finished.
ATM it's being tested by some people (I hope :) )
If the tests run OK it will be released in a few days.

@dbloom
The database as it is now is not record based anymore, it's just a continuous compressed bitstream.

@STR
HC will always be a 2 pass encoder.

@Trahald
Chapters (forced I-frames with closed GOPS) will be in the next version I think.
Tried to implement it but still have some problems to get it to work with the scene change detection.

[Trahald]

@hank315
great to hear!

[hank315]

New release: HC015

- better motion estimation
- compressed database
- added *BIAS command to tweak the compression curve
- added *SHUTDOWN command
- as usual some bugs fixed

Get it at http://hank315.dyndns.org/HC_015.zip

Will also be in the download section soon.

[LigH]

Thank you, hank! That's a fine birthday gift for me today! ;)

[hank315]

@LigH
Congratulations, 30, a real milestone... Already a long time ago for me :)

And thanks again for testing to get the W2K problem solved.
Thanks also go to Katjarella and Jeanl for this.

[lunario]

thanks you for this version hank :)

about bias settings for best quality what do you thing is the most used setting ? RB uses 25 in cce by default ( i think) is that a good setting in HC ?

hc gui default is 0

any help on this would be apreciated

[feedback]

I know the CCE guys don't like to hear it but I use your encoder almost exclusively.

HC is just as good in quality output as CCE IMHO.

I do test different Matrices and get mixed results of course (some not made for mpeg2). I use HC in DVD-RB Pro. and the Newer Rockas Rebuilder Editor v.0.1.8 which now directly imports (don't have to convert to text files anymore) .CQM and .XCM Matrices that you can use in DVD Rebuilder Pro v.93.2

Regards,:)

P.S. I like LigH's Custom Quant. Matrix Editor 1.0, also.

[jdobbs]

Great job, as usual.

:D

[hank315]

@lunario
Hard to say what the "best" setting is, it will depend on the source etc.
Did some test encodings with bias 30 which looked pretty good.

@jdobbs
Thanks...

Some additional info:
The new database used will be much smaller, for 1 hour video at full DVD resolution the size is approx. 250 MB.
For those who need it, dc precision 11 can also be used, just remember it's NOT DVD compliant.
On my system (P4) the new version runs a little bit faster, about 5%
Differences in quality for the encoding profiles are now smaller, so using NORMAL instead of BEST should give some speed gain.

[lunario]

thanks hank :)

[i4529]

To Hank315
Great encoder!!

Would the new release 0.15 fixes the "Hcbatch.exe -Application Error" below ?

[The Instruction at "0x08ea8bfc" referenced memory at "0x08e139c8".The Memory could not be "read".
Click on Okto terminate the program]

[Frameserver]

@LigH
Congratulations, 30, a real milestone... Already a long time ago for me :)

And thanks again for testing to get the W2K problem solved.
Thanks also go to Katjarella and Jeanl for this.


@LigH congratulations to your birthday, 30 is indeed a milestone :cool:

@hank315 will test that version, especially the gui version, because I'm one of the "W2k won't work" members. Let's see what happens this time :sly:
Btw. hcbatch works fine all the time, use it much, great encoder. :thanks:

frameserver

edit: Test succeeded hc_015 works on W2k with avisynth now. Great Work !! :D

[DK]

my first experiences with 015 are:

- 015 works considerably slower

a 100min source took me:
Phase II ENCODING completed in 234 minutes with 015

Phase II ENCODING completed in 156 minutes. with 014


at first i didnt realize it had to do with the new version, because i was building a new system and thought: *high time you optimized your system*

i did 4 conversions with different machine settings using 015 until i tried 014 on the new hardware again and i was reliefed to find the hardware was fine :rolleyes:

[freelock7]

Good job Hank. HC15 becomes slowly my favourite encoder.
Picture quality is increased once again!
What about the new bias setting? Same as CCE?

[vhelp]

:helpful: with HC.

FWIW, (and after 10 pages) I finally got HCenc to work on my w98 gold.
So, this is my test with HC.

I did notice quality was very good :)

However, I have three issues of concirn I found so far:

Does HC have problems with windows 98 resources ?? ..because I found that
it had depleated my resource down to 3% and when I closed HC to regain,
my system hung, and I had to reboot. This was happening when I was
working w/ an AVS script on a short 60 second VOB file as a test file.
If memory serves me, I peformed about 8 encoding scenarios, before the
resources gave out.


Can not open .d2v files. I put the "DGDecode.dll" inside my \windows\system
folder, and even in HCenc folder, where my VOB; *.d2v; and AVS script files
are located, but it does not want to open the .d2v file.
.
The error message is:
can not load movie, error in D2V file ?


YV12 color space. When I was finally able to work with it through
AVIsynth script, and opened the script
and encoded it with HC, and once encode to MPEG, the color space seemed
to be too bright.. sort of washed out.
.
Here is my script history.. After making my .d2v file from dvd2avi v1.76
I opened in VFAPI and converted to psuedo avi file. Then, through a script,
I imported the avi file, and finally, encoded inside HC.
.
--> VOB --> dvd2avi 1.76 --> VFAPI --> AVS script --> HCenc GUI
.
I was hoping to utilize the directness of HC's power when working with
sources in same format (ie, YV12 sources) and the .d2v route I thought
was the best one to use, because I'm under the assumption that the .d2v
file color space is YV12 -- now, I'm not sure, either way, if the above
proves too light, will the .d2v route prove the same results ??


One question ...

If I check this box, does that mean, [ ] Constant Q = CBR ??


Another fwiw.., I found that interlace doesn't seem to agree with HC
(for the moment) [u]But, progressive sources seem to encode without much
blocks, if any. And, the speed is perfectly acceptable, given my
CPU: AMD XP 1800+ setup w/ 512mb ram. I am impressed with this new encoder, HC :)

-vhelp

[Guest]

Regarding the D2V file opening, you have to make the D2V with the DGIndex program that matches the DGDecode you are using. You can't use DVD2AVI. Get the matched DGIndex/DGDecode pair here:

http://neuron2.net/dgmpgdec/dgmpgdec.html

For best results, don't use VFAPI or open the D2V directly. Make the D2V with DGIndex and then make an AVS script like this:

loadplugin("dgdecode.dll")
mpeg2source("yourfile.d2v")

Then load that script into HCEnc. This will deliver YV12 to HCEnc with no color space conversions. It also gives you all the flexibility and power of Avisynth scripting.

[vhelp]

Ah, yes. Thank you. I will D/L as you suggested.

However, my W98 resources still remain worrysome to me. It seems to
be depleating on me.. now down to 19% and counting. If I close HC down,
I regain 8 percent, brining me back up to 27% percent. hmm.. Anyways.

.
.

Later on..

I wanted to share my current status so far.

The YV12 color space brightness problem has been adressed and solved..
things are looking much better now. Thanks Neuron2.
.
oh, fwiw.., I did not use your suggestion w/ going the script route.
Since I do not have a need to filter my source (ie, ivtc; resize; color
space; or function creativity purposes, etc) I prefer to go the direct
route. In my recent endeavors w/ HC, I fed the .d2v file directly, and
all worked well. Color space was back to normal, according to my usual
taste.

.
.

The encoding is working fine though. And, the quality is looking better
than before. But, it seems that [ ] Constant Q does not equal CBR. Oh well.
.
I won't say that HC passes TMPGenc in terms quality, but it is definately
approaching it. My weighing scale is different than most. Since I do
mainly CBR 9000 bitrate in TMPGenc, I can't make any fair comparisons.
So Hank315, if you're lurking, you're out of the heat. Nah..
just kidding.

-vhelp

[Guest]

HCEnc doesn't do CBR.

Hank will have to address your resources issue. Why don't you upgrade to XP?

[LigH]

A constant quantizer cannot result in constant bitrate, as long as the content of the video changes: Low detailed scenes will compress much better, high detailed scenes (due to more different quantization values for high frequencies) instead harder.

[vhelp]

Regarding the resource issue. I was just mentioning it here fwiw,
since it wasn't reported.

And as for the reason why I don't install XP ..
(will try and use minimal words) Although I do have XP (home edition)
it is too bothersome for me to deal with my tipcle crashes, and
reinstallations and setups for my multi-pupose system, and then having
to resort to re-registering it to MS every time I crash, or make a change.
There are other reasons beside these.


Thanks LigH, for your CQ explanation. I was too quick to realize this.

In the mean time, I will continue to play with HC.. ( under win98 )

-vhelp

[Guest]

XP crashes for you more than 98? That's a shocker.

[hank315]

What about the new bias setting? Same as CCE?Don't know, can't look at their source code :)
Short explanation: HC calculates a minimum bitrate using the values for max. and average bitrate.
Using this min. and max. bitrate the (VBR) compression curve is calculated.
Using the bias command means the compression curve is pulled towards the average bitrate.
I don't know if the method and scale is the same as CCE, using a value bias=30 produced pretty good results during my tests.

@vhelp
HC doesn't use that many resources, if the GUI version is started you can see it will use about 125 MB virtual memory.
Because HC itself doesn't use dynamically allocated memory it will NEVER use more...
If I start an avisynth encode memory use goes up to 260 MB on my system,
starting a d2v project memory only goes up to 148 MB.
I also tested HC with DVD Rebuilder on a system with 256 MB installed using XP without any problem.
And you may have your reasons still using W98 but I can still remember all those restarts I didn't ask for...

[zilexa]

Don't know, can't look at their source code :)
Short explanation: HC calculates a minimum bitrate using the values for max. and average bitrate.
Using this min. and max. bitrate the (VBR) compression curve is calculated.
Using the bias command means the compression curve is pulled towards the average bitrate.
I don't know if the method and scale is the same as CCE, using a value bias=30 produced pretty good results during my tests.


I guess, if your average bitrate is high enough (3mbps or 4 or higher?), you could set the bias to 30... but if your average bitrate is lower (like 1.9-2.5 with cce when encoding 8 episodes of Lost(xvid) to 1 DVD) perhaps it is advicible to put the bias a little bit lower? like 20?