Log in

View Full Version : HC encoder


Pages : 1 2 3 4 5 6 7 8 [9] 10 11 12 13 14 15 16

Rumbah
22nd August 2005, 14:33
For some people it helped to install the Helix YUV Codec (http://www.lillevold.com/files/yuvcodecs-1.2.exe).

LigH
22nd August 2005, 19:08
Other people installed DivX 5, or XviD, or enabled YV12 conversion in ffdshow's "raw video" processing. No need for outdated simplest micro "codecs" ;)

Rumbah
23rd August 2005, 16:49
Why is the Helix Codec outdated? It is old, but small and fast and can do everything you want and I never had a problem with it. For some people also deinstalling Xvid and installing another YV12 capable codec helped, but I think installing the Helix codec is the simplest solution. It works and does not interfere with other codecs.

LigH
23rd August 2005, 18:07
Okay - not really "outdated"; but I presonally prefer a few multi-purpose codecs over many tiny specialized ones. The Helix codec may be one tiny helper codec, being able to convert YV12. The ATI YUV codecs are another alternative of tiny specialized codecs.

But indeed, I never needed them, because DivX 5 or XviD 1.x (correctly installed) already offered just the same functionality, and ffdshow is configurable to support it.

And: How could they "interfere"? With what?

Rumbah
23rd August 2005, 19:07
Yes, of course you could use Xvid or Divx. But first, there were some versions that did not handle YV12 correctly for example if the dimensions were no multiple of 16. And second, they both do not support saving in YV12.
And for the interference, well, I had some problems having Divx, Xvid and ffdshow installed that sometimes lead to strange results.

Asmodeus
26th August 2005, 09:55
First of all I must admit that I didn't read whole thread.
I got one question to hank: is it possible to set fixed quantizer wor each frame type, egz. Q2 for I, Q3 for P, Q4 for B ? or is it possible to implement this in new build.
Thx

sorry for my english

Pookie
26th August 2005, 21:38
Wondering about HCenc and HDTV Resolutions. Is it a lot of work to modify HCenc to be able to do 1280x720 @ high high bit rates ?

Regardless, thanks again for this great encoder :)

Pasqui
27th August 2005, 11:19
@hank315
When do you expect to release HC016? I'm a bit worried about the non compliancy of MV restriction you talked about in this post (http://forum.doom9.org/showthread.php?p=702377#post702377). Is current version safe to use ?
Thanks for all your work !

Pasqui.

hank315
30th August 2005, 22:51
When do you expect to release HC016? I'm planning to release a HC016 batch test version in a week or so.
One of the things that have been changed is the way avs scripts are handled, hope this will solve the crashes some people report.
Also the MV range will be fully MPEG compliant again so the blocky playback on some HW players will be solved.

Wondering about HCenc and HDTV Resolutions. Is it a lot of work to modify HCenc to be able to do 1280x720 @ high high bit rates ?It's quite alot of work but yes, it's one of the things I want to implement but not restricted to 1280x720, max. resolution will be 1920x1152, some day :D

is it possible to set fixed quantizer wor each frame type, egz. Q2 for I, Q3 for P, Q4 for B ?This also will be in one of the future versions but not in the next version.

Zeul
30th August 2005, 23:12
I for one am looking forward to the next build. :D

Marius-the-Mad
30th August 2005, 23:21
You're not alone. :D

hank315 - thank you again for all your work. HC is amazing. :)


<edit>

P.S. May I have a feature request ? :) How about something like matrix.ini, where we could store additional matrices, which would then be selectable from within the programme ? ;)

LigH
31st August 2005, 12:46
In case of need, don't hesitate to ask for specific beta tests, hank. ;)

Xeno86
1st September 2005, 17:01
@hank315
And what about resource leak in mode preview in Win 9x? Did you fix it?
Thanks again for your great encoder :)

john3voltas
3rd September 2005, 15:36
Also the MV range will be fully MPEG compliant again so the blocky playback on some HW players will be solved.

Hi Hank,
I've been using your encoder on a regular basis for quite some time already and may I let you know that I find it wonderful.
But now I see a few remarks that it might not be 100% MPEG compliant and that gave me the creeps.
Does that apply to all HCEnc versions or just the latest version?
Because you see, I'm still using 0.14 ;).
Other than that keep up with such a wonderful work :thanks: .

Carpo
4th September 2005, 11:58
i have done many encodes with hanks encoder and my rather picky standalone plays them all - so it should be ok :)

john3voltas
4th September 2005, 16:14
@Carpo
I see, but it obvious that HC has compliancy issues otherwise Hank wouldn't have posted such words.
And that scares me a bit.
So I just wanted to know if these issues have been there all the way from the early releases or if they were just introduced in the latest releases.
Cheers

hank315
8th September 2005, 19:06
New release, a test version of HCbatch016 only.
Just want to see if it runs more stable now and all options are working OK.
But also some improvement in quality and speed.

link: http://hank315.dyndns.org/HCbatch_016_testrelease_08-09-2005.zip

Changes:

- improved auto gop option
- fixed a bug in Motion Estimation which could cause blocky playback on some HW players
- for fixed GOP 15-1, GOP length sometimes was set to 16-1, fixed
- added bias and dc precision as parameters for HCbatch
- small GUI changes
- as usual some other bugs fixed

Recommended avisynth version: 2.5.6 RC1.
It should work OK with DVD-RB.

Frameserver
8th September 2005, 19:13
New release, a test version of HCbatch016 only.
Just want to see if it runs more stable now and all options are working OK.
But also some improvement in quality and speed.

link: http://hank315.dyndns.org/HCbatch_0..._08-09-2005.zip

Changes:

- improved auto gop option
- fixed a bug in Motion Estimation which could cause blocky playback on some HW players
- for fixed GOP 15-1, GOP length sometimes was set to 16-1, fixed
- added bias and dc precision as parameters for HCbatch
- small GUI changes
- as usual some other bugs fixed

Recommended avisynth version: 2.5.6 RC1.
It should work OK with DVD-RB.

Hi hank315,
can you check the link?
Got a http 404 error - file not found.

cu
frameserver

hank315
8th September 2005, 19:18
@Frameserver
You're right, didn't work, should be fixed now.

Frameserver
8th September 2005, 19:39
@Frameserver
You're right, didn't work, should be fixed now.

@hank315
sorry I don't think so.
But this time server not found or dns error.
Server busy/down? :confused:
I try it later again
cu
frameserver

jeanl
8th September 2005, 19:41
Same here, can't connect to hank315.dyndns.org, timing out...
jeanl

LigH
8th September 2005, 19:47
@ Frameserver:

hank315.dyndns.org

is usually a PC at home, which gets an IP address for at most 1 day by the Internet Service Provider. Each time a new IP arrives, this new IP must be published to the DNS servers worlswide, this may take some time. Although it may just take minutes... Try regularly each few hours until you got it.

Zeul
8th September 2005, 19:47
:)
All d/l and now to try

hank315
8th September 2005, 20:13
My connection is very unstable at the moment :(

@LigH
Yes it runs from my home server but I use a static IP, just checked it, it's still the same as in the dyndns database.

Just saw the download counter go up again :) seems I'm back online but the speed is low...

katjarella
8th September 2005, 20:56
Unfortunately, still same like under: http://forum.doom9.org/showpost.php?p=698348&postcount=387

I am surprised because it "almost" functioned in the last internal Beta.

see Result: http://katjarella.gleitz.info/hc/

Darksoul71
8th September 2005, 21:11
@frameserver:
FYI: Download worked fine 4 me.

-D$

hank315
8th September 2005, 21:17
@Katjarella
DirectShowSource is still a PITA :mad:
Sometimes it can not be rewinded to the first frame after the first pass, I tried to unload and load the clip again for the second pass but that didn't work if you use avisynth directly.
I tried a wmv clip with 2800 frames using DirectShowSource, if I let HC encode the whole clip it didn't work, with the command *FRAMES 1 2600 the first 2600 frames are encoded OK.
Seems if DirectShowSource reads until the end something is screwed up.

jdobbs
8th September 2005, 21:44
@frameserver:
FYI: Download worked fine 4 me.

-D$Not for me...

hank315
8th September 2005, 21:51
Not for me...Strange :confused:

link again: http://hank315.dyndns.org/HCbatch_016_testrelease_08-09-2005.zip

katjarella
8th September 2005, 23:02
@Katjarella
DirectShowSource is still a PITA
http://katjarella.gleitz.info/smilies/smilie20.gif YES YES YES.

1: load AVS -> 1st pass -> close avs
and then
2: load AVS -> 2nd pass -> close avs
does not work ?

jdobbs
8th September 2005, 23:23
Strange :confused:

link again: http://hank315.dyndns.org/HCbatch_016_testrelease_08-09-2005.zipWorks now...

Frameserver
9th September 2005, 17:03
Works now...

Jep - got it too

jmvolfan
9th September 2005, 17:16
In testing HC .16 beta with DVDRB Pro I am seeing a strange problem. During encoding HC pops up and encodes a segment but when it starts a new segment it will not start until I click on the program link in my start bar. Once I click on it HC opens on the screen and the encoding begins and runs normally. After is finishes it closes and a new segment is loaded but does not encode until I click on it again.

I am running Win XP Pro SP2 with an AMD 64. I have the latest DVDRB Pro release and using the 3 click method.

I ran the same data with CCE and everything ran fine.

Thanks for the great encoder Hank. Hope this info helps with testing.

2COOL
9th September 2005, 18:22
In testing HC .16 beta with DVDRB Pro I am seeing a strange problem. During encoding HC pops up and encodes a segment but when it starts a new segment it will not start until I click on the program link in my start bar. Once I click on it HC opens on the screen and the encoding begins and runs normally. After is finishes it closes and a new segment is loaded but does not encode until I click on it again.

I am running Win XP Pro SP2 with an AMD 64. I have the latest DVDRB Pro release and using the 3 click method.

Though I'm using 0.94, I can confirm that this is happening to me too. I just did King's Ransom R1 back to back. The first one with with v0.15A and the next was v.16 beta (test release). I had no "hanging" problems with 0.15A.

HC Encoder seems to hang alot during the encode process. On numerous occasions, I had to abort the encode and restart at last encode to get it going again. I'm using WinXP Pro SP2.

OS: Windows XP Professional Service Pack 2
BIOS: A M I - 11000306
Motherboard: ASUSTeK Computer Inc. P4P800
CPU: Intel(R) Pentium(R) 4 CPU 3.00GHz
RAM: 1024 MB SDRAM
Video: NVIDIA GeForce FX 5700 Ultra

DK
9th September 2005, 23:01
is *suppress warning prompts* UNchecked in your rb versions?

DK
10th September 2005, 00:05
i didnt test this latest 016 test release before and i do experience the same flaw:

hc stays minimized and does not keep going until i clicked on it


the 016 version you sent me a couple of days ago, hank315, did not show this behaviour and worked flawlessly

DK
10th September 2005, 00:54
for anyone interested in testing this latest release with dvd rebuilder, use this workaround:

go to setup and DISABLE *run encoders minimized*

this way hc will stay on top and keep encoding one segment after the other

at least this works for me this very minute

zambelli
10th September 2005, 07:37
A question about interlaced encoding:

Do the field order checkboxes in the GUI refer to the field order of the input source or the output MPEG?

I'm encoding DV video to DVD and I'm not sure what to set it to. I know DV is BFF, but I also know most (all?) DVDs are TFF. So should I reverse the field dominance manually in Avisynth, or should I just set this to checkbox to BFF and expect the DVD player to handle it correctly?

Ebobtron
10th September 2005, 15:10
@zambelli

BFF TFF settings in HC are used to set flags in the final .m2v.
HC will not change the field dominance.

Boulder
10th September 2005, 15:30
A question about interlaced encoding:

Do the field order checkboxes in the GUI refer to the field order of the input source or the output MPEG?

I'm encoding DV video to DVD and I'm not sure what to set it to. I know DV is BFF, but I also know most (all?) DVDs are TFF. So should I reverse the field dominance manually in Avisynth, or should I just set this to checkbox to BFF and expect the DVD player to handle it correctly?
Your standalone should play the video fine even if it's encoded and flagged as BFF, I don't think TFF is mandatory.

Marius-the-Mad
10th September 2005, 19:45
I tried the new beta out. There are some problems with the picture, however - some strange blocking artifacts appear. :(

I tried a few decoders and here are the results:


The source:
http://img253.imageshack.us/my.php?image=source341tl.png

ffdshow - libavcodec:
http://img253.imageshack.us/my.php?image=ffav348ts.png

ffdshow - libmpeg2:
http://img227.imageshack.us/my.php?image=ffl2341wc.png

MPC - libmpeg2, AFAIK (please ignore the colour differences):
http://img227.imageshack.us/my.php?image=mpc346lg.png

DGDecode 1.4.4 b3 (based on heavily patched reference decoder):
http://img227.imageshack.us/my.php?image=dg340gn.png


It seems only DGDecode can play the file correctly. I made a sample available here:

http://uploadhut.com/view.php/311163.m2v

And here's the log (hmm... should the number of underruns really be that high ?):
http://uploadhut.com/view.php/311228.log


I haven't watched the film from the start to the end and am not sure whether those black blocks appear in other parts of the frame. :(


I'm using Athlon XP 1700+. Hank, do you need any additional information ?

Boulder
10th September 2005, 20:34
What's your script?

Marius-the-Mad
10th September 2005, 21:10
The script is not necessarily very sane - I was experimenting both with some functions and new HC. As you can see from the screenshot however, there are no blocks in the source.

LoadPlugin("ColorMatrix.dll")
LoadPlugin("DGDecode.dll")
LoadPlugin("RemoveGrain.dll")
LoadPlugin("MaskTools.dll")
Import("soothe.avs")
Import("pctv_smooth.avs")


AVISource("sw.avi")
s1=last.PC2TV_Smooth()
s2=s1.ColorMatrix(mode="rec.601->rec.709", mmx=false)
s1=s2.BlindPP()
Soothe(s2, s1, 50)
Lanczos4Resize(704,416)
ColorYUV(opt="coring")
Crop(0,0,0,0,true).RemoveGrain(mode=1)
AddBorders(8,80,8,80)

# making the screenshot
#
#Lanczos4Resize(1024,576)
#ConvertToRGB32(matrix="rec709")

Boulder
10th September 2005, 21:27
Did you try just feeding the clip to HC without any additional filters? That would doublecheck if it's really an HC issue. I'm not even sure if I can tell what black blocks you mean (wouldn't be the first time though :D )

Marius-the-Mad
10th September 2005, 21:35
Hmm... No, not yet. :) And as for the blocks... Look at the right side of the source image, and then try the one decoded with libavcodec. ;)

Boulder
10th September 2005, 21:45
The encoded clip shows fine here, I tried PowerDVD, VDubMod and Media Player Classic (Elecard MPEG2 decoder decoded the clip).

Marius-the-Mad
10th September 2005, 22:00
So, it seems only libavcodec and libmpeg2 have problems with it, after all. Which is a bit scary nevertheless, because these are probably two the most popular libraries out there. :(

I guess we have to wait for Hank to verify whether the problem lies on the encoding or the decoding side.

<edit> Thank you very much for additional test, Boulder. :)

<edit #2> WinDVD decodes it fine, too.

hank315
10th September 2005, 23:39
Well, I played it with:
- PowerDVD --> no problem
- BSplayer (libavcodec) --> wrong
- BSplayer (libmpeg2) --> wrong but not in the same way as libavcodec
- BSplayer (Elecard) --> no problem

MPEG Stream Eye (Elecard) also OK
MPEGAnalyser (don't know what decoder) also OK

This is a strange encode, average non-linear quant = 1.9, so very low.
Seems the source has very little information so the quants must be low to keep up with the required average bitrate.
This also explains why there are so many underflows, with quants between 1 and 2 it's very hard to predict the bitrate...
The MacroBlocks with display errors all have quant=1.

Maybe you should try to encode it again (or only that part) with a lower average bitrate or even better a fixed quant (4 should be a good value for testing).

Marius-the-Mad
11th September 2005, 00:11
This is a strange encode, average non-linear quant = 1.9, so very low.
Seems the source has very little information so the quants must be low to keep up with the required average bitrate.
This also explains why there are so many underflows, with quants between 1 and 2 it's very hard to predict the bitrate...The source is not particularly detail-rich, indeed. Thank you for checking the sample, Hank. :)

However, I need a statement from you. :) Could you, please, tell me, whether I should file a bug report to libavcodec developers or wait until you check HC further ? It seems HC outputs fully correct MPEG 2 elementary stream, but I need to be sure - so no one will flame me on ffmpeg-devel. I already found them one problem yesterday. ;)


One session with VirtualDubMod later...

Hank, am I right the maximal GOP length in my sample equals 18 ? HC displayed 18, too, but I wasn't sure it was the correct value, since it's beta after all. If I'm not mistaken, HC beta makes GOPs one frame longer than specified. :(

Sorry for finding a different problem to bug you! :devil: :) And terribly sorry if I managed to set something wrong.

DK
11th September 2005, 01:38
@ hank315

the latest016 TR version you sent me, works fne with dvd-rb and option *run encoder minimized* set