AutoMKV 0.98.4

[ayfa]

Thanks.
I left all at default.
Just replaced the CPU. With the older CPU I had almost 100%.
Well, I convert my 720p videos, shot with my Samsung VP-HMX10.
I'll see, that I get some VOBs for testing. What are recommended filters? Remove Grain Mode 2 is always set. Resize Filter is SimpleResize(neutral soft). Where else are settable filters?

[chrisway]

Hi
I've been using AutoMKV 0.98.4 to convert some DVDs of mine after ripping with DVD decrypter, but wanted to speed up the process by running it all automatically.
I tried using the command line options for the AutoMKV exe but nothing seems to be working. I've tried this via cmd.exe and by adding the options in a shortcut.
I think I'm using the correct format specified in the /? option: eg
"C:\Program Files\AutoMKV\AutoMKV0984.exe" DVD.ini H:\THE_DEPARTED\TEMP\DVD\VIDEO_TS\VTS_02_0.ifo 1 0 0 0 start kill
It seems to get as far as selecting the DVD.ini profile, but doesn't load the IFO file at all, and hence nothing turns up in the log.
I've also tried this with VOB files, and both work as normal by selecting via the GUI.
I think somewhere along the line, updating the program has broken the command line functionality. Either that or I'm just doing it wrong.
Running Vista by the way. Any help would be greatly appreciated.

Thanks

[ayfa]

Sorry if I spam around, but I just don't get it working properly.
I just created a .ts of one of my blurays and loaded it into automkv. But no matter what settings I use, I don't get more than 40% CPU usage now, what ends up in processing 4fps. OS is Vista, but also tried on XP. Something is not ok here.
Can someone please post his/her settings to create a 1920 x xxx(auto crop) mkv file with a ts as source.
Current log:
20:44:48 0.98.4 - - - - - - - - - - - - - START JOBS - - - - - - - - - - - - - -
20:44:48 Selected Input file: D:\HD\movie.dga
20:44:48 Selected Output dir: C:\HD\temp\
20:44:48 Selected Final Name: movie
20:44:48 Selected Container type: MKV
20:44:48 Selected Encoder: X264
20:44:48 Selected Size: 10800 / 1
20:44:48 Selected Profile: 2_Pass_Balanced.xml
20:44:48 Selected Resizer: SimpleResize (neutral soft) / AutoCrop Only
20:44:48 Selected Filters: RemoveGrain(mode=2)
20:44:48 Selected Deinterlacer: NONE
20:44:48 Selected Field Order: AUTO
20:44:48 Track Language 1: movie.dga PID 1100 3_2ch 48KHz 640Kbps DELAY 0ms.ac3
20:44:48 Audio Codec 1: Copy Original
20:44:48 Track Language 2: NONE
20:44:48 Audio Codec 2: Coding Tech. AAC
20:44:48
20:45:38 Extracting audio track 1 movie.dga PID 1100 3_2ch 48KHz 640Kbps DELAY 0ms.ac3 with ffmpeg
20:45:39 Renamed file movie.dga PID 1100 3_2ch 48KHz 640Kbps DELAY 0ms.ac3 to audio.ac3 -1 file size
20:45:39 DGA File Size: 3772622 bytes
20:46:45 Number of Frames: 199033
20:46:45 Framerate: 23.976025
20:46:45 Movie length in Seconds: 8302
20:46:45 Movie Width/Height: 1920/1080
20:46:45 Starting X264
20:46:45 Command Line 1' Pass X264:
C:\AutoMKV0983\exe\encoder\x264.exe --pass 1 --bitrate 10258 --stats "C:\HD\temp\.stats" --ref 1 --progress
--keyint 250 --bframes 3 --qpmin 10 --qpmax 51 --no-psnr --filter -1,-1 --subme 2 --direct auto --vbv-bufsize
14000 --vbv-maxrate 45000 --me dia --no-ssim --level 4.1 --b-adapt 2 --no-chroma-me --partitions none --threads auto
--thread-input --aud --aq-mode 1 --aq-strength 1 --psy-rd 1.1:1.0 --sar 1:1 -output NUL "C:\HD\temp\movie.avs"
20:46:45 Command Line 2' Pass X264:
C:AutoMKV0983\exe\encoder\x264.exe --pass 2 --bitrate 10258 --stats "C:\HD\temp\.stats" --progress --keyint 250
--bframes 3 --qpmin 10 --qpmax 51 --no-psnr --ref 3 --filter -1,-1 --subme 9 --direct auto --vbv-bufsize 14000
--vbv-maxrate 45000 --me umh --no-ssim --level 4.1 --weightb --b-adapt 2 --no-chroma-me --8x8dct --threads
auto --thread-input --aud --aq-mode 1 --aq-strength 1 --psy-rd 1.1:1.0 --sar 1:1 --output "C:\HD\temp\movie.264"
"C:\HD\temp\movie.avs"
Now I have 40% CPU usage and an average 7fps

[buzzqw]

@chrisway

yep.. the command line help is a bit outdated..

but you can do this

open a vob , set all settings.. and add to queue
now check the queue.bat and the queue01.ini
you should found your answer :)

@ayfa
you are encoding at 1080/xxx
unless you have a fast quadcore the speed seems reasonable

also... remember that x264 is very optimized for multiprocessor.. but avisynth NO
so you could be vinctim of unsufficent decoding power

just for test you could try with megui and confron the speed (should not be very different)

BHH

[chrisway]

Thanks so much for the quick reply. You've solved my problem.:)

[ayfa]

@ayfa
you are encoding at 1080/xxx
unless you have a fast quadcore the speed seems reasonable

also... remember that x264 is very optimized for multiprocessor.. but avisynth NO
so you could be vinctim of unsufficent decoding power

just for test you could try with megui and confron the speed (should not be very different)

BHH
Yes, the file was 1080p.
My CPU is a quad core Intel Q6600 with 3GHz per core.
I will try megui then. Hope it's not complicated!
Funny thing is: when I run 1 Pass fastest profile, then speed is ok compared to my E6300 cpu at 2.8GHz but at 2pass, speed of both cpus is equal.

[Chengbin]

How come when I specify AR input, the encoding goes much faster? I usually get 33fps in first pass, I'm getting 41fps now. I don't know about the second pass though.

[Chengbin]

buzzqw, AutoMKV seems like it can't read an AAC audio file, in MKV or MP4. It just acts like I've selected to have no audio in my video.

[Jackie]

Hi,

no matter which file I try, I always keep getting this error message:

---------------------------
Attention
---------------------------
Unable to copy G:\source\movie.dga to O:\temp\movie.dga
---------------------------
OK
---------------------------

What's wrong?

My movie is not named movie.dga, I think this is a file created by AutoMKV. In both directorys, I have full (write) access, and more than enough free disk space.
Thanks alot.

[ayfa]

Well, I just ignore that message. It comes each time I start a job.
EDIT:@Chengbin
Can't confirm. Same FPS no matter manual selection or auto.

[Chengbin]

Well, I just ignore that message. It comes each time I start a job.
EDIT:@Chengbin
Can't confirm. Same FPS no matter manual selection or auto.

Yeah, I take that comment back. It doesn't make it faster.

[Chengbin]

What kind of resolution should I choose when I'm encoding anamorphic videos? I saw some anime are encoded in 720x480 for 16x9. How about 2.35?

[buzzqw]

Miscellaneus answer

buzzqw, AutoMKV seems like it can't read an AAC audio file, in MKV or MP4. It just acts like I've selected to have no audio in my video

aac audio is detected and processed, please try just remuxing your file into a new mkv (or try autox264)

faac ?
faac audio is extract wrong by ffmpeg, so automkv is unable to process. Please try updating the ffmpeg.exe

Unable to copy G:\source\movie.dga to O:\temp\movie.dga
try this. Launch automkv, and point as input file the O:\temp\movie.dga (or try autox264)

How about 2.35?
there isn't resolution as 2.35, but 16/9 or 4/3
2.35,2.40 are just 16/9 with or without black bars

BHH

[Chengbin]

AutoMKV detects the AAC audio, but does nothing to encode it, even when I choose copy original. It is correctly muxed (the IDs are in order), and I've tried it on videos I've encoded with AutoMKV with AAC sound, and it didn't work.

AutoX264 looks interesting. Can it use AutoMKV's profile?

[buzzqw]

AutoX264 use MeGui profiles and the full package already incude it

on autox264 audio is extracted by eac3to (from mkv)

BHH

[Chengbin]

How do I turn off VBV in the profile? How do I set x264 to use unlimited bitrate fluctuations?

[buzzqw]

are you sure you haven't already asked this question ?...

check vbvmax...

BHH

[Chengbin]

are you sure you haven't already asked this question ?...

check vbvmax...

BHH

What do I change it to? Some extremely large number or something?

[Chengbin]

Hi,

no matter which file I try, I always keep getting this error message:

---------------------------
Attention
---------------------------
Unable to copy G:\source\movie.dga to O:\temp\movie.dga
---------------------------
OK
---------------------------

What's wrong?

My movie is not named movie.dga, I think this is a file created by AutoMKV. In both directorys, I have full (write) access, and more than enough free disk space.
Thanks alot.

I got the same message, but if you wait a bit, AutoMKV will start encoding.

[buzzqw]

What do I change it to? Some extremely large number or something?
leave blank

BHH

[ayfa]

@ayfa
you are encoding at 1080/xxx
unless you have a fast quadcore the speed seems reasonable

also... remember that x264 is very optimized for multiprocessor.. but avisynth NO
so you could be vinctim of unsufficent decoding power

just for test you could try with megui and confron the speed (should not be very different)

BHH
Hi, I had no luck using MeGUI. It's too complicated for me.
But I tried RipBot.
Using it, CPU is @ 100% usage even in the first pass. I also have more fps. Same sample file is encoded in approx. half of the time.
Don't know whats wrong. Does RipBot use a different encoder for HD files?

[buzzqw]

no, it just use different profiles... (and so different quality!)

BHH

[ayfa]

Can using the same codec but different profile cause a cpu usage difference of 50%?

[buzzqw]

even more!

please try the 2 pass balanced or a faster one's (the 1 pass)

BHH

[ayfa]

I did that already. I wrote it in a post a few messages ago.
Damn, I don't know why.
Another question: Should the first or second pass be faster (on balanced for example).

[Chengbin]

I did that already. I wrote it in a post a few messages ago.
Damn, I don't know why.
Another question: Should the first or second pass be faster (on balanced for example).

First pass is faster, and it will only use 2 cores. It is faster because it uses whats called a "turbo" mode. First pass is mainly frame type decision and other stuff, so you can use ultra fast settings for some settings, which brings significantly faster encoding speed that hurts very little quality.

[cosmos]

There seems to be some problem with the autoresizing. I've selected xvid/avi output. In the advanced settings the following are different compared to the default values:
* Use CRF for autoresize routines is UNCHECKED
* Force ITU resize is CHECKED
* HQ Deint is CHECKED

With the disabled CRF setting, I'd expect a compressibility test to take place, based on bitrate, utilizing xvid_encraw. However, the following two strange things happen:
1) A x264 encoding pops up at some point during the process! It seems like the CRF tests always take place?

2) Checking the log, it seems like that autoresizer suggests a movie width of "1200 based on sample size". Isn't this quite optimistic? I mean, if I enable the CRF autoresize routines, a width of 448 will be selected! This is a vast difference between the two auto-resize routines. A bug somewhere perhaps?

21:12:10 0.98.4 - - - - - - - - - - - - - START JOBS - - - - - - - - - - - - - -
21:12:10 Selected Input file: C:\AutoMKVRips\MainMovie\test\VIDEO_TS\VTS_01_0.IFO
21:12:10 Selected Output dir: C:\AutoMKVRips\temp\
21:12:10 Selected Final Name: test
21:12:10 Selected Container type: AVI
21:12:10 Selected Encoder: XviD
21:12:10 Selected Size: 1/2 CD / 1
21:12:10 Selected Profile: 2_Pass_Balanced.xml
21:12:10 Selected Resizer: SimpleResize (neutral soft) / Automatic
21:12:10 Selected Filters: RemoveGrain(mode=2)
21:12:10 Selected Deinterlacer: AUTO
21:12:10 Selected Field Order: AUTO
21:12:10 Enabled CRF/CQ AutoRes at 2.5 value
21:12:10 Activate PGC Ripping: 1
21:12:10 Using Matrix: Sharktooth's EQM v3LR.cqm
21:12:10 Track Language 1: 80: English (AC3 2ch, 0xBD 0x80) [0]
21:12:10 Audio Codec 1: Lame
21:12:10 Audio Quality 1: 112
21:12:10 Audio Channel 1: Stereo
21:12:10 Track Language 2: NONE
21:12:10 Audio Codec 2: Lame
21:12:10 Advanced Audio Settings DUMP Track 1
21:12:10 Nero Option Profiles: AUTO --- Encoding Mode: Quality
21:12:10 CCT Option Profiles: AUTO --- Channel Options: AUTO
21:12:10 Lame Settings: ABR
21:12:10 Audio Normalization: 1
21:12:10 Frequency: AUTO --- Tempo: NONE --- Pitch: NONE --- Other Add:
21:12:10 Activate Force ITU Resizing
21:12:10
21:12:11 Activate PGC processing: PGC 1 [0:45:19:01]
21:12:11 Dgindex CMD: C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\dgindex\dgindex.exe -FO=0 -OM=1 -TN=80 -YR=2 -AIF=[C:\AutoMKVRips\MainMovie\test\VIDEO_TS\VTS_01_1.vob] -OF=[C:\AutoMKVRips\temp\movie] -exit -minimize
21:12:51 D2V file size: 317394 bytes
21:12:51 Audio 1: movie t80 2_0ch 192kbps delay 0ms.ac3
21:12:51 Pre Audio 1' Size: 65256960 bytes
21:12:51 Applying DelayAudio of 0 to audio file: movie t80 2_0ch 192kbps delay 0ms.ac3
21:15:28 Encoded 1' Audio Track: movie t80 2_0ch 192kbps delay 0ms.mp3 35905848 bytes
21:15:30 Number of Frames: 67965
21:15:30 Framerate: 25.000000
21:15:30 Movie length in Seconds: 2719
21:15:30 Movie Width/Height: 720/576
21:15:30 DgIndex AR: 16:9
21:16:33 Source is declared tff by a margin of 28/26.
21:16:33 Advanced Deinterlace Routines Log
21:16:33 Analyzed 399 sections of five frames
21:16:33 Found Interlace Sections: 0 0.00%
21:16:33 Found Progressive Sections:399 100.00%
21:16:33 Found Telecined Sections: 0 0.00%
21:16:33 Found Zero Mov: 100.00% 399
21:16:33 Found One Mov: 0.00% 0
21:16:33 Found Two Mov: 0.00% 0
21:16:33 Found Three Mov: 0.00% 0
21:16:33 Found Four Mov: 0.00% 0
21:16:33 Found Five Mov: 0.00% 0
21:16:33 The Movie is declared to progressive
21:16:33 AutoResize 1' Pass XviD: "C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\encoder\xvid_encraw.exe" -pass1 -bitrate 968 -turbo -progress -max_bframes 2 -imin 3 -imax 16 -quality 6 -vhqmode 1 -pmax 16 -bmax 16 -bquant_offset 0 -kboost 20 -odegrade 9 -nopacked -nochromame -bvhq -threads 0 -lumimasking -qmatrix "C:\Users\Καραγκιόζο\Documents\AutoMKV\profiles\XviD-profiles\matrix\Sharktooth's EQM v3LR.cqm" -i "C:\AutoMKVRips\temp\resmovie.avs" -type 2 -avi "C:\AutoMKVRips\temp\resmovie.avi"
21:16:33 AutoResize 2' Pass XviD: "C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\encoder\xvid_encraw.exe" -pass2 -bitrate 968 -progress -max_bframes 2 -imin 3 -imax 16 -quality 6 -vhqmode 1 -pmax 16 -bmax 16 -bquant_offset 0 -kboost 20 -odegrade 9 -nopacked -nochromame -bvhq -threads 0 -lumimasking -qmatrix "C:\Users\Καραγκιόζο\Documents\AutoMKV\profiles\XviD-profiles\matrix\Sharktooth's EQM v3LR.cqm" -i "C:\AutoMKVRips\temp\resmovie.avs" -type 2 -avi "C:\AutoMKVRips\temp\resmovie.avi"
21:17:09 Finished 1' Pass XviD
21:17:43 Finished Autosize procedure
21:17:43 Tested total movie: 5%
21:17:43 AutoResize movie size: 15560184 bytes
21:17:43 Movie Width suggested: 1200 value based on sample size (166.67% of original width)
21:17:43 FIRST pass XviD: "C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\encoder\xvid_encraw.exe" -pass1 -bitrate 968 -turbo -progress -max_bframes 1 -imin 3 -imax 16 -quality 6 -vhqmode 1 -pmax 16 -bmax 16 -bquant_ratio 66 -bquant_offset 0 -kboost 20 -odegrade 9 -nopacked -nochromame -bvhq -threads 0 -lumimasking -qmatrix "C:\Users\Καραγκιόζο\Documents\AutoMKV\profiles\XviD-profiles\matrix\Sharktooth's EQM v3LR.cqm" -i "C:\AutoMKVRips\temp\movie.avs" -type 2 -avi "C:\AutoMKVRips\temp\movie.avi"
21:17:43 SECOND pass XviD: "C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\encoder\xvid_encraw.exe" -pass2 -bitrate 968 -progress -max_bframes 1 -imin 3 -imax 16 -quality 6 -vhqmode 1 -pmax 16 -bmax 16 -bquant_ratio 66 -bquant_offset 0 -kboost 20 -odegrade 9 -nopacked -nochromame -bvhq -threads 0 -lumimasking -qmatrix "C:\Users\Καραγκιόζο\Documents\AutoMKV\profiles\XviD-profiles\matrix\Sharktooth's EQM v3LR.cqm" -i "C:\AutoMKVRips\temp\movie.avs" -type 2 -avi "C:\AutoMKVRips\temp\movie.avi"
21:26:00 Starting Second Pass
21:34:51 Only Movie Size: 329073152 bytes
21:34:51 Using VirtualDubMod for Muxing!
21:35:07 Encoding finished: 00:09:07 elapsed time
21:35:07 -----------------------------------------------------------------------
21:35:07 Script AutoCrop.log
21:35:07
21:35:07 Crop(0,2,720,572)
21:35:07 -----------------------------------------------------------------------
21:35:07 Script Resmovie.avs
21:35:07
21:35:07 #loadpluginstart
21:35:07
21:35:07 #test1
21:35:07
21:35:07 #importstart
21:35:07
21:35:07 #test2
21:35:07
21:35:07 movie = mpeg2source("C:\AutoMKVRips\temp\movie.d2v")
21:35:07 function getOrder(clip c) {
21:35:07 order = GetParity(c) ? 1 : 0
21:35:07 Return order }
21:35:07 #filter1start
21:35:07
21:35:07 #test3
21:35:07
21:35:07 last=movie.Crop(0,2,720,572)
21:35:07 fixed_aspect = 1.836111
21:35:07 out_width = 704
21:35:07 #Aspect_Error = 0.152%
21:35:07 out_height = 384
21:35:07 #resizestart
21:35:07
21:35:07 #test4
21:35:07
21:35:07 SimpleResize(out_width,out_height)
21:35:07 #filter2start
21:35:07
21:35:07 #test5
21:35:07
21:35:07 RemoveGrain(mode=2)
21:35:07 function Comptest(clip c, float percent)
21:35:07 {
21:35:07 frange=floor(14./(percent/100.))
21:35:07 Return SelectRangeEvery(c, frange, 14)
21:35:07 }
21:35:07 Comptest(last,5)
21:35:07 -----------------------------------------------------------------------
21:35:07 Script movie.avs
21:35:07
21:35:07 #loadpluginstart
21:35:07
21:35:07 #test1
21:35:07
21:35:07 #importstart
21:35:07
21:35:07 #test2
21:35:07
21:35:07 movie = mpeg2source("C:\AutoMKVRips\temp\movie.d2v")
21:35:07 function getOrder(clip c) {
21:35:07 order = GetParity(c) ? 1 : 0
21:35:07 Return order }
21:35:07 #filter1start
21:35:07
21:35:07 #test3
21:35:07
21:35:07 last=movie.Crop(0,2,720,572)
21:35:07 fixed_aspect = 1.836111
21:35:07 out_width = 704
21:35:07 #Aspect_Error = 0.152%
21:35:07 out_height = 384
21:35:07 #resizestart
21:35:07
21:35:07 #test4
21:35:07
21:35:07 SimpleResize(out_width,out_height)
21:35:07 #filter2start
21:35:07
21:35:07 #test5
21:35:07
21:35:07 RemoveGrain(mode=2)
21:35:07 -----------------------------------------------------------------------
21:35:07 Script mkvmaudio.avs
21:35:07
21:35:07 LoadPlugin("C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\filter\NicAudio.dll")
21:35:07 LoadPlugin("C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\filter\SoundOut.dll")
21:35:07 LoadPlugin("C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\BeSweet\BassAudio.dll")
21:35:07 LoadPlugin("C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\filter\FFMS2.dll")
21:35:07 NicAC3Source("C:\AutoMKVRips\temp\movie t80 2_0ch 192kbps delay 0ms.ac3")
21:35:07 DelayAudio(0.0000)
21:35:07 #Applying STEREO downmixing routines
21:35:07 function stereo(clip a)
21:35:07 {
21:35:07 flr = GetChannel(a, 1, 2)
21:35:07 fcc = GetChannel(a, 3)
21:35:07 lfe = GetChannel(a, 4)
21:35:07 lfc = MixAudio(fcc, lfe, 0.2071, 0.2071)
21:35:07 mix = MergeChannels(lfc, lfc)
21:35:07 lrc = MixAudio(flr, mix, 0.2929, 1.0)
21:35:07 blr = GetChannel(a, 5, 6)
21:35:07 Return MixAudio(lrc, blr, 1.0, 0.2929)
21:35:07 }
21:35:07 #
21:35:07 6==Audiochannels() ? stereo() : last
21:35:07 function addvideo(clip c) {
21:35:07 blankclip(length = Int(100 * AudioLengthF(c) / float(audiorate(c))), fps=100)
21:35:07 Return audiodub(last,c)
21:35:07 }
21:35:07 Normalize()
21:35:07
21:35:07 Soundout(output="cmd",type=0,format=0,autoclose=true,executable="C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\besweet\lame.exe",prefilename=" --abr 112 - audio.mp3",postfilename="",nofilename=true,showoutput=false)
21:35:07 -----------------------------------------------------------------------
21:35:07 Script VirtualDubMod.vcf
21:35:07
21:35:07 VirtualDub.Open("C:\\AutoMKVRips\\temp\\movie.avi",0,0);
21:35:07 VirtualDub.RemoveInputStreams();
21:35:07 VirtualDub.stream[0].SetSource("C:\\AutoMKVRips\\temp\\movie t80 2_0ch 192kbps delay 0ms.mp3",0x00000202,1);
21:35:07 VirtualDub.stream[0].DeleteComments(1);
21:35:07 VirtualDub.stream[0].AdjustChapters(1);
21:35:07 VirtualDub.stream[0].SetMode(0);
21:35:07 VirtualDub.stream[0].SetInterleave(1,500,1,0,0);
21:35:07 VirtualDub.stream[0].SetClipMode(1,1);
21:35:07 VirtualDub.stream[0].SetConversion(0,0,0,0,0);
21:35:07 VirtualDub.stream[0].SetVolume();
21:35:07 VirtualDub.stream[0].SetCompression();
21:35:07 VirtualDub.stream[0].EnableFilterGraph(0);
21:35:07 VirtualDub.stream[0].filters.Clear();
21:35:07 VirtualDub.video.DeleteComments(1);
21:35:07 VirtualDub.video.AddComment(0x00000005,"ISFT","Done with AutoMKV! 0.98.4 http://forum.doom9.org/showthread.php?t=140178");
21:35:07 VirtualDub.video.AdjustChapters(1);
21:35:07 VirtualDub.video.SetDepth(24,24);
21:35:07 VirtualDub.video.SetMode(0);
21:35:07 VirtualDub.video.SetFrameRate(0,1);
21:35:07 VirtualDub.video.SetIVTC(0,0,-1,0);
21:35:07 VirtualDub.video.SetCompression();
21:35:07 VirtualDub.video.filters.Clear();
21:35:07 VirtualDub.SaveAVI("C:\\AutoMKVRips\\test.avi");
21:35:07 VirtualDub.Close();
21:35:07
21:35:07 - - - - - - - - - - - - FINISHED JOBS - - - - - - - - - - - - - - - - -

[Tiberiust]

Given that there seems to be quite a bit of uncertainity about the time AutoMKV and its components take, and encoding time generally, I thought I’d do a shoot out and here are the results. I took the same 3.9GB ‘Vobmerge’d VOB file and just specified bit rate to 2000 and 2Pass Extreme Profile to an MKV x264 and then ran it on three different PCs. The resulting 1.2 GB MKV file was to my eyes 95pct as good as the original VOB. (IOW you really had to look hard to see any diff) Comments and log files (case I messed up somewhere) underneath.

PC1: Asus P5N-ESLI. NVidia chipset (my home PC) Core 2 Duo E4400 at 2.65GHz, 2GB OCZ DDR2 800 RAM, 2 x 400GB Striped. 1x8500GT. XP 32bit. SisSandra ALU 18,500 MFlops.
Start to finish time: 6 Hrs 8 mins

PC2: HPackard (off the shelf type, my office PC) Core 2 Quad 6600 at 2.4GHz. 2GB DDR2 800Ram, 1 x 200HD. XP 32bit. SisSandra ALU 29,600 MFlops.
Start to finish time: 3 Hrs 47 mins

PC3: WoW extreme gamer mate of mine….check out this PC:
Case: Thermaltake Armor+ VH6000BWS
PSU: Corsair HX1000W
Mobo: Asus P6T Deluxe X58
Cpu: Core i7 920 2.66Ghz @ 4.02 Ghz
AirCooling: Noctua
Ram: Corsair Dominator 3X2Gb 1866Mhz @ 1907Mhz
VGA: 2 X Asus EAH4870X2 CrossfireX
HDD: 4X500Gb Deskstar E7K1000 Raid-0,
EIZO: FlexScan 28''
SisSandra ALU 102,000MFlops (yep…not a typo)
Start to finish time: 1 Hr 42 mins

- PCs1 and 3 did the encode exclusively. PC2 was running MS Office 03 and Messengers concurrently but was not stressed.
- all PCs had identical file and component versions. (I copied all files from PC1 to the others). They loaded the same .ini file
- used latest x264.exe
- PC1 had average CPU usage on both cores ttl about 70pct
- PC2 abt 45 pct on all 4 cores
- PC3 abt 65 pct on all 8 cores

- I thought Core i7 would be faster, Sandra ALU certainly suggests that, and especially considering recent start at Nehalem specific work done on x264 which I read about. Maybe it’s sommat to do with 64bit operating system??? Dunno, beyond the sphere of my understanding but my gamer mate prob willing to try something else if anyone has any suggestions. If yu can get him off WoW for long enough that is. Would appreciate some feedback on this bcos myself and prob many other encoders seriously thinking about upgrading to Nehalem, but the above suggests prob not worth the cash at the moment since almost 2 hrs encoding makes it still an overnight job IMHO for batch work anyway.
- Out of int, PC3 has a 1000watt power supply and when going at full power in Azeroth (or wherever it is they play) it uses all of it and is very stressed. He uses air cooling only – 15 fans in total. ‘It’s quite noisy’ he says, and apparently the lights dim in his room when he’s playing…..Whatever dude!!….

T

[nurbs]

I have a few suspicions, but it's hard to say anything definite till your attachments are approved. Next time just post them in a code block, like cosmos did in the post above yours, so people don't have to wait. Generally synthetic benchmarks like SisSandra ALU are pretty useless if you want to know about real life performance.

[Tiberiust]

Nurbs,
Tks. I did that initially but it got rejected bcos it was too long (max 16k characters) and I thought it better to post it in one complete post rather than in 2 or 3 parts for ease of reference. I'm very interested to hear any response. No hurry - if it takes more than 24hrs for approval I'll post them in code blocks in 3 posts. I thought I'd post these results bcos reading thru this entire thread and in other places where I go abt encoding, it seems people have no idea just how long video encoding takes and they then spend ages trying to work out what's wrong with their setup. SiS ALU alone isn't much good I know, I just put it in as a guide but on PC3 rig as yu can imagine from the specs, EVERY benchmark is 'out there' as far as I am concerned. My mate reckoned he could fly thru anything - this process started with me proving to him that nothing flies thru video encoding. Would be great if someone reading could point out a few basic errors to speed me up.
T

[Chengbin]

buzzqw, how do I make MVDegrain 1 multithreaded? Is MVDegrain used only in first pass or both passes?

[nurbs]

@Tiberiust:
On video encoding the core i7 is about 2.35 times faster then PC2. That's a bit less then I would expect, the core i7 is generally up to 50% faster at video encoding at the same clock speed as the core 2, but since you are encoding SD the cores can't all be used that effectively. You are also using --b-adapt 2 and with frametype decision not being threaded yet that slows down first pass greatly, especially for the quadcore cpus because a lot of the time only one of the cores will be used which is more noticable with four cores than with two. It would be interesting to know how much of the total encoding time was used by the first pass. Also you used slightly different audio setting on PC2.

[buzzqw]

@cosmos

Use CRF for autoresize routines is UNCHECKED
as you can see in log
21:16:33 The Movie is declared to progressive
21:16:33 AutoResize 1' Pass XviD: "C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\encoder\xvid_encraw.exe" -pass1 -bitrate 968 -turbo -progress -max_bframes 2 -imin 3 -imax 16 -quality 6 -vhqmode 1 -pmax 16 -bmax 16 -bquant_offset 0 -kboost 20 -odegrade 9 -nopacked -nochromame -bvhq -threads 0 -lumimasking -qmatrix "C:\Users\Καραγκιόζο\Documents\AutoMKV\profiles\XviD-profiles\matrix\Sharktooth's EQM v3LR.cqm" -i "C:\AutoMKVRips\temp\resmovie.avs" -type 2 -avi "C:\AutoMKVRips\temp\resmovie.avi"
21:16:33 AutoResize 2' Pass XviD: "C:\Users\Καραγκιόζο\Documents\AutoMKV\exe\encoder\xvid_encraw.exe" -pass2 -bitrate 968 -progress -max_bframes 2 -imin 3 -imax 16 -quality 6 -vhqmode 1 -pmax 16 -bmax 16 -bquant_offset 0 -kboost 20 -odegrade 9 -nopacked -nochromame -bvhq -threads 0 -lumimasking -qmatrix "C:\Users\Καραγκιόζο\Documents\AutoMKV\profiles\XviD-profiles\matrix\Sharktooth's EQM v3LR.cqm" -i "C:\AutoMKVRips\temp\resmovie.avs" -type 2 -avi "C:\AutoMKVRips\temp\resmovie.avi"
21:17:09 Finished 1' Pass XviD
21:17:43 Finished Autosize procedure
automkv will do 2 pass autoresize (or 1 pass if you are doing a 1 pass encoding)

A x264 encoding pops up at some point during the process! It seems like the CRF tests always take place?
it's due to movie analysis and interlace analysis, it's run only for parsing the movie and obtain informations

2) Checking the log, it seems like that autoresizer suggests a movie width of "1200 based on sample size". Isn't this quite optimistic?
when disabling automatic resize maybe it's better to enlarge sample analyzed , 10% or 20%

@Chengbin
applyed on both pass
automkv don't actually support MT plugin

@Tiberiust AND nurbs
thanks for your test Tiberius, feel free to suggest better profiles :)
and thanks nurbs for your explanation. it's correct to the bone

BHH

[cosmos]

@buzzq: thanks for all the information.


when disabling automatic resize maybe it's better to enlarge sample analyzed , 10% or 20%
Will try it, thank you :)

[Chengbin]

@Chengbin
applyed on both pass
automkv don't actually support MT plugin

Why not? Will the MT plugin be supported in the future? MVDegrain is taking longer than me tesa and me range 32.

[Tiberiust]

@nurbs

Tks very much.

but since you are encoding SD the cores can't all be used that effectively
SD ?

It would be interesting to know how much of the total encoding time was used by the first pass

Is that not in the logs? Just subtract the times against the lines? Like so for PC1
00:37:32 Command Line 2' Pass X264: C:\Program Files\AutoMKV\exe\encoder\x264.exe --pass 1 ****
02:37:06 Finished 1' Pass X264

PC1: 2 Hrs 0 mins
PC2: 1 Hr 47 mins
PC3: 53 Mins

Not a huge difference there. It's mostly all in Pass 2:

PC1: 4 Hrs 3 mins
PC2: 1 Hr 54 mins
PC3: 48 mins

Can anyone point me to somewhere where I can see all the x264.exe command line options and an explanation of their effects? I can't seem to find anything concise on it.

Main point of my test was to see whether it was worth laying out the cash for a Core i7, or not. So unless I've really messed up by using a load of non-multi thread options which would reduce the relative speeds I don't think the difference really justifies it at the moment for the casual user. PC3 is running at 4GHz and the end result was 'only' 3.5 X faster. Not enuf IMO....
T

[nurbs]

SD = standard definition; low resolutions like 720*576 or 720*480 and below. Encoding with more threads generally works better at higher resolutions. If all you do is encode DVDs your core 2 quad is probably fast enough for you. I'm doing Blu-Ray to 720p at 2.5 fps. With that core i7 even at stock speed my encodes would be at least 6 times faster. If you are interested in doing a test with a HD source but don't have anything lying around you could try the Tech ARP benchmark (http://www.techarp.com/showarticle.aspx?artno=520). It uses an old version of x264 (not optimised yet for Core i7, Phenom,...) so the speeds you get will be a bit lower than with an up to date version, but it's still good at comparisons between different CPUs.

As I said frametype decision (which is done in first pass) isn't threaded, that's why the first and second pass take almost the same time at the quadcores. You can work around that by using --b-adapt 1 (faster b-frame decision method) at the expense of some quality. It looks like that will be threaded soon anyway so the problem will probably go away.

You can have a list of the x264 options by typing "x264 --longhelp", but that doesn't really explain them. I'm sure you can find explainations if you search the forum.

[noee]

I have found this "mapping" useful: FFMpeg -> x264 (http://ffmpeg.x264.googlepages.com/mapping)

[Tiberiust]

@nurbs

Tks - a great help. I'll find a Blu-Ray from someone and I'll run it on all 3 PCs. Then I'll update my original post which I now know is doing the Core i7 and the recent x264 optimisations a disservice. Have yu any suggestions for a few basic settings I should use to optimise the process for the 4 and 8 cores?

There's some debate around about 1080 vs 720 and whether the diff is really that visible. Since you're doing BluRay > 720, what do your eyes tell yu?

@noee
Tks - that's enuf for me to sink myself into for a few days

[nurbs]

720p vs. 1080p depends on what you want to do I guess. I do 720p mainly because I want small filesize while still maintaining good quality. I often travel by train and like to watch tv shows on the laptop then and I also have a 1080p HDTV with a WD TV media player. I just find it more confortable to browse the folders then looking through my DVD and Blu-ray collection. Personally I'm pleased with the quality even on the HDTV. The detail lost depends on the source. If it's very clean and sharp to begin with you really have to look hard do discern the difference. I don't do any preprocessing currently because my rig is too slow, but I guess with some grain stabilizing and a little sharpening the outcome would be even better. You should look around in the Avisynth section of the forum. The possibilities with the various plugins like LSFmod and MVdegrain are basically endless.
I don't do straight 720p but use the equivalent number of pixels. So for 16:9 movies it's actually 1280*720 and for 2.35:1 movies something like 1472*624.
It definitly looks way better then DVD while still being smaller in filesize, probably because the source is much better.

I use the following setting:
--crf 22.0 --keyint 96 --min-keyint 12 --ref 3 --mixed-refs --bframes 3 --b-adapt 2 --weightb --subme 9 --trellis 1 --psy-rd 1.0:0.2 --partitions all --8x8dct --vbv-bufsize 25000 --vbv-maxrate 25000 --me umh --threads auto --thread-input --progress --no-psnr --no-ssim --output "output" "input" --level 4 --colormatrix bt709
The short keyint is for faster seeking and doesn't cost much quality. I would definitly recommend --subme 9, because according to the x264 devs it helps a lot if you use --psy-rd and the speed loss isn't bad compared to say --trellis 1 vs. --trellis 2. If you use --subme 9 and --trellis 1 one the gain from the first may outweigh the loss from the latter while you still come out with a net speed gain. I actually tested the speed difference, but I can't remember. I reduced psy trellis (the second value in the --psy-rd command) to 0.2 because a) high values produce too much ringing in my opinion and b) combined with trellis 1 and the short keyframe interval high values tend to trigger frame flashes (not sure if that's the right word; you have one frame where quality is noticably worse then all the preceding and following frames). b) hardly happens with high values when you use trellis 2 and I haven't noticed it with trellis 1 and 2pass encodes either. It's probably best if you test it yourself with a scene where there is a long and detailed close up of a face. My encodes tend to come out between 3 and 6 Mbps. The size depends much more on grain and noise in the source then on detail.

I made some pictures from my encodes for you. Don't have the source files on my HD for comparison, sorry. The bitrates of those encodes are around 3 to 4 Mbps. Sources are very clean and detailed (Casino Royale and Terminater not as detailed as the others).
http://www.megaupload.com/?d=NCYWSMGD

edit:
I forgot to get better speed you can try to play around with the number of threads a bit. --threads auto uses number of cores *1.5 but that's more a rule of thumb. You could maybe try 2 times the number of cores. Some people on the forum have been able to get some noticeable speed gains. Raising the number of threads reduces quality a little, but I wouldn't worry about it for HD content. For SD you shouldn't go higher. As I meantioned a threaded frametype decision patch is in the works and will probably be ready soon (guesstimate couple of weeks to 2 months). When that is included in x264 you will see great speed gains on the first pass and maybe also with crf encodes on multicore systems.

[delacroixp]

- PC1 had average CPU usage on both cores ttl about 70pct
- PC2 abt 45 pct on all 4 cores
- PC3 abt 65 pct on all 8 cores
As nurbs points out ... SisSandra can be pretty useless when it comes to a real-life workload.
That said, your very-low % usage is more than a little disturbing ... regardless of your final speed.
1000 x 540 is more than enough resolution to keep X264 pumping at full throttle.
I'm busy encoding the World at War (http://en.wikipedia.org/wiki/The_World_at_War) series at 560 x 432 on a 2.4 GHz dual-core and my encodes never drop below 95%.

I have found that certain filters will affect your encode efficiency adversely.
I tried MipSmooth recently and was surprized to see the CPU running at 70%.
RemoveGrain() (http://automkv.a.wiki-site.com/index.php/Installation#RemoveGrain), however, supports SSE2 and runs beautifully on my system.

Unfortunately, the downside is that I avoid inefficient filters, eventhough I may be running at framerates that would be the envy of most encoders.


It's all good
:):devil::D
Pascal

[nurbs]

He didn't say when and how he measured the cpu usage. If it was during first pass numbers for PC2 and PC3 are actually higher then I would expect.
Filters being slow is not the same as them being inefficient. Since Avisynth isn't treaded any filter that runs slower on one core than you can encode the frames will bring down CPU usage. Removegrain works well to remove light grain, but with the stronger modes it kills too much detail compared with other often slower filters, which is the price you pay for using a simple grain removal method.

[Gamezcdr]

I have the same problem: AutoMKV use only 40-45% of 4 cores Phenom II in pass1 even I set Realtime for x264.exe in Windows Task Manager, in pass 2 use about 70-80% enconding DVD to H264. Why not 100% ? :)

[Tiberiust]

@nurbs
Tks again, I'll go thru your post in details later when I've a bit more time (supposed to be working atm!) Those pictures look superb though - I assume yu end up with a smooth picture as well, no jerks at all?. Approx what kind of filesizes do yu get for those quasi 720p codes?

@delacroixp
Tks. I 'measured' the CPU usage by opening XP Task Manager and observing an average over 20 mins or so. Yes, it was first pass only. Lemme know if there's a better way of doing it.

[delacroixp]

He didn't say when and how he measured the cpu usage.

Filters being slow is not the same as them being inefficient. Since Avisynth isn't treaded any filter that runs slower on one core than you can encode the frames will bring down CPU usage.

Removegrain works well to remove light grain, but with the stronger modes it kills too much detail compared with other often slower filters, which is the price you pay for using a simple grain removal method.
You're quite right ...
Avisynth isn't multi-threaded, RemoveGrain() isn't the sharpest tool in the shed and there is a difference between speed and efficiency.

It's all relative ...
I guess encoding 1080p on an oct-core will be slower than 480p divX on a single core.
I like to maximise my CPU usage even if I don't have a $million supercomputer.

There are a lot of very good filters out there ...
Using the right tool for the right job probably separates the men from the sheep.


It's all good
:):devil::D
Pascal

[delacroixp]

@delacroixp
Tks. I 'measured' the CPU usage by opening XP Task Manager and observing an average over 20 mins or so. Yes, it was first pass only. Lemme know if there's a better way of doing it.
I'm sure that your results pretty much reflect the reality.

That said, using the Constant_Quality profile will radically reduce encode times compared to 2-pass.
Even the Constant_Quality_CQ_Optimised becomes a viable alternative for "quality junkies".


It's all good
:):devil::D
Pascal


btw
1000 x 540 is wastefully high.
anamorphic (720 x 576) => 414,720 pixels <= (864 x 480) square pixels

[nurbs]

The movies are between 3 and 4 GB, the TV series is about 1 GB all including at least one 5.1 audio track (aac). Everything is smooth. I have had movies come out with up to 6 GB and TV series episodes with up to 2.

[mrmrmrmr]

I did that already. I wrote it in a post a few messages ago.
Damn, I don't know why.
Another question: Should the first or second pass be faster (on balanced for example).

@ayfa:
I have a similar problem.
I installed a new quad core cpu (Q9400) and did not observe too much speed improvement compared to my previous Pentium D.

I am converting 1080p MKV file to 720p or 1080p wmv file.

How much time should it take for such conversion of 2 hours movie at 1 pass fast encoding ?

[ettin]

Ive been trying to encode a dvd rip that i merged using vobmerge, but when i try to encode it with automkv it starts encoding in 4000 frame chunks and does this over and over again, and ive even left it on for 12 hours for a 2 hour movie, and still had it not finish. I check the temp folder after each 4000frame pass and it starts at 160mb, then builds up to 170mb by the end of the one, and then goes back to 160mb afterwards.

Not sure what else to do, i did this on a small vob file and it will do one small chunk, and then a big chunk which is the rest of the vob, and worked fine, but it wont do it on the merged vob file.

Anyways, if you cant think of a solution off the top of your head, i could really use a recommendation for a better x264 encoder.

[Chumbo]

@ayfa:
I have a similar problem.
I installed a new quad core cpu (Q9400) and did not observe too much speed improvement compared to my previous Pentium D.

I am converting 1080p MKV file to 720p or 1080p wmv file.

How much time should it take for such conversion of 2 hours movie at 1 pass fast encoding ?
I just wanted to add similar results with my Q9400 and i7 920. The results compared to my old P4 660 were pathetic to say the least.

Can't wait to see this thing support multi-cores.