QuEnc 0.72

[tamahome]

get rid of the -mpeg2 switch, probably.

i've had no problem on the 20 odd computers i've run it on :)

No, commandline is correct (quenc would not start with encoder)
i use -mpeg2MUX
Little video of problem
http://dl.free.fr/GuFJqwmT/problem.avi

commandline use -b 1683 -maxbitrate 8000 -dc 8 -priority 2 -2 -nohq -vbr -scene -notrell -nocgop -nointerlaced -noextreme -gopsize 15 -maxbframes 2 -qlb -aspectratio 4:3 -mpeg2MUX -scene -auto -close

[tamahome]

it would seem that quenc bug with the sources h.264/avc, the problem would come from,
because with the divx/xvid source or dvd he goes well.

as an information I have haali splitter and coreavc to read the mkv/avc (I have to also test with ffdshow same result quenc plant)

[Mr_Odwin]

Is this more of an Avisynth problem than a QuEnc problem then?

[tamahome]

I do not think because the same file with hcencoder /CCE passes perfectly

[Mug Funky]

my bet is on directshow as the problem, but it is strange that it's working in CCE and HC, and older versions.

have you played with setmemorymax, or setlegacyplanaralignment in avisynth? i'm clutching at straws here :)

[dragongodz]

-mpeg2MUX -scene
mpeg2mux actually requires a value after it, ie. vob. without it it will just use default(vob) anyway.

it would seem that quenc bug with the sources h.264/avc, the problem would come from,
because with the divx/xvid source or dvd he goes well.
can you provide a smallish sample that reproduces the problem so i can test against ?

also what is the exact avisynth script you use ?

[shirohamada]

is it just me, or quenc don't use any quant 3 and 5, or odd numbers?

[hank315]

is it just me, or quenc don't use any quant 3 and 5, or odd numbers?No it doesn't, QuEnc uses a linear quantization scale, values 2,4,6,8,10,12 .... 62.

[dragongodz]

is it just me, or quenc don't use any quant 3 and 5, or odd numbers?
depends. what are you looking at the stream with ?

for example if you look with elecard streameye you can see quant 3 etc frames. however if what you are using just shows quantifier scale then no you wont see quant 3 etc.


tamahome - still waiting on something i can test with. please PM me with more information if nothing else.

[shirohamada]

too lazy to look for other tools. so i just feed it through ffdshow and turn on osd -_-

[MrC]

@Nic,DGZ

Any chance to see an update to your wonderful creature? Should we consider the project stopped?

:confused::)

Bye

[dragongodz]

Any chance to see an update to your wonderful creature?

yes. i had actually planned to have a new version out long before now. unfortunatly real life kept getting in the way. for example in the last week 2 major things have happened, neither good, which means finalising things i was testing in QuEnc is not even in my mind at the moment.

[tom942]

dragongodz, any advice to avoid blocks in fast action or motion scenes or some of the improvements that you are working on goes this way?.

BTW, thank you for your nice app and hope that your problems go away soon :).

Regards.

[dragongodz]

any advice to avoid blocks in fast action or motion scenes or some of the improvements that you are working on goes this way?.

using a lower bitrate matrix may help reduce them.

and hope that your problems go away soon

one is nearly sorted out and the other(the death of my mother) will take a little time(funeral arrangements etc).

[Boulder]

one is nearly sorted out and the other(the death of my mother) will take a little time(funeral arrangements etc).I'm sorry to hear this :( My condolences to you and your family.

[MrC]

Condolence to you from me too.

Bye

[tom942]

As Boulder said, I'm so sorry and my condolences to your family and you :(.

[dragongodz]

thanks guys.

[tom942]

@dragongodz

I have used lower matrices, such as yours QLB or AVAMAT6 & 7, but the blocks are still there.

The truth is, that I can see them just when I play the video frame by frame, not in normal playback.

That's the reason that I asked you if you were tweaking something related with that, but I hope that you "play" with it when you feel better :).

Regards.

[Ebobtron]

Doing some quick last minute encodes and I suddenly noticed that QuEnc 72 doesn't like avisynth to feed it 24 or 32 bit sound. Result is a rail to rail hiss. 16 bit works just fine. Oddly though I don’t ever remember feeding it anything other than 16 bit sound.

Just needed an excuse to say merry, merry and happy, happy and sorry about your mom.

[dragongodz]

ok for those interested and feeling a little adventurous i present a little xmas and new years present .... QuEnc 0.75 alpha.

http://w14.easy-share.com/13664921.html

WARNING!!!!! its alpha for a reason. it is not intended for permanant use or to be added to other software collections etc. it is purely for those interested in doing a little testing etc.

changes to test

trellis does not enable trellis, instead enable 2 threads. disabling trellis turns back to 1 thread. this is done this way so you can test it using other programs that have the trellis option when calling QuEnc.

slight possible quality change to normal and hiq modes. hiq you may not see much or even anything. hopefully normal should gain a little though.
if you compare against QuEnc 0.72 with psnr and/or ssim then please compare with your eyes aswell. also please compare the speed, with trellis disabled(so 1 thread) for 0.75alpha.

these are the only 2 major things i left in this release for people to play with.

Ebobtron - i will look at that later. thanks.

[Fishman0919]

Thank You dragongodz

[MrC]

Thanks a lot DGZ, I will try it ASAP and I'm looking forward for a stable release to add to AVStoDVD.

A (maybe) stupid question: why 0.75 and not 0.73?

:) :cool:

Bye

[dragongodz]

A (maybe) stupid question: why 0.75 and not 0.73?

no not a silly question.

basically there have been other alpha/test versions. they were only given to a few people however to test specific changes. so they were numbered differently for my own record of them.

so if i had not released this here the next stable would be either 0.73 if only a few changes were included or possibly 0.80 if it contained many, many changes or something i felt was major enough.

however i felt being the time of year it is it may make a nice little present and also get a bit broader feedback. :)

[Fishman0919]

Quality seems about the same as .72 with HQ on.... maybe a bit better in high motion scenes. Norm is better for me on same scene.

Speed is slower then .72... same movie .72 avg about 38.5 fps where .75a avg 31 fps


Edit: I just did a 1 min clip from Transformers HD down converting it to SD DVD. It's the opening scene at the military airport... dark and high motion.
The clip is at 3600k with HQ on. To me .75a looks better then .72

[manolito]

Thanks DGZ for the new build,

just did a short torture test in 1-pass VBR mode, normal quality, AVG bitrate 8500, Max bitrate 9000. I am mainly interested in getting DVD compliant output (if Muxman has no complaints then I assume it is compliant).

With V. 0.72 Muxman reported a max bitrate of 10888, with 0.75 it was 11434. Still Muxman did mux and author it without errors, and of course it did play in my SA without any stutter.

Happy New Year!

Cheers
manolito

[dragongodz]

.72 avg about 38.5 fps where .75a avg 31 fps


is that normal or hq ? yes i knew there would be a speed hit. quality increases usually come at some sort of cost. hopefully threading should help when not using a program that does multiple encoding processes(meaning splitting the encode and running several instances of QuEnc at once).

1-pass VBR mode

having a quick look there is still a couple of changes that i didnt revert back for this so they may be the cause of the max bitrate increase and not the other changes.

[Fishman0919]

is that normal or hq ?

HQ mode... the speed loss is not bad for the quality gained.
A few more scene I did with a lot of high motion look better then .72... little... but better.

The couple of clips I did with norm mode look quite better then .72 norm and maybe close to .72 HQ mode

I'm away from home now and using a single core AMD Sempron 3400+ laptop so "trellis" 2 threads is not an option right now.

[smok3]

Thanks DGZ for the new build,

just did a short torture test in 1-pass VBR mode, normal quality, AVG bitrate 8500, Max bitrate 9000. I am mainly interested in getting DVD compliant output (if Muxman has no complaints then I assume it is compliant).

With V. 0.72 Muxman reported a max bitrate of 10888, with 0.75 it was 11434. Still Muxman did mux and author it without errors, and of course it did play in my SA without any stutter.

Happy New Year!

Cheers
manolito
iam having a similar problem with 1-pass vbr, adobe encore refuses to build since there is error (max bitrate overflow).

[dragongodz]

I'm away from home now and using a single core AMD Sempron 3400+ laptop so "trellis" 2 threads is not an option right now.

thats cool. the other feedback/tests is helpful.

similar problem with 1-pass vbr

now repeat after me. ALPHA, AAAAAALPHA. ;)
i did mention why 1 pass may be like that. i may do another alpha with that fixed and maybe a different quality increase(too see get an opinion by others which would be more desirable) on the weekend.

[smok3]

actually i was referring to 0.72.

[dragongodz]

actually i was referring to 0.72.

ahh you should have been a bit more concise in what you meant. :)

hmm ok 17 downloads and only feedback from a few people. why doesnt that surprise me anymore ? :sly:

[Fishman0919]

May a new thread with a "QuEnc .75 Alpha" title... I just happened to see you had posted something in this thread and wanted see what you had wrote. Some people might not be so concerned with .72 and without a new thread title, may not look in.

[dragongodz]

well i never expected or wanted mass downloads etc. however i did hope, against expectations, that of those who did spot it and downloaded then hopefully at least half would make some comment.

[jeffy]

well i never expected or wanted mass downloads etc. however i did hope, against expectations, that of those who did spot it and downloaded then hopefully at least half would make some comment.
Happy New Year and all the best! It might be good to add the feedback wanted note to the post with the latest link, eg. please post your comments, wishes etc. here.
It might be disappointing for the programmer, but rest assured, some people really appreciate the effort, although they remain quiet.

[smok3]

ahh you should have been a bit more concise in what you meant. :)

with this setting iam getting non-compatible dvd output (0.72):
[HKEY_LOCAL_MACHINE\SOFTWARE\QuEnc]
"crBkgColor"=dword:00aaaaaa
"crLtBkgColor"=dword:00bbbbbb
"crVLtBkgColor"=dword:00cccccc
"m_crForeColor"=dword:00000000
"m_nGopSize"=dword:0000000c
"m_nBFrames"=dword:00000002
"m_nBitrate"=dword:0000125c
"m_nCodec"=dword:00000001
"m_bUseHQ"=dword:00000000
"m_bUse2Pass"=dword:00000000
"m_bUseTrellis"=dword:00000000
"m_bUseVBR"=dword:00000001
"m_bUseClosedGOP"=dword:00000000
"m_bQLB"=dword:00000000
"m_bExtremeSettings"=dword:00000000
"m_bInterlaceEnc"=dword:00000000
"m_bSceneDetect"=dword:00000000
"m_bAutoMaxBitrate"=dword:00000000
"m_nMaxBitrate"=dword:00001d4c
"m_nDCPrecision"=dword:00000000
"m_nThreadPriority"=dword:00000020
"m_nAudioBitrate"=dword:00000007
"m_nAudioCodec"=dword:00000000
"AspectRatio"=dword:00000001
"m_bTopFieldFirst"=dword:00000000
"m_bPulldown"=dword:00000000
"m_csCustomMatrixFile"=""
"m_csMuxProfile"="DVD"
"m_nMPEGSel"=dword:00000001
"m_bKVCDNotch"=dword:00000000
that's what i meant :)

p.s. do 'we' really need to use registry for storing setting btw?

[dragongodz]

It might be good to add the feedback wanted note to the post

however the simple fact it is an alpha and has a big warning plus mentions whats worth testing should be a pretty clear sign to only download if a person actually plans to test it. if thats a given then it shouldnt be unreasonable to expect/hope that some will then report their finding from their testing.
as i said i have been around too long to expect everyone that downloads it to report anything but it is a little dissapointing that Fishman0919 seems to be the only one willing to go that extra step to test and report on the things mentioned ... and he is not even at home. :eek:

EDIT: no doubt some have been busy and may report later aswell of course.

that's what i meant

out of curiosity did you try anything else besides adobe encore(muxman, tmpgenc dvd author etc) ?

do 'we' really need to use registry for storing setting btw?

is there a good reason not to ?

[Fishman0919]

I've done 14 diff clips... ranging from 1 to 5 min's
With every clip .75a looked better then .72

Here is the order I would rank them...

4) QuEnc .72 Norm
3) QuEnc .75a Norm (very close to .72 HQ)
2) QuEnc .72 HQ
1) QuEnc .75a HQ

QuEnc .72 was faster in all clips in both mode but like I said before... the loss is not bad for the quality gained

Sorry I don't have encoding times but busy with real world work stuff and limited with my resources here

Edit: Side note... not sure if it's something you changed or maybe just the movies I encoded but .75a seems to size better then .72

[smok3]

out of curiosity did you try anything else besides adobe encore(muxman, tmpgenc dvd author etc) ?
is there a good reason to ?

[Fishman0919]

I've had issues with the muxing process of Adobe Encore...where Muxman didn't fail or get a problem.

Just my 2 cents.

[Fishman0919]

@dragongodz,

I just finished encoder an HD-DVD clip to SD-DVD and got an debugging error at the end of the 2nd pass that pop-up Microsoft Visual Studio 2008.

I closed it by mistake before I could grab a screenshot. I will run it again later

[Tdr]

@dragongodz and all,

is there a special Thread, Forum or Helpfile, where a Newbie to Quenc can post some - possible - stupid Questions?

@dragongodz

I`ve tested - as far as I can see - all freeware encoders with libavcodec and I ended allways by your Quenc. Audio inside!, great Work!!!

Is there a list, where I can add some whishes for further versions? (just kidding)

Thanks
Tom

Ps: last used my english about ten years ago, so please ...

[dragongodz]

got an debugging error at the end of the 2nd pass that pop-up Microsoft Visual Studio 2008.

did the encode actually work or bad file at the end ?

i did install the vs 2008 trial and played with it for a couple of day. wasnt impressed so i removed it except the .net crap.

.net compact framework 2.0 sp2 - 97mb
.net compact framework 3.5 - 86mb
.net framework 2.0 sp1 - 195mb
.net framework 3.0 sp1 - 250mb
.net framework 3.5 - 29mb

man i could save some room removing all that crap. :)

hmm i see i still have crystal reports 2008 installed (188mb). bloody loits of crap vs 2008 installs and you have to uninstall it all separately. hows that for a pain ?

so avcodec was compiled with mingw/gcc and QuEnc compiled with vs 6 with intel compiler. vs 2008 should have had nothing to do with it.

[dragongodz]

is there a special Thread, Forum or Helpfile, where a Newbie to Quenc can post some - possible - stupid Questions?

right here. :)

if i dont answer others may.

Ps: last used my english about ten years ago, so please ...

no worries. if i dont understand what you mean i will say so and if i misunderstand then i will just ask you to be more precise in what you mean.

[Fishman0919]

did the encode actually work or bad file at the end ?

so avcodec was compiled with mingw/gcc and QuEnc compiled with vs 6 with intel compiler. vs 2008 should have had nothing to do with it.

Seemed to work fine.
On my laptop vs 2008 pops-up for debugging when a program errors. It made a QuEnc.slh and QuEnc.suo file for debugging. But since I don't have the source... the files are better much empty

[ricardo.santos]

Source:
Video: 5 minutes VOB = 6200 bitrate
Audio: 6 ChannelAC3 = 448k

Avisinth script: (Thanks MrC)
Video = DirectShowSource("C:\TESTE\VIDEO_TS\VTS_01_1.VOB", convertfps=true, audio=false)
Audio = GetChannel(DirectShowSource("C:\TESTE\VIDEO_TS\VTS_01_1.VOB", video=false), 1, 3, 2, 5, 6, 4).Normalize()
AudioDub(Video, Audio)
ConvertToYV12

====================================================

TEST SETTINGS:
VIDEO BITRATE: 3500
AUDIO BITRATE: 192 AC3

Quenc 0.72 (2 pass, variable bitrate)
Conversion time= 7.50
Final size= 138 megs

Quenc 0.72 (2 pass, High Quality mode, variable bitrate)
Conversion time= 11.15
Final size= 138 megs

Quenc 0.75 (2 pass, variable bitrate)
Conversion time= 12.20
Final size= 138 megs

Quenc 0.75 (2 pass, High Quality mode, variable bitrate)
Conversion time= 16
Final size= 138 megs

I couldnt see any difference between the four videos, not even when watching the videos really close to tv. ill try a lower bitrate later today, any ideas for a bitrate value i should try?

[Tdr]

@dragongodz,

Thanks for the offer, for first I have 2 Questions:

1.
I have an XVID-Action-Movie and have cut out a sequence of 5 Minutes with VirtualDubMod (Direct-Stream Copy) for testing; I searched for an extrem scene with very fast Cars.

I want to understand the function bitrate (kpbs) and max bitrate (advanced Options), how they effect each another.

In the setting you can find :"QuEnc does not limit it to that maximum though so you are better off setting
it manually for things like DVD or SVCD etc.".

So, for example, I used 2600 in max bitrate (advanced Options), to produce a SVCD.

When the Bitrate (not under Advanced Options) is then set for example to 2500, is this 2500 the max or the average (I can get for example out of Fitcd, I use it for resizing to 480 x 576) bitrate?

When its average, then a setting 2600 in Bitrate and 2600 in the max bitrate would be stupid .

2.
I have tested serveral encoded Films with various Bitrates (1 pass, VBR) and my short clip.

When I use Bitrate Viewer, there is a problem, that I cant understand.

When I encode a complete Film, the Bitrate, used under Advanced Options, is as max Bitrate ok. There are some small peaks (4-5 in 90 minutes) over, but just for 200-300 kpbs. Therefore I can live with.

When I encode my short Action Clip, Bitrate Viewer shows the first 4-5 seconds extrem peaks, over 6000 kpbs, although max bitrate is set for example (for SVCD) to 2600. After 10 seconds the peaks are ok, same as complete film.

On my Standalone I recognised this problem several times, the picture jumps. Its allways the same, when a movie starts very fast with action scenes.

Is this a problem, used by my settings, a bug, or is it the general problem, that libavcodec should have with the ratecontroll.

Or is it a Problem of 1-pass?

Hope somebody can understand the questions.

Thanks a lot
Tom

PS: @dragongodz, would it be much work, to code a shutdown in the Gui? I often encode overnigth.

[Fishman0919]

I had stated before that QuEnc .75a seems to be sizing better .72 and the encodings I was doing were between 2500k and 3000k.

I just did a few higher bitrates... 4000k-5000k and .75a seems to be running into encoder saturation more so then .72

I ran these test but I tried encoding the Simpsons R1 NTSC with DVD-RB using CCE Smooth Matrix and QuEnc .75a. It undersized a bit 4.02 GB (4,323,149,824 bytes)

I'm running the same movie again thru DVD-RB with .72 now...

Edit: QuEnc .72 = 4.18 GB (4,499,406,848 bytes)

[dragongodz]

any ideas for a bitrate value i should try?

depends on the source. the more complex(action etc) the more bitrate is usually required.

is then set for example to 2500, is this 2500 the max or the average

the bitrate in the main window is target/average bitrate. the one in advanced is maximum bitrate.

Or is it a Problem of 1-pass?

i would always reccomend 2 pass for tighter constraints such as svcd or for lower bitrate encodes. 1 pass encoding is generally fine for higher bitrate encodes but even then 2 pass will probably look better 99.999% of the time.

to code a shutdown in the Gui?

i may add that later. depends on time and if i think about it. :)

the same movie again thru DVD-RB with .72 now

how did they look in comparison ? was 0.72 better/the same/worse with the higher bitrate ?

[Fishman0919]

how did they look in comparison ? was 0.72 better/the same/worse with the higher bitrate ?

Very hard to tell... being that it's Anime/CGI... I may want to say .75a is better but after looking at several scenes several times... I'd have to say the same. Being that it's smaller (only a bit smaller) and looks the same is not bad :)