quenc & speed

[ncage]

Hi guys. I messed with CCE & DVD-Rebuilder toooo long. I wanted it to work because i know that is the defacto standard :). The guys over at afterdawn told me not to even mess with it because trying to encode files with the trial version of CCE because it is very error prone and i would end up almost pulling my hair out which i almost did :). They said if i wanted to use CCE go buy cce basic and don't even mess with the trial. I don't know about you guys but i don't like buying anything before i know how it works. I decided to try QuEnc. I can tell you what...i put all the high settings on and the copy is beautiful. It seems like a VERY good quality encoder. I was impressed to say the least There is only one downfall to it....its VERY VERY VERY VERY VERY SLOW. Did i say it was slow yet :). Its about 3-4 times slower than cce on my machine. I have an AMD 64 3000+ overclocked to 2200mhz which for you guys that don't know puts it a 3200+ level. I have 1gig of RAM. I also have a very very fast 160GB Hitachi 7200RPM 8meg cache SATA drive. I always restart and don't have anything running before i start dvd-rebuilder. Do you guys thing the cost of CCE Basic is really worth it? How much better quality will i really get. BTW, i only have a toshiba 32 in tv....no HDTV or big screen here. I really love quenc quality but hate its speed. I seen a server farm you can create for quenc...i was thinking on maybe trying that at some point since i have two amd 64 systems. My secondary system is an amd64 2800+ with 512meg ram.
e
ncage

[pg55555]

There is not any problems with the CCE trial. They work very well. The only issue is that they put a small logo in your encoded video.

But for testing quality of the output, just go for it.

The only issue with CCE Basic is that it only does two passes (one analysis (.vaf file) and one encoding.

But a lot of people thinks more passes in CCE are a waste of time.

[squinty]

I too had problems with getting the trials to work correctly. I was either getting video errors or an oversized dvd. :confused: So, one night when I was drunk, I decided to just go ahead and buy cce basic. I have to say, I am very happy with the quality of the movies so far. Most of what I do is Anime titles with moderate to low bitrates. The quality is simply amazing on these. On regular movies its also the best I've seen. I was using Clonedvd before this and cce basic (even in 2 pass mode) blows it away.

I have an AMD 64 3000+ with 1GB of ram and sata harddrives. The total encode (2 pass) takes roughly 2.5 hours of a 6.5GB DVD.

EDIT: oops, you do have an AMD 64, lol

[ncage]

Originally posted by pg55555
There is not any problems with the CCE trial. They work very well. The only issue is that they put a small logo in your encoded video.

But for testing quality of the output, just go for it.

The only issue with CCE Basic is that it only does two passes (one analysis (.vaf file) and one encoding.

But a lot of people thinks more passes in CCE are a waste of time.

There are many many people who have trouble with CCE's trial. I am not the only one. Just browse the forums. There are a ton of people that report there encodes are huge after encoding with the trial or some other problem. I am not a stupid computer user. I am a programmer by trade and have been working with computers since i was 10. But the question wasn't.....is there any problems with CCE trial....it was more in line with quenc speed & if you would see a difference in quality with CCE Basic.

[pg55555]

There are many many people who have trouble with CCE's trial. I am not the only one. Just browse the forums
I've been following this Rebuilder forum since I started using RB around 7 months ago, and actually I have not noticed such "many people" having trouble with CCE's trials.

As I'm not a programmer I'm not dare to use CCE without RB, so I've not visiting the CCE forums,where probably your troubles were posted

The problems I saw with RB were almost always due to a bad setting:
- Bad paths
- Not using EclCCE, or not setting it properly
- Using a CCE version not supported by RB at the time
- Incorrect Avisynth installation

And I would say almost all these problems would be the same with the full CCE version (except for the EclCCE issues, as the full versions above 2.50 do not need it)

But going to your question, I just want to encourage you to try the trial. It will do No harm.

And regarding speed and quality: Try it in your computer and watch in your TV: With your setup to encode a 7 GB original should be around 2 hours with CCE Basic

[jdobbs]

@ncage

I'd recommend you try it without all the high settings. From my experience they don't make that much difference anyway -- but really, really slow you down. If you keep those set to the defaults you'll find QuEnc is only slightly slower than CCE.

It's not a bad way to go either -- as Nic really gives it a lot of attention and it is based on a library that is continually improving.

I'd recommend you use QuEnc v0.57 -- as there have been some problems reported with DVD-RB and v0.59 alpha.

[manolito]

@ncage

And if you want it even faster, try FreeEnc 0.31 using the "Incredible_KDVD" template. (The default template has a tendency to undersize.) I would also suggest to edit the Incredible_KDVD template and replace the included Notch KVCD matrix with Dragongodz' QLB matrix.

QLB matrix formatted for FreeEnc:
-----------------------------------
intra_matrix=8,16,19,22,26,27,29,34,16,16,22,24,27,29,34,37,19,22,26,27,29,34,37,39,22,22,26,27,29,34,38,42,22,26,27,29,32,36,40,50,26,27,29,32,36,40,50,61,26,27,29,35,40,50,59,73,27,29,35,40,50,59,73,87
inter_matrix=16,17,18,19,20,21,22,23,17,18,19,20,21,22,23,25,18,19,20,21,22,23,24,26,19,20,21,22,23,24,26,28,20,21,22,23,25,26,28,29,21,22,23,24,26,28,29,31,22,23,24,26,28,29,31,32,23,24,25,28,29,31,32,34


With this setup FreeEnc is even a little faster than CCE 2.67 on my machine, and the quality is at least as good as with QuEnc - sometimes better.


Cheers
manolito

[jdobbs]

Does it work with DVD-RB?

[DK]

Does it work with DVD-RB?


renaming freenc to quenc seems to do the job

i'm running a test with freenc right this very minute and i'm looking forward to seeing the result

[jdobbs]

Let me know how it comes out. Does it support the QuEnc command line options? Some of them are very necessary.

[ADDED] Hmm... I just did a little reading on it. It does support the QuEnc command line switches -- and also has an INI file for tailoring other libavcodec parameters. Interesting.

All you would have to do to make it work is point the QuEnc path in DVD-RB's setup screen to FreeEnc.exe to give it a test.

[dragongodz]

a word of warning if using Freeenc with DVD-RB, dont blame DVD-RB if you get some funny results.

note that some of Freeencs commandline options are just dummies to copy QuEncs for compatability. eg. dc precision, qlb and trellis settings do nothing with freeenc from command line instead the settings in the INI are used, the dummy ones were listed on the kvcd forum before it went closed since that is where Freeenc started but i couldnt see it listed on its own site now. maybe someone would like to ask Vmesquita what ones are actually used and what ones are just dummies. also i see in 0.31 it says top_field_first added to INI settings which means to change to BFF you have to change the INI etc.

a comment about trellis and extreme&slow settings. i really wouldnt bother with trellis personally, its far to slow for any small quality increase. the extreme&slow however is good for low bitrate encoding but not really needed otherwise so you have to decide for yourself if you need to use it or not. :)

[jdobbs]

I just tried it. It also doesn't appear to be as fast as QuEnc. I'll just stick with what works.

[DK]

hmmm

my resultat was this:

freenc produced a final output of roughly HALF the size of what was desired

i havent looked into the material yet but i wouldnt expect anything decent

[manolito]

Originally posted by dragongodz
note that some of Freeencs commandline options are just dummies to copy QuEncs for compatability. eg. dc precision, qlb and trellis settings do nothing with freeenc from command line instead the settings in the INI are used, the dummy ones were listed on the kvcd forum before it went closed since that is where Freeenc started but i couldnt see it listed on its own site now. maybe someone would like to ask Vmesquita what ones are actually used and what ones are just dummies. also i see in 0.31 it says top_field_first added to INI settings which means to change to BFF you have to change the INI etc.

No need to ask Vmesquita about the command line parameters, a simple "freeenc.exe -h" will do.

IMHO it is a very good idea to use dummy parameters to ensure QuEnc compatibility and let the user edit a template file to change settings. QuEnc is still in Alpha state, and the QuEnc CLI has changed quite a bit recently, so any appication that supports QuEnc has to catch up constantly. If you want to change GOP length or DC precision, you cannot do this from DVD R-B, you have to execute QuEnc and make those changes there. So why not agree on a minimal set of command line parameters and make all other changes in the encoder?

@DK
If you get an undersized encode with FreeEnc, avoid using the default template by all means. Delete the file "Default.ini". In the "Incredible_KDVD" template the value for the vbv buffer has been tweaked, and this seems to get rid of the buffer underruns.

For the speed differences between QuEnc and FreeEnc, it is very hard to do a fair comparison because it is simply impossible to use the same parameters. GOP length, Trellis, matrix and number of passes can be matched, but FreeEnc does not have parameters for High Quality, Scene Detection or DC precision. QuEnc 0.59 seems a little faster than FreeEnc when HQ is off, but with HQ on it is considerably slower.

As for quality, I mainly do very low bitrate SVCD encodes, and FreeEnc looks a little better than QuEnc on dark scenes (less blocking).

Other issues: Both encoders have the libavcodec rate control issues, but the B-frame thing that QuEnc suffers from is definitely not present in FreeEnc.


Cheers
manolito

[DK]

i tried with the incredible.ini before using your modified parameters but i found freenc using way to low bitrates while encoding

so i stopped this testrun after some 30% or so and used the default.ini instead which too gave me an undersized result


+++++++++++++++++++++++++++++++++++++++++

2.nd run

this time with the standard incredible.ini WITHOUT the parameters you posted and now the bitrate looks fine

i'll report back

[dragongodz]

IMHO it is a very good idea to use dummy parameters to ensure QuEnc compatibility and let the user edit a template file to change settings.
oh yes i understand why he did the dummy parameters since he said why at the time. as for the INI, uh well thats fine so long as you think everyone is going to understand all the settings and not stuff anything up. :) besides an INI is no good if a program wants to change any of the dummy parameters is it. ok DVD-RB doesn change DC precision(for example) at the moment but that doesnt mean it or another program wont in the future.

Both encoders have the libavcodec rate control issues, but the B-frame thing that QuEnc suffers from is definitely not present in FreeEnc.
and this has been identified in the QuEnc thread so should be fixed next release. not exactly a killing bug though.

[jdobbs]

DVD-RB does change the aspect ratio, bitrate, and other options during execution -- so they have to be accepted from the command line. There are lots of DVDs that use both 4:3 and 16:9 on the same disc (for example).

[manolito]

Originally posted by jdobbs
DVD-RB does change the aspect ratio, bitrate, and other options during execution -- so they have to be accepted from the command line. There are lots of DVDs that use both 4:3 and 16:9 on the same disc (for example).

And of course FreeEnc does support these command line parameters.

[jdobbs]

Is the FreeEnc project still active?

I just noticed that it is broken by the new QuEnc command line options. If there is continuing support for it I'll add it as a true encoder option.

Has anyone done any testing on it? Does it size properly, etc?

[DK]

a couple of days ago i used freenc as well as quenc and both did a pretty good job

maybe asigning freenc a separate encoder entry might indeed be a good idea

[gavin051]

What kind speed are you talking about? I only get around 5 fps using quenc and dvdrebuilder, even with the trelis and high quality turned off. To me this is unacceptable (will take a number of days). Before i got a dvd burner i used dvd2svcd in which i also used quenc, and this returned a much better fps (around 10 i think but can't really remember, but i know it would finish a whole dvd in around a night).

Any reason why it is so slow? My computer is a 3200+ amd64 with 512mb ram.

Thanks
Gavin

After some more playing around and reading I realised that I had checked the extreme and slow box in the advanced settings. Unchecking this gave me around 30 fps with the same options. I guess this was never ticked in dvd2svcd, and is not normally ticked for dvdrebuilder.

Therefore, my problem solved.
Thanks anyway.

[manolito]

Originally posted by jdobbs
Is the FreeEnc project still active?

I just noticed that it is broken by the new QuEnc command line options. If there is continuing support for it I'll add it as a true encoder option.

Has anyone done any testing on it? Does it size properly, etc?

I use FreeEnc on a regular basis, it does size properly, it is pretty fast and gives very good quality especially at low bitrates, and yes, the new -mpeg2mux noaudio | nomux parameters cause it to quit with a "Bad Parameter" error.

I do not know if VMesquita has plans to update the encoder, the current version 0.31 dates from Nov 2004. For myself I just modified my little utility that makes NuEnc compatible to QuEnc, but for DVD Rebuilder I would like to propose a slightly more radical approach:

Now that even Hank315 considers command line support for his HC encoder, just adding options for new encoders in Rebuilder will end up in a constant catch up with new encoder versions. Why not add a generic option for any command line enabled encoder and get the necessary commands from a user editable INI file? Or maybe less complicated just issue the QuEnc command line parameters and us an INI file to translate the arguments.

Example:
FreeEnc only does not like the audio parameters. The INI file would look like this:

" -mpeg2mux" -> ""
" noaudio" -> ""
" nomux" -> ""

For NuEnc it would be something like this:

"-1" -> "-pass 1"
"-2" -> "-pass 2"
" -hq" -> ""
" -nohq" -> ""

and so on...

With this approach the problem of supporting new encoders would be solved once and for all. What do you think?

Cheers
manolito

[Trahald]

Originally posted by gavin051
What kind speed are you talking about? I only get around 5 fps using quenc and dvdrebuilder, even with the trelis and high quality turned off. To me this is unacceptable (will take a number of days). Before i got a dvd burner i used dvd2svcd in which i also used quenc, and this returned a much better fps (around 10 i think but can't really remember, but i know it would finish a whole dvd in around a night).Any reason why it is so slow? My computer is a 3200+ amd64 with 512mb ram.After some more playing around and reading I realised that I had checked the extreme and slow box in the advanced settings. Unchecking this gave me around 30 fps with the same options. I guess this was never ticked in dvd2svcd, and is not normally ticked for dvdrebuilder. Also keep in mind 720x480/576 encodes slower than 480x480/576. but yeah.. when i use quenc , it processes in about the same time as cce ( not exact measurements but its done in the morning and last encode end time is similar)

[jdobbs]

I'm usually getting in the neighborhood of 56fps. I use the default settings.

[dragongodz]

these are my personal observations.

.the defaults should be fine most of the time.
.high quality setting does generally increase quality ,especially for lower bitrates.
.extreme setting is for masochists. ;)