What is the CCE SP best encoding type and prediction?
 

What is the CCE SP better way related to encoding type and prediction.

Are the following tools useful?
http://www.autoq.net.tc/

I use Tylo's CCEFront. I create an ecl file in CCE as usual and then run CCE via CCEFront with parameter -roba.

It will predict and do an OPV encode and a sizing VBR pass if necessary. You can set the upper and lower limits for the sizing pass to occur. I've found out that 3% sample size and 3% size adjustment hit the target more often than with the default 1% and 1.4% respectively. The good thing is that the defaults can be set via an ini file so you'll just need the -roba parameter if you already have an ecl file.

IMO AutoQ is a piece of crap. I really don't think Amnon82/Avalon/whatever ever really tested his prediction engine. He just did the predictions but never ran a full encode. AFAIK his program also doesn't do the sizing pass so you'll have to do it manually unless the program hits close enough to the target with the OPV pass.

@Boulder

Hi, mate! Could you post a .ecl file that you have got used it ? In this way I could to use as template for my encondings using new CCE 2.70 and CCEFront by Tylo?

TIA,

There's really nothing special about the ecls I use. Besides, you'll need to adjust the ecl according to the source. The average bitrate is what CCEFront will target in the OPV and sizing pass.

I use CCE 2.70 since a few weeks and the old reg that vmesquita did for 2.67 works... almost. The only thing to adjust is the encoding mode (even if it is set to OPV in the reg, CCE 2.70 shows it as CBR -> change this) and change the DC precision also.

Once you loaded, fixed and saved back the template, you can create your own ecl.

@Boulder

Thanks a lot, fella :!: :D

@Dialhot

Merci, camarade :!: :D

I don't know nothing about CCE. How can I must to do a ecl file?

If you use CSP prior to 2.66, and ecl is the only way to open an avs.
Starting with 2.66, it's not more necessary but an ecl can sttil be done, to save the parameters. It's like saving a tmpgenc project into a text file : you do not need to do that, but you have the possibility to do..

Yes, but what parameters set, when that window pop up?
encoding mode, etc., etc.
Also I know how to change the matrix inside ecl. :D

All right, all right.

http://www.saunalahti.fi/sam08/progressive.reg
http://www.saunalahti.fi/sam08/interlaced.reg

Thank you mate.
Excuse this noob, please. :D

Go on vmesquita site, and retrieve the guide he did for KDVD using CCE. You will find with it 2 reg file that define templates for KVCD and KDVD into CCE. All settings are done in them, including the matrix :)
(the guide explain how to use them)

Yes, but I didn't understand what I have to change (that you say in a post before).

So you have CCE 2.70 ?
Okay. Load the template form VM into your registry. Then open the CCE. In the template menu you will find a new entry for KDVD and one for KVCD ([b]BE CAREFULL: VM reg files remove all templates delivered by CCE. But they are useless IMHO)

Select the KDVD template twice, this will open the window to change the template. There you can see that the encoding mode is CBR. Change it to 1-pass VBR. Change the DC precision to 8. Save the template.

Now, to encode a KDVD, just check that "KDVD" template is selected in the template menu and simply drag/drop you avs in the CCE window. All settings of the selected template will be applied to this job.

Thanks Phil.

@Boulder

I agree with you on AutoQ I have tested but was overly impressed.

I use Johnny's QCCE I like it's prediction method and it's pretty close for getting predicting results.

How do you find QCCE Boulder, do u like this tool also?

I did use QCCE quite a lot before CCE 2.70 came along. Amnon's tool is a ripoff actually :wink: The only thing I found a little annoying was that the prediction is a bit slow due to the binary approach, and that you can't run a sizing pass if the OPV encode is off the target.

OT-On:
I just keep wondering why doom (owner) permits Avalon (alias Amnon) on his site, when he is clearly aware of all the things he has ripped off :roll:
OT-Off

-kwag

He's probably noticed that in the end they tend to be crap. A pile of bugs and weird behaviour - and it looks like he doesn't care much about MPEG2 stuff anyway. That OPV comment on the D9 front page was a little weird :roll:

@Boulder

ccefront is really a killer app ! 8O But I just wonder why in the sizepass the app don't call the rejig (transcoder) if I´ve chosen mode=2 ("transcode if opv size > 0%). For now, ccefront is doing all encoding again (no transcoding).

I´ve used:

and the roba's section in ccefront.ini follows down:

log.txt for first OPV:

log.txt for "sizing pass":
note: the encoding hasn't finished !!!

Should be a bug in ccefront?

TIA,

Try adding quotes to the ReJig path, or alter the path so that it doesn't contain any spaces :
rejig_exe="C:\Arquivos de programas\DVD-RB\ReJig\ReJig.exe"

@Boulder

Thanks, mate! I'll follow your suggestion and I let you know about that.

Cheers,

Hi Boulder,

Did you mean that AutoQ is a bit slow if so then I agree AutoQ for me runs reallyi slow which is one of the reasons I dislike it.

I agree, I remember the problems he caused here with his rip-offs and he's still going.

AutoQ just doesn't do the work it's supposed to do. The prediction engine doesn't work, you end up with under- or oversized encodes 99% of the time. There's some weird rounding when the sample size is calculated and this screws up things before they even start. Besides, from what I noticed in the posts I read in the AutoQ thread, Amnon mostly did predictions, he didn't do many encodes with those predicted values.

And it is slow as it uses the binary method for calculating the next Q. That is, in range 1-100 first encoding at Q50, then either Q75 or Q25 depending on the result and so on. It's slow and very frustrating if you notice that Q48 would have hit the target but the prediction goes 50->25->37->43->46->48.

I pointed him to Tylo's spreadsheet which would have helped him to speed up the process but he either didn't understand it or whatever :roll:

Hi Boulder,

Yeah I read all those posts over at D9 complaining about the undersizing prediction and compared to QCCE produced smaller than desired sizes.

Also like you said it is soooo SLOW! I don't use because of this.

The thing that annoyed me over at D9 is that Anmon didn't listen to any of the people's advice when they tried to tell him to use the algorithms that Johnny used in QCCE and stuck with his own inferior methods of calculating.

The only thing that puts me off CCEFront is that is CLI based I guess im lazy and like nice GUI's. :D

Sorry for the off-topic there guys. :oops:

Tylo is preparing a simple GUI for the program, but I don't know when it will be out.

Thanks Boulder, that will be great for lazy peeps like me. :lol:

Actually there's not much you need to do with a GUI. I run the program with only the -roba parameter, all the default values can be set in its ini file. I adjust all the necessary CCE settings in CCE itself and save the ecl file which I use with CCEFront.

Then, is it enough one ecl for progressive and another for interlaced material, or each one must to be adjusted for each source?

The ecl contains all the information, that is, everything you see in the CCE GUI itself. Thus, the same one cannot be used. You must use the templates for the initial settings, load the avs script and then just change the average bitrate you desire and save the ecl file. With CCEFront you must use VBR mode in CCE, that's where the program takes the target bitrate (unless you specify things manually as a commandline parameter).

Consider CCEFront the same for CCE as CQMatic is for TMPGEnc.

Thanks Boulder. I would like recall your attention, if i'm not wrong, about that those reg files, have QLB matrix inside, isn't it?

About your spreadsheet of your last post, could you send me copy or post here equations.
Thank you.

The matrix is QLB with the notch. I like to quantize less so I took QLB and added the notch pattern. You can change it if you like, of course. Just open the template (kind of "select the template twice" and you get to edit it), change the matrix to your liking and choose "Replace".

Here's the spreadsheet : http://www.saunalahti.fi/sam08/kbps_eng.xls . It should be pretty straightforward, but ask if you can't figure out something. The average bitrate you should use for a movie is in the lower part (Avg kbps for final video).

Thank you Boulder. I have learn already how to change the matrix, but it took some time until I have seen that it was not the KVCD notch matrix. Not even knew that you can take any matrix and put the notch pattern inside.

Is it a good practice to use rejig to downsize an oversized stream?

Rejig is to m2v what DVDShrink is on a complete DVD.
You can use also the recoder include into DVDLab-Pro. Or you can author the DVD without reducing the size of the M2V and you will apply DVDShrink on the generated DVD.

Rejig is like all other tools in this category : nothing very good to wait from them, but for a little ratio (90% or 80% maximum), they make you save a lot of time.

Hi guys,
I was trying to download these templates but the server says that they are no longer hosted.
Could somebody please upload the templates again.
Thanks guys

Take it

http://s21.yousendit.com/d.aspx?id=0...81N2PQY9FETT6T

Thanks Pro :)
I owe you one :lol:.
Cheers

Your welcome.