Sorry, even worse here with the Duron, 30-40 min. per prediction cycle. Tried with different and shorter movies too, but they are also very slow. Will take hours to find the correct CQ. How do I configure the Prediction & Spacing page in ToK to use the same sample cycle as in CQMatic? I'd like to try and see if there's any difference in speed when using ToK.

took now a 53 min, wanted it to have it (for tests) on 300mb ;

calculated avg bitrate , with 631.

here the log

using cq 60
encoder started...
encoder end...
difference=-5439988.500000
using cq 90
encoder started
encoder end
differnce=-5439988.500000
cq above watermark , cq set to 90
using cq 90
cqmatic complete

hmm strange :)

Set Length of sample to 6, and number of samples per minute to 4.

-kwag

Load your project (.tpr file) in TMPEG manually, and hit "Start" to encode. See what you get :!:

-kwag

?
don´t know exactly what you mean ?
when i hit start he starts encoding ...

got another test running now, so can´t do another one...

what i wanted to say :
cqmatic is getting information on cdsize, nr of cds...about the average bitrate , i think?

so why does it take cq 90 , when i like to have the movie on 300mb ?

but thats not the most important thing,its more important that it runs stable, finds the cq as fast as possible, so that everybody can use it :)

Kwag, tigger , please:

i have doubts ...
tigger wrote:
"took now a 53 min, wanted it to have it (for tests) on 300mb ;
calculated avg bitrate , with 631."

this size (300mb) is not too low for 53 minutes?
and the average is not too low?
:?

maybe i misunderstand....seems strange for me!
:?

you´re right of course, would never do that :)

but for 1 test, i chose such values...

i thougt that cqmatic would calqulate the cq very low...perhaps 20-30 as cq or sth, but it chose 90

and thats strange for me :)

Just finished running the test.
Source was 104 min. 576 x 304 PAL AVI movie.
Using the same script for both tests, and destination was MPEG2 - 480 x 576 PAL.

One prediction cycle with CQMatic took 16 min 40 sec, with a source range of 1 min 39 sec.
One prediction cycle with ToK took 6 min 17 sec, with a source range of 1 min 44 sec.

I have checked the project file over and over again, and as far as I can see, everything looks ok. This really puzzles me....

Forget about the current beta. Wait for the next release. Don't bother testing testing BETA 7.

This WILL be fixed today on BETA 8 :!: as I'm duplicating the problem now. Until later today, no more BETAS :!:

-kwag

If somebody want beta 7 here the right link:

http://www.kvcd.net/cqmatic-1.0b7.exe

Kwag forgot the "-" ...

Great, looking forward to it! :D

I think this will be the right link ... then:


http://www.kvcd.net/cqmatic-1.0b8.exe

:wink:

hope beta8 will be there soon :)

Kwag I'm still getting CQ prediction loops with beta 7. It starts at CQ 60, then 50.25, 40.83, 40.66, 40.57, 40.53, 40.51, 40.50, 40.50, 40.49...

And that's where it gets stuck. The value for Difference = also doesn't change. It also SEEMS like prediction samples take a longer time to encode than with ToK, but I haven't done a direct comparison.

I'm encoding a widescreen avi, 640x272, 152 min long, avg bitrate 608, into a KVCDx3.

avg 608 @ KVCDx3???
Is the quality still good at these settings?

Oh... CQMatic calculated CQ is wrong... I want to encode a 100 min movie... but the CQ is 100MB too big... :cry:

please fn86,
post big details....
wanted size,CQ found,average,etc.

:)

Try CQMatic RC-1. You shouldn't get those numbers, and it should predict in around 4 or 5 tries :)

-kwag

Actually, I jumped to RC-1 (Release Candidate), because I believe it's now almost ready for prime time :)
Only final prediction accuracy tweaks "might" be needed :cool:
Here's the link, in case some of you haven't noticed it:
www.kvcd.net/cqmatic-1.0RC1.exe

-kwag

Hello Kwag:

Cqmatic is now working for me. The video stream error message that I was receiving, as well as the strehed screen, were corrected following krassis and your answer to previous posts. On Cqmatic v6 I was seeing up to 2 hours of CQ checking, prior to encoding. I am presently running v RC-1, and so far is in its third pass. It takes about 12 minutes per pass, for an 80 minutes dvd. Will let you know how many passes it takes prior to the actual encoding, which in this oicture is about 13.5 hours.....
Question: What is the shortest film, in minutes, that can I test with no problems? On previous versions I have used 4 minutes, but Moviestacker was given me erroneous numbers for the average kbs. ( anywhere from 8,000 to over 240,000- Theshortest the film, the higher the number. Obviously I am doing something wrong here, perhaps I should do this manually no?)

Thanks again for your help

Totonho03

Hi Totonho03,

You should be able to throw any (time) size movie at CQMatic, and it will either encode at a max of 90, or at a min of 1 ( if you like very low quality videos :lol: )
The program should protect MIN,MAX bundaries now :)

-kwag

I'm giving it a try right now. I forgot to count how many attempts it took but it no longer got stuck at 40.49 and instead zeroed in on 37.53. This was on T2 special edition, trying to fit 153 min on one 82 min cdr (with overburn) as KVCDx3. I'm going to try some at lower resolutions and see what my CQs are like, then encode some samples and see how they look.

The only gripe I have with cqmatic is the longer sample times than ToK. I guess I'm just used to Tenra's fast predictions, but I could normally zero in on a CQ in 20 min tops. But now it takes me up to 11 min a sample, and if it takes 4-5 samples well that just adds up! (guess I shouldn't look a gift horse in the mouth but.... :wink: )

;)
Well, If we can get 98% accuracy every time, then I wouldn't mind waiting a ~15% extra overhead for prediction :!:
I'd rather wait that extra time, than find out the next morning that it's under 40MB or over 20MB :!:
But then again, only if it's a consistent ~2% precision every time :!:

-kwag

Kwag:

Sorry for this, but I am getting a loop with vRC 1.
It started at CQ 60, then it went to:
68.19>72.28>73.40>72.31>72.37>72.32>73.34>72.35>73 .31>72.37>73.29>72.38> 72.37> 72.40 and is still going. I have checked and rechecked my numbers, and the only thing that I did wrong was to use 2500 as max number, instead of 2000. Would this create the loop?
Each time it encoding takes about 13 minutes, and the prediction is now over 2.5 hours.........

Totonho

P.s.- Would you loke for me to send you the log? Is a long one......

Hi Totonho ,

I'm now too tired to think :!:
I quit for today. Tomorrow I'll tackle this.
This doesn't make any sense 8O
And the test/predict cycle is really boring :!:, so I have to wait a long time too before I get results.
And the worse thing is that I don't get a single movie to fail on me, and that makes things worse, because I can't reproduce the problems :roll:
To be continued tomorrow .....

Later,
-kwag

Something very odd happened:
I stopped the TMPEGenc, it was still in a loop, printed the log, and surprise, the printed log is longer than the log that the CQmtic was showing win their screen.........No, I only had 4 glasses of Merlot, and am not hic, drunk......... The log even states that CQmatic completed its cycle at CQ 73.054291, (the difference between that one and the previous one is 0.217361, which is less than the .25 tha kwag was talking about............

I will start all over again, but am not sure if I am going to wait a couple of hours for it to finish, so I will report manana por la manana.......

Totonho03

With RC1, my problem seems to be gone :!:
Here's my log (same source as always):
I've tried with a first CQ of 60, i'll let it do a full encode and then i'll try with different CQ's at start.
Thx Kwag for working so hard on this :!:
And thx for the logging feature (very important for me :D ) :drink:

Well, my doubts where right. With another CQ at start, final CQ is different. I've done the second prediction exactly with the same settings except first CQ. This time i started with 90. Here's the log:
EDIT: Same behavior with avs-script with filters. Logs above were done without filters.

sorry kwag, but i still have the problem with short movies, its getting me crazy :)

well still have the problem with short movies

http://www.kvcd.net
CQMatic Version 1.0 RC-1
Copyright Softronex Corporation, 2003.
All rights reserved.

Ready!
Project: C:\--== DVD RIPPING ==--\Band Of Brothers 10 Teile\Teil 4\bob4.tpr

Creating: CQMatic.tpr


Execute.
Movie Time: 57
Average Bitrate: 1785
Prediction Only mode
Executing Prediction Phase...
Using CQ of 80.00
Encoder started...
Encoder end.
File size difference = -16550299.000000
Last CQ = 80.000000
Current CQ = 90.000000
CQ difference = 10.000000
Using CQ of 90.00
Encoder started...
Encoder end.
File size difference = -16550299.000000
CQ ABOVE watermark. CQ set to 90.0
Using CQ of 90.00
CQMatic complete!

will try a lower cq as start value...lets see

Good morning /afternoon gentlemen:

As mentioned, I left the tool running last night, with the max number set at 2,000, and the results were positive:
It went through its calculating routine 6 times, and received a final CQ of 78.3089600 and a CQ difference of .218658.4, the CQ values were:
I printed the previous log as well, and it shows the point where I stopped the encoding. However, the log keeps on going, perhaps looking at the looped numbers...I do not know, I must have also asked to dump the log twice, that explains the 2 sets showing completed CQmatic)

Kwag, something that perhaps would have answered this without doubts is a clock. Do you think that a clock could be placed at the encoding starting and ending points? Just a question. (I just asked CQmatic to dump the info from my recent test to the clipboard, and I now have 2 sets of the same log, it appears that the command is not clearing the contents of the clipboard, if CQmatic has not been closed.)

Total prediction time appears to be around 1 hor and 30 minutes, this for an 80 minutes film and an average bit rate of 1149
Thanks Totonho

Hi Totonho,

can you try the same prediction with a low starting-CQ :?:
I would like to know if it is predicting the same or a similar value.

Hi Krassi.

Will do that, I will start in about 15 minutes, the result will be there in about 1.5 hours.

Totonho03

hi krassie...
well started preditcion of 57min with low cq, 30
jumped directly in the 2nd pass to 90

Kwag: I notice that CQMatic takes a much longer time to encode the same number of frames in its prediction samples than with tmpgenc on its own. For example, the movie I'm on now takes 12 min to encode 3240 frames. But if I add the line trim(0,3240) to my avs and just load that directly into tmpgenc, it only takes 2 min! Sometimes it takes 5-6 tries to get the right CQ, which would mean 60-72 min, whereas I'd "expect" it to only take 10-12 min. Can you shed some light on what this may be due to?

Hello Krassi:

I re-run CQmatic, starting at 70, and this is what I have:

The previous, at CQ 60 was
Hope this helps

Totonho03

And this is my latest log:

Do not think that I need to quote the previous log, but I will do so if you think it is needed.

Thanks

Totonho03

hmmm wondering :)

set moviestacker to 90mincdr, wanted to have 120min movie on 90min cdr, videostream SHOULD be ca. 780mb, but its 820 mb ??

thought that the filesize woulder NEVER go above the maxiumum?

perhaps, next time, i try to configure moviestacker 88mincdr ;)

anyone tested running cqmatic for 90min cdr configuration?

Please get RC-2 :!: ;)

-kwag

Hello kwag:

I just finished running a test with RC-2. Here is my log ( I have changed its format a bit to save space):

I also did a test using CQ 70 on the previous software, RC 1, and the final CQ was:

Which as you can notice it is exactly the same thing.

Hope this help some.

Totonho03

Thanks totonho03 :)
But use RC-2, because it fixes many problems with short (High CQ) encodes that can fail miserably with ;)

-kwag