digitalFAQ.com Forums [Archives]

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)
-   Video Encoding and Conversion (http://www.digitalfaq.com/archives/encode/)
-   -   Bitrates: CQMatic Log (http://www.digitalfaq.com/archives/encode/8635-bitrates-cqmatic-log.html)

Aielman 03-16-2004 10:39 PM

CQMatic Log
 
Kwag,

I hope I am not imposing, but I am experiencing a problem with CQMatic that you can help me (and others having similar problems) with. So far, in every single encode I have tried with CQMatic the final file size is too big to fit on a 700 mb cd-r, even though I always use the X3 prediction mode. So what I do now is get the CQMatic prediction and then do manual encodes, gradually reducing the CQ until I find the highest CQ that will produce an acceptable file size. In one case I did 4 manual encodes before I could get a file size that will fit on the cd. Obviously, this is very very inefficient.

I notice that in each of its iterations the CQMatic log prints the FILE DIFFERENCE (between predicted file size at that CQ and available space, I think). Can you make the log also print the predicted file size at each iteration? That would be very useful if one has to do a manual encode. Thanks.

Dialhot 03-17-2004 05:10 AM

I wonder what difference you have. Give examples of wanted and obtained values please. Just to see what can be done.

Aielman 03-17-2004 11:59 PM

Quote:

Originally Posted by Dialhot
I wonder what difference you have. Give examples of wanted and obtained values please. Just to see what can be done.

Phil:

Below is a listing of the log from a sample prediction I did today. I have taken the liberty of adding to the log places where I think the predicted file size would be useful. If after a completed prediction one reviews the log, it will be apparent how the predicted file size would change with changes in the CQ. That would be very useful if a subsequent manual encode is required.

***********************************
http://www.kvcd.net
CQMatic Version 1.2.06
Copyright Softronex Corporation, 2003.
All rights reserved.
Time: 21:38:11 Date: 03/17/2004
Ready!
Project: C:\Video\Oceans.11.tpr

Creating: CQMatic.tpr

C:\Video\oceans.11.m1v
Project resolution: 352x240
Execute.
Movie Time: 117
Average Bitrate: 809
Prediction Only mode
Executing Prediction Phase...
Process started at 21:38:41
On 03/17/2004
CQ set for prediction
Setting up initial sampling.
Using CQ of 60.00
Prediction cycle #1
Encoder started...
Process time: 13.83 minutes.
Encoder end.
Predicted File Size: xxx,xxx
File size difference = 1.857940
Low fence: 60.000000
High fence: 90.000000
Last CQ = 60.00
Current CQ = 75.00
CQ difference = 15.000000
Using CQ of 75.00
Prediction cycle #2
Encoder started...
Process time: 13.63 minutes.
Encoder end.
Predicted File Size: xxx,xxx
File size difference = 1.463339
Low fence: 75.000000
High fence: 90.000000
Last CQ = 75.00
Current CQ = 82.50
CQ difference = 7.500000
Using CQ of 82.50
Prediction cycle #3
Encoder started...
Process time: 14.22 minutes.
Encoder end.
Predicted File Size: xxx,xxx
File size difference = 1.098089
Low fence: 82.500000
High fence: 90.000000
Last CQ = 82.50
Current CQ = 86.25
CQ difference = 3.750000
Using CQ of 86.25
Prediction cycle #4
Encoder started...
Process time: 13.97 minutes.
Encoder end.
Predicted File Size: xxx,xxx
File size difference = 1.023782
Low fence: 86.250000
High fence: 90.000000
Last CQ = 86.25
Current CQ = 88.30
CQ difference = 2.051216
Using CQ of 88.30
Prediction cycle #5
Encoder started...
Process time: 13.72 minutes.
Encoder end.
Predicted File Size: xxx,xxx
File size difference = 1.019652
Low fence: 88.301216
High fence: 90.000000
Last CQ = 88.30
Current CQ = 89.15
CQ difference = 0.849388
Using CQ of 89.15
Prediction cycle #6
Encoder started...
Process time: 13.28 minutes.
Encoder end.
Predicted File Size: xxx,xxx
File size difference = 1.017849
Low fence: 89.150604
High fence: 90.000000
Last CQ = 89.15
Current CQ = 89.58
CQ difference = 0.424698
Using CQ of 89.58
Prediction cycle #7
Encoder started...
Process time: 13.33 minutes.
Encoder end.
Predicted File Size: xxx,xxx
File size difference = 1.016937
Low fence: 89.575302
High fence: 90.000000
Last CQ = 89.58
Current CQ = 89.79
CQ difference = 0.212349
Using CQ of 89.79
Prediction cycle #8
Encoder started...
Process time: 13.05 minutes.
Encoder end.
Predicted File Size: xxx,xxx
File size difference = 1.016463
Low fence: 89.787651
High fence: 90.000000
Last CQ = 89.79
Current CQ = 89.89
CQ difference = 0.106178
Final CQ value is 89.79
Predicted File Size: xxx,xxx
CQMatic complete!
Total minutes of process: 109.08
Process ended at 23:27:46
On 03/17/2004
[/b]

Dialhot 03-18-2004 04:35 AM

Okay I understood what you would want to have in CQMatic. This answer belongs to Kwag (thos maker of the tool).

Now what I asked to you : can you give me an example of final encoded size that you had and that is too big to fit on a disc ?
With the length of the movie, the resolution used and the average bitrate you used ?

Thank you.

kwag 03-18-2004 11:09 AM

Re: CQMatic Log
 
Quote:

Originally Posted by Aielman
Can you make the log also print the predicted file size at each iteration? That would be very useful if one has to do a manual encode. Thanks.

Hi Aielman,

I think what you mean for predicted file size, is really the sample size that CQMatic want's to try and match on every iteration.
This is exactly ( or more accurate) that trying to do a manual prediction :)
So there is really no predicted file size. The file size difference, is the difference from the encoded sample on every iteration to the fixed "wanted" calculated sample size that needs to match.
The accuracy of CQMatic to "Exit" prediction, is 0.5%.
That means that the file size encoded on iterations must be no more that 0.5% over or under the size of the encoded sample. To achieve that 0.5% manually, is very painful.
The problem with final size prediction is that even though I match the wantes sample by <= 0.5%, the final file size can be way off, because of other parts of the movie that are not sampled in the slicing interval.
So if your predictions are being way off, it's because the movie you're trying to encode has a very mixed distribution of activity.
I'm looking into other techniques to find a fix for this.

-kwag

Aielman 03-18-2004 08:31 PM

Okay. Thanks for the response, Kwag. I hope for all of us that you can find an elegant solution.

Aielman 03-18-2004 08:49 PM

Quote:

Originally Posted by Dialhot
Okay I understood what you would want to have in CQMatic. This answer belongs to Kwag (thos maker of the tool).

Now what I asked to you : can you give me an example of final encoded size that you had and that is too big to fit on a disc ?
With the length of the movie, the resolution used and the average bitrate you used ?

Thank you.

Phil, here are the numbers from the latest encode I did. Because of hard disk space limitations I always delete the m1v file before trying again. I hope the following will give you enough information.

Audio File: 136 min
MP2 File Size: 108 mb

CQMatic Prediction (X3):
Time = 136 min; Average Bitrate = 690.9
Predicted CQ: 83.47

TMPGEnc Manual Encodes:
Maximum Bitrate: 1500
Minimum Bitrate: 300

1st Encode: CQ = 83.27, File Size (muxed): 847 mb
2nd Encode: CQ = 82.80, File Size (muxed): 825 mb
3rd Encode: CQ = 82.00, File Size (muxed): 807 mb
4th Encode: CQ = 81.80, File Size (muxed): 796 mb

I burned the 807 mb file size with VCDEasy and with overburn enabled. The burn was successful (according to vcdeasy) but when I tried to play the movie in PowerDVD it stopped in the middle with an error message. So I lowered the CQ to get a smaller file size and tried again. I have tested the burned 796 mb file in PowerDVD and encountered no problems, but I haven't yet had the chance to watch it on the standalone. Hope this helps.

Dialhot 03-18-2004 10:21 PM

Okay; It seems you did prediction with audio at 112. Do it with 128. This way if CQMatic find a CQ too high, you will have the possibility to do your audio in 112 and obtain a correct size for video+audio.

Note: if you 8078 video stop in the middle that is not due to the overburning. You should be abble to burn 807 MB with no real problem on every burner.

kwag 03-18-2004 10:36 PM

Quote:

Originally Posted by Dialhot
Note: if you 8078 video stop in the middle that is not due to the overburning. You should be abble to burn 807 MB with no real problem on every burner.

... Because if there was an overburning problem, then your "freeze" would be at the end. Not in the middle ;)

-kwag


All times are GMT -5. The time now is 09:53 PM  —  vBulletin © Jelsoft Enterprises Ltd

Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.