Bitrates: Testing CQmatic Versions - Page 6

vhelp · #**101** 08-08-2003, 12:18 AM

Hi Kwag..

Quote:

Originally Posted by kwag

Select "Prediction Only" in CQMatic. When the process is finished, run TMPEG. You'll have the last CQ already set. Just clear your range, select "System (Audio+Video") and add your audio in the "Audio Source" of TMPEG.

Now you can encode full Video + Audio.

-kwag

Sorry old chap, but it's not working !!

It's encoding the whole thing (all 36,513 frames of the 21 min. vob)
I need it to encode the same prediction frames, but only the audio.

What I CAN do is let it go through the predict once, w/ auido and video,
and then when TMPG comes up again w/ MPEG-2 warning message, I can
copy the mpeg file to an new name, and then cancel out of CQM and TMPG
and then de-mulplex and reMUX - problem solved, but at an expense.

I guess, you hadn't accounted for anyone wanting to include audio during
the prediction phase ??

-vhelp

vhelp · #**102** 08-08-2003, 12:23 AM

@ Kwag..

regarding my previous post above, actually, what's happening is this.

It's going through the prediction phase just fine, but as soon as the last
prediction frames are processed, it stays on one frames, and continues to
encode blank frames, for the remaining 36,513 frames. If that helps any.

Note, the first few frames (from CQM prediction phase) are the same.

-vhelp

kwag · #**103** 08-08-2003, 12:27 AM

Did you Select your range of frames in TMPEG

If you don't, it will continue to encode for the length of the audio

-kwag

vhelp · #**104** 08-08-2003, 12:34 AM

Kwag..

Quote:

Originally Posted by kwag

Did you Select your range of frames in TMPEG

If you don't, it will continue to encode for the length of the audio

-kwag

No, no... I think you misunderstood me

Ok, picture the scenario...

A - I use the "Matrix" rip (first 21 minute VOB) for the CQM predict testing.
B - I've now completed the (A) and have a small 2mb file thats only .m2v

But, I'd like to include the audio w/ this .m2v sample (frame for frame w/
audio)

So, you tell me, it can be done, and just redo it, but this time, w/ the setts
(audio + video) but it doesnt work that way.

What I want, is (B) above, and the audio to match it with. So, my theory
was that if I re-ran the CQM test w/ the same as in previous (video only)
test, but this time, either with Audio only, or Audio + Video.

If (Audio + Video) then I can de-MUX it and re-MUX - - but only if I have to.

Now do you understand what I'm asking ??

-vhelp

vhelp · #**105** 08-08-2003, 01:29 AM

Hi guys..

Sorry, I'm exhausted, w/ trying to get Audio either included w/ CQM's
predictions (w/ same video frames used in predict test encodes) or with
both audio and video together, but not the whole video, just the predicted
encoded video (and audio) all for MUXing purposes really. Anyways..

Since this can't be done (I've tried to no avail) below is what I encoded w/
CQM's predicted CQ of 39.86 !! So, it looks like it would take a CQ 39.86
to fit 136 minutes to fit on one 800mb cdr (based on 21 minute test VOB
computations)

* matrix.mpg.sample.21min.cq(39.86).rar

Note, make sure the extension name is .rar (not .mpeg) some browsers
will think it's an .mpeg and name the file that way. Just rename to .rar
and use WinRAR to extract the actual .mpg file !!

Be sure Powerdvd "Config Setup" properly for best viewing 16:9 ...
-------------------------------------------------------
* 4:3 Pan&ScanScreen...
[change to 1024 x 768]
[x] ask me before change

* 16:9 Widescreen...
[change to 1024 x 768]
[ ] ask me before change

* Screen Control...
[ ] Run full screen
[x] Keep screen Aspect Ratio
-------------------------------------------------------

Windvd gives very poor results w/ 16:9, and not to mention incorrectly too,
unless it's from an actual DVD disk.

I'm pooped, so good night !!

See you guys on TGIF day !!
-vhelp

kwag · #**106** 08-08-2003, 01:33 AM

Quote:

Originally Posted by vhelp

Now do you understand what I'm asking ??

-vhelp

Yes, but you can't run Audio + Video with CQMatic, because the formula used is for Video stream only.
I'll take that into consideration for a future version

When I integrate the calculator, I'll set an "Audio" option, to take into consideration if it's a "Video Only" stream, or if it's a "Video + Audio" stream

-kwag

J-Wo · #**107** 08-08-2003, 12:50 PM

Wooaaaaah is something wrong with 1.1.04 here or what? So I'm encoding Terminator 2, Divx source, split into two parts. I'm working on part 1 right now which is 79 min and an avg bitrate of 1227 (from MS). First I ran prediction using 1.1.03 and it gave me a CQ of 83.76. So then I loaded the same project file into 1.1.04 but overode default CQ so it would start at 83.76, and it predicted CQ 58.91 after only 1 try! Then I did it again but this time started at default CQ 60 and after one prediction run it predicted that as the final CQ. Something smells fishy here? I'm going to encode at all three CQs to see which is actually accurate. Here's a copy of my log:

Code:

http://www.kvcd.net
CQMatic Version 1.1.04
Copyright Softronex Corporation, 2003.
All rights reserved.
Time: 13:38:25 Date: 08/08/2003
Ready!
Project: E:\Movies\T2 Special Ed Divx\part1.tpr

Creating: CQMatic.tpr

E:\Movies\T2 Special Ed Divx\part1.m1v
Project resolution: 528x480
Execute.
Movie Time: 79
Average Bitrate: 1227
Full Encode mode
Executing Prediction Phase...
Process started at 13:38:46
On 08/08/2003
Setting up initial sampling.
Using CQ of 83.76
Prediction cycle #1
Encoder started...
Process time: 1.93 minutes.
Encoder end.
File size difference  = 0.703311
Low fence: 2.000000
High fence: 83.760002
Last CQ = 83.76
Current CQ = 58.91
CQ difference = 24.850662
Using CQ of 58.91
Prediction cycle #2
Encoder started...
Process time: 1.90 minutes.
Encoder end.
Encoding set to Full encode.
Full encode start...
CQMatic complete!
Total minutes of process: 3.85
Process ended at 13:42:37
On 08/08/2003
Project: E:\Movies\T2 Special Ed Divx\part1.tpr

Creating: CQMatic.tpr

E:\Movies\T2 Special Ed Divx\part1.m1v
Project resolution: 528x480
Execute.
Movie Time: 79
Average Bitrate: 1227
Full Encode mode
Executing Prediction Phase...
Process started at 13:43:59
On 08/08/2003
CQ set for prediction
Setting up initial sampling.
Using CQ of 60.00
Prediction cycle #1
Encoder started...
Process time: 1.92 minutes.
Encoder end.
Encoding set to Full encode.
Full encode start...
CQMatic complete!
Total minutes of process: 1.92
Process ended at 13:45:54
On 08/08/2003

Bchteam · #**108** 08-08-2003, 03:03 PM

@ J-Wo

As Kwag already said, the prediction for Divx isn't accurate, because of Codec - related problems. You probably should use Tok for DivX Encodes.

@ All

The first full encode, which I predicted with CQMatic Version 1.1.0.4., is going to be done in 5 - 7 Minutes. It looks very promising, i'm at 98 % right now and the current Size is 674 MB. The target Size is 688 MB. I think, we found a solution. Cheers

Bchteam · #**109** 08-08-2003, 03:14 PM

Wohoo....

The Process is done. The target size was 704.512 KB and the Final Size is 706.063 KB. Damn, that's so f*ckin' close. It's an awesome feeling, to have reliable prediction

Krassi · #**110** 08-08-2003, 03:16 PM

Hi Bchteam,

great to hear that

Can you post some details

I would like to know how you calculated your average bitrate.

Bchteam · #**111** 08-08-2003, 03:18 PM

I used Moviestacker 2.0 for the average Bitrate.

What exact details you wanna here ?

Krassi · #**112** 08-08-2003, 03:22 PM

Did you use the MA-script

I ask because i'm currently testing without filters. The value given by moviestacker is giving me a slighty oversized file. I think CQMatic does a good job, maybe the problem is the given average bitrate.

Bchteam · #**113** 08-08-2003, 03:25 PM

That's the Script:

Code:

## Defined Variables and Constants ## 
# 
MaxTreshold = 1.50 
nf =  0 # Current frame. 
# 
#### 

## Main section and static filters ### 
# 
Mpeg2Source("J:\2009\2009Project.d2v") 
#
Telecide()
Decimate()
undot() 
Limiter() 
asharp(1, 4) 
BicubicResize(528, 480, 0, 0.6, 0, 0, 720, 576)
STMedianFilter(8, 32, 0, 0 ) 
MergeChroma(blur(MaxTreshold)) 
MergeLuma(blur(0.1)) 
# 
# 

## Linear Motion Adaptive Filtering ## 
# 
# ( Portions from AviSynth's manual ) - This will apply temporalsoften to 
# very static scenes, and apply variable blur on moving scenes. 
# We also assign a variable - and this is why a line break is inserted: 

SwitchThreshold = (Width<=352) ? 4 : (Width<=480) ? 3 : 2 
ScriptClip("nf = YDifferenceToNext()"+chr(13)+ "nf >= 2 ? \ 
unfilter( -(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100)) ) : \ 
TemporalSoften( fmin( round(2/nf), 6), round(1/nf) , round(3/nf) ,1, 1) ") 

# 
# 
# 

LetterBox(16, 16, 24, 24)
Limiter() 

# 
# 
## Functions ### 

function fmin(int f1, int f2) { 
  return (f1<f2) ? f1 : f2 
}
#
####

I can't confirm that Moviestackers Avg. Bitrate is causing too high file sizes.

kwag · #**114** 08-08-2003, 10:32 PM

Just finished my "Styx - Return to Paradise" encode.

Processed directly via .d2v (no filters) and used TMPEG's internal deinterlacer (double).
CQMatic 1.1.06 calculated CQ of 31.32. Low because of 528x480 "Full Screen" encode, but this was just for a final file size test.
Moviestacker wanted file size: 724,391.25KB
FInal encoded file size: 735,946KB
For a difference of +1.57%

-kwag

totonho03 · #**115** 08-09-2003, 09:10 AM

Hello Kwag:

Yesterday, I tried a couple of things, using CQm v1.1.06:
1.- I started doing a prediction on a short 20 minutes clip, but Moviestacker gave me an average bit rate of 5,000+. After I while I aborted this test, thinking that the prediction was erroneous. In addition to that, the min bit rate would have been higher than the max range of 2000, and I did not know what results CQmatic would have given me had I used such high average bit rate.
2.- I also started a second prediction, this time with the full 120 minutes, using avi (d2v) as my source. The resolution was 528 X 480, and it gave a prediction of about 41.+. Actually I stopped it in its 5th run, because I thought that it was given me a low CQ.
3.- Next was the full encode, on the same film that I was testing before, using avi as source, and a resolution of 352X480. The results are:

Target range: 682 Meg (Moviestacker)
Encoded size: 668 Meg. Which is within 2% of what I wanted
Which is great. You may remember that I was getting CQs ranging from 51 to 90, with file sizes anywhere from 1.2 G to 900+ megs. Now it is within 2%!!!!!!!!!!!!

There is an issue that I would like advise:
In viewing the M1V file, via your kvcd player, I noticed that the video is horizontally stretched. The movie is anamorphic, and the ratio setting that I am using in Tmpgenc is 16:9. Is this correct or should I use different settings? ( As in 1:1?)

Thanks an regards

Totonho

kwag · #**116** 08-09-2003, 11:26 AM

Quote:

Originally Posted by totonho03

There is an issue that I would like advise:
In viewing the M1V file, via your kvcd player, I noticed that the video is horizontally stretched. The movie is anamorphic, and the ratio setting that I am using in Tmpgenc is 16:9. Is this correct or should I use different settings? ( As in 1:1?)

Hi Totonho,

The KVCD player uses the same engine as WMP. So the aspect will look wrong in both players. Use WinDVD or PowerDVD for playback

-kwag

vhelp · #**117** 08-09-2003, 12:20 PM

Hi totonho,

Quote:

There is an issue that I would like advise:
In viewing the M1V file, via your kvcd player, I noticed that the video is horizontally stretched. The movie is anamorphic, and the ratio setting that I am using in Tmpgenc is 16:9. Is this correct or should I use different settings? ( As in 1:1?)

If you wnat TRUE'est aspect ratio play for 16:9 encodes (encoded to 16:9)
and, which I label as OUT(16:9) you should really use only PowerDVD.
I use v3.0 but I would assume/suspect that any version over v3.0 would do.
.
.
and follow it w/ these config setting, for max TRUE'est 16:9 playing pleasure:

Be sure Powerdvd "Config Setup" properly for best viewing 16:9 ...
-------------------------------------------------------
* 4:3 Pan&ScanScreen...
[change to 1024 x 768]
[x] ask me before change

* 16:9 Widescreen...
[change to 1024 x 768]
[ ] ask me before change

* Screen Control...
[ ] Run full screen
[x] Keep screen Aspect Ratio
-------------------------------------------------------

And, you can test this view out, w/ the sample I have here:
* SAMPLE Encodes.. KDVD; KVCD; CVD; SVCD; VCD and "x"
...and, look for #6, which is my latest one.

It's a very low bitrate encode, based off a CQM topic, but do giev it a try
anyways, to get the feel for what I'm talking about !!

playing in normal windowed size (16:9 non-fullscreen view) really bring out
the 16:9 feel, not to mention the FullScreen 16:9 view too, but the windowed
view will deffinately tell you the TRUEth of 16:9 encodes.

Note, this is for MPEG-2 encodes to 16:9 output

If you play your encodes w/ anything else WinDVD or Zoom or any other
app that claims 16:9, then you're only jipping yourlsef w/ thoughts of:

and

DYSAF, and follow my regimin, and you'll thank me (perhaps, later)

-vhelp

totonho03 · #**118** 08-09-2003, 01:16 PM

Hello gentlemen:

@kwag: I am wieving the film with Powerdvd, and it looks stretched indeed. I have viewed both the original vob file, as well as the M1V and mpg versions of the same film. The vob file looks very good in Power DVD, but the encoded M1V and Mpg files are stretched.

@vhelp: I have setup my powerdvd with the settings you provided. and it did not help. as mentioned above, if the vob file plays ok, then the encoded file should also play Ok, if it was encoded properly. The input to the encoder came directly from dvd2avi, with no filters, and it was encoded using the 16:9 ratio. Perhaps I need to encoded using different ratio settings. While I may not be able to experiment today, I will do that tomorrow afternoon.

At least CQmatic is working!!!!!!!!

Thank you guys for your help and advise.
Regards

Totonho03

vhelp · #**119** 08-09-2003, 01:30 PM

hi totonho,

Quote:

@vhelp: I have setup my powerdvd with the settings you provided. and it did not help. as mentioned above, if the vob file plays ok, then the encoded file should also play Ok, if it was encoded properly. The input to the encoder came directly from dvd2avi, with no filters, and it was encoded using the 16:9 ratio. Perhaps I need to encoded using different ratio settings. While I may not be able to experiment today, I will do that tomorrow afternoon.

The reason why it did not play, is because there is no 16:9 suport for VCD,
or MPEG-1

I know that TMPG has the 16:9 setting (even for MPEG-1 encodes) but I think
that this may be an oversight, that it includes this features setting for MPEG-1
or that maybe, perhaps its there just for the "Aspect Ratio" of things, but
nothing more.

MPEG-2 has the flag for 16:9, but AFAIK, MPEG-1 does not, hence your
proof during play, in PowerDVD and standalone DVD player (as in my
experience too, hence my conclusion as well on this 16:9 issue for MPEG-1
sources)
.
.
But, don't be fooled by the 16:9 output for MPEG-1, when you view it in
Zoom player or any other SW player out their. It's just stretching it to look
like 16:9 (giving you the illusion of 16:9 w/ those letterbox bars)

That's why I made certain, I stated that the 16:9 settings and all, for the
SW PowerDVD v3.0 was for MPEG-2 encodes only !!

Now, you can restle this all you want, till you're BLUE in the face, but you'll
get nowheres, when you finally play it in your standalone DVD player !!

Have you at least even looked at the sample I posted, to get the feel for
what I am on to here ??

FWIW here, when you see me writing like this:
* IN(16:9) means your source is 16:9 aspect ratio
* OUT(16:9) means you encoded the IN(16:9) source to output 16:9 !!

Sorry for the ..off topic.. here!

Hope that helps a bit

-vhelp

totonho03 · #**120** 08-10-2003, 03:27 PM

Hi vhelp:

Thanks for your answer. I tried downloading your sampler, but for some reason my computer thinks that this is an mpeg file, and I was not able to downloaded as a RAR file. I will try again later on today.

Thanks

Totonho