digitalFAQ.com Forums [Archives]

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)
-   Video Encoding and Conversion (http://www.digitalfaq.com/archives/encode/)
-   -   Mencoder: Using Bitrate Viewer (http://www.digitalfaq.com/archives/encode/10131-mencoder-bitrate-viewer.html)

maurus 06-07-2004 05:55 AM

Using Bitrate Viewer
 
I encoded a chapter of a movie with new Kwag's MEncoder adaptive params and the same sample with the previous params using unsharp.

I don't know use Bitrate viewer... :oops:

How I can use Bitrate Viewer for determinate the quality of both encodes?

Thanks.


-Maurus

Dialhot 06-07-2004 06:18 AM

Bitrate viewer, as reported in the name, is only a tool to view (graphically) the bitrate all along your movie. Unfortunalyl that has NOTHING to do with quality.

You can compare bitrate usage for both sample, and also the Q curve but both samples can be very bad or very good whatever the bitrate curves displayed by bitrate viewver.

maurus 06-07-2004 06:20 AM

What is good Q curve?
If it is plain?
if it is more high than another?
If it is down or upper the bitrate curve?

Thanks.


-Maurus

incredible 06-07-2004 06:27 AM

Bitrateviewer can only be used to see "how" bitrate behaves. And how Q (RLE encoding descrambling) comes out. Conclusion .. Bitrateviewer is (as already mentioned by phil) is a) not acurate b) it says nothing about the real quality.

So the best comparing tool are your own eyes by watching and comparing parts of the video.

Example: Tmpgenc low bitrate encodings do very often got in Bitrateviewer a higher Q factor, means more quantisation, means more artefacts/ringings etc. then compared with CCE.
But TmpgEnc low bitrate ncodings for instance "look" better when playing back then the ones from CCE.
See the point?

:) Inc

EDIT:

A Q curve is relative and SHOULDN't be flat! As thats one of the mpeg2 advantage purposes which bases in a dynamic Q curve. The best would be if Q values on dark parts do fall (=less quantisation = less blocks on flat dark areas) and rise on High action parts ... but still in a way that blocks etc. won't result.

I think that line above does answer all your 3 questions.

Dialhot 06-07-2004 06:29 AM

Quote:

Originally Posted by maurus
How is good a Q curve?
If it is plain?
if it is more high than another?

By experience :

1/ a flat curves is not the garant of a godd quality but genally, if this flat curves is arround a low value (let say up to Q=4) then the sampel is good.

2/ the lower the Q value is, the better is the (supposed) quality of this point in the movie. But there are good video with AVG Q to 8 and bad ones with AVG Q to 5 :!:

incredible 06-07-2004 06:36 AM

Thats exactly the point Phil!
Thats why TmpgEnc encodings with higher Qs often got a better visual output. So Q curve in Bitrateviewer means nothing related to real "visual" final quality.

A Q curve in Bitrateviewer ONLY shows how the inverse DCT transformation worked on the encoding, means when it will again be decoded for showing on the PC system or SA Player/TV.
If during the decoding many equal numbers do result (which is a purpose of a DCT matrix to get continous similair numeral values for better compression) then a low Q curve appears. Said in small words (cracks would hit me now): BV shows therefore only some kind of decoding quality but it cant show the encoding quality as it doesnt know how the stream state ws BEFORE encoding :wink:

maurus 06-07-2004 06:45 AM

Thanks to both.

I will use my eyes and I see best the sample encoded with the last Kwag adaptative parameters. It's a little, not too much. With a bitrate of 1310 for my KSVCD the quality difference is not big, but I see best quality.

But the time of encodig is double also...

-Maurus

Dialhot 06-07-2004 07:03 AM

Quote:

Originally Posted by maurus
But the time of encodig is double also...

That is one of the reason that made me put mencoder on the side for a while : to achieve a behaviour equivalent to what we had with tmpeng for KVCD you must use a command lien that drop so much the speed than on my last attempt, mencoder need 6h per pass (12h in total) where tmpgenc did the job in 8 :!:

For sure, as told, the is ONE of the reason as speed is, and never has been, a major issue in KVCD process.

maurus 06-07-2004 07:49 AM

But I love the automatic proccess with Mencoder, BeSweet and Mplex. The time is not a big problem if the computer works for me all the night... 8)

And with the last AviSynth MA script and MakeAVIs in my computer the speed is 1/2 fps. With the Kwag's last adaptive parameters of Mencoder the speed is 5/6 fps.

-Maurus

rds_correia 06-07-2004 08:16 AM

Quote:

Originally Posted by maurus
And with the last AviSynth MA script and MakeAVIs in my computer the speed is 1/2 fps. With the Kwag's last adaptive parameters of Mencoder the speed is 5/6 fps.

Hi Maurus,
Yes that could be an argument.
Using Mencoder you always need 2 pass, so 5/6 frames divided by two passes equals to 2,5/3 frames.
If the total encoding time of 1/2 frames is 15 hours then 2,5/3 frames should be around 8/10 hours, and that's noticable.
What about video quality using your eyes?
Is it similar?
I'll run some tests on it tonight.
But always remember: Mencoder is good for 2 (or sometimes 3) movies KDVD encoding.
It's not a very good option when it comes to KSVCD.
Cheers

Dialhot 06-07-2004 09:04 AM

rds, both speed maurus was refering was for mencoder. The 6 fps is with usage of internal filters only, the 0.5 was with usage of an avisynth script, via makeAvis. And we all know that this made drop the speed by half or more.

So in fact he is encoding at 0.25 fps to 3 fps.

rds_correia 06-07-2004 12:00 PM

Quote:

Originally Posted by Dialhot
rds, both speed maurus was refering was for mencoder. The 6 fps is with usage of internal filters only, the 0.5 was with usage of an avisynth script, via makeAvis. And we all know that this made drop the speed by half or more.

So in fact he is encoding at 0.25 fps to 3 fps.

Hi Phil,
In fact I understood that Maurus was doing:
- D2V > AVS & MA script > makeAVIS > Mencoder :arrow: between 1 and 2 frames per sec.
- Single VOB > Mencoder & MA mimick :arrow: between 5 and 6 frames per sec.
What I definitly didn't think of is that all the figures he provided were most probably took from one of the passes
he was doing wether he was doing single-pass or 1st pass of 2-pass.
So there wouldn't be a reson for all the calculations I did :oops:
Maybe Maurus can clear this up.
Cheers


All times are GMT -5. The time now is 02:19 AM  —  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.