DebugView analyzer for XviD codec v0.9

[MoonWalker]

Another new version :)

v0.9
Optimized the code.Now the programm is faster about 70%
Added Report of Consecutive I-Frames

To-Do (Among others :) ) :

Report concecutive I-Frames that are 1-10 in distance(for use with Koepi's 27/07/02 build and over)

You can get it following the link in my signature

MoonWalker

[scorchED]

how do the analyzer calculate the quality of the XviD movie?

[MoonWalker]

Originally posted by scorchED
how do the analyzer calculate the quality of the XviD movie?

As it were in Nandub/DivX3.11.

Quality=200/Average Quant

MoonWalker

[Marc FD]

Qual = 200/avg quant

that's what i name "heavy statistical advanced calculations" ;)

(PS : imagine a quant1 encode :D 200% quality :cool: whow, amazing !)

[MoonWalker]

Originally posted by Marc FD
Qual = 200/avg quant

that's what i name "heavy statistical advanced calculations" ;)

(PS : imagine a quant1 encode :D 200% quality :cool: whow, amazing !)

:D Well is was derived from Nandub quality indicator...

It is 200/avg quant because quality 100% is with quant 2(the 1-pass is done with quant2=maximum quality)..

Personaly I never look this value cause it's theoritical, but it is a request :p

MoonWalker

[Marc FD]

I already knowed that :p
(keep in mind i'm always lurking on the forums :cool: )

BTW, did you tried xla ?

[MoonWalker]

Originally posted by Marc FD
I already knowed that :p
(keep in mind i'm always lurking on the forums :cool: )

BTW, did you tried xla ?

:cool: :D Check your mail

[scorchED]

thanks guys for that explaination. i read that somewhere but forgot it :(

[serbersan]

I suppose the question of scorchED was about compressibility.

But how do the analyzer calculate the quality of the XviD movie? I mean now Xvid Quality. I have a movie that have 40% compressibility and 50% Xvid quality. What the last value means exactly? and/or what it means related to compressibility(40%)?

...and there is a good range? Well the 50% means for me a minimum value, less isn't good enough. But is 65% (for example) a good value to raise the resolution?


Thanks and continue so good job¡

[scorchED]

serbersam

the quality of an XviD movie is calculated by the formula Quality=200/averageQuant. the averageQuant is the result when you add all Quants and devide it by the number of the used Quants. In the first pass of XviD all frames where encodet by Quant 2 - so the average quant is 2 -> 200/2=100% -> so the first pass is the most reachable quality.
The compressibility is the factor between the first pass and second pass. If the first pass output needs 1000MB and the second pass output is 650MB so you reached a compressibility of 65%.

[Marc FD]

where do you get a compressibility value ? GKnot ?

[scorchED]

and i think a quality of 65% is not a reason to increase the resolution. i would increase the res by 85% or above a little bit for a high quality ripp.

[scorchED]

Marc FD;

you can read the compressibility value at the end of the "DebugView Analyzer" results

[Marc FD]

my goal is quant 2 satured 1CD encodes...100% quality :cool:
but i think there is tons of work to achieve before we reach that... :D

[scorchED]

i think its very hard to encode a full movie on 1CD with Q2.
You need very strong filtering and no action scenes or always slow motion. and sthe movie has to be as dark as possible, because higher contrasts will increase the size.
thats not the goal for a good Ripper to encode the movies in that way:p
i think if the movie has Q2 and Q3 and sometimes Q4, it's okay and good looking.

[Marc FD]

it was just a joke :D
(of course quant 2 with 3-4 in action scenes is a must ! but try to get it on 1CD ... ;) )

[Aktan]

If 200/avg quant = quality, then wouldn't it take a 200 avg quant to equal 1% quality? But of course its not possible to get an avg quant of 200. So I thought if 2 avg quant = 100% and 31 avg quant = 0% why not make the equation like this instead?:

(avg quant)(-100 / 29) + (3100/29) = (quality)

What do you think?

[MoonWalker]

Originally posted by Aktan
If 200/avg quant = quality, then wouldn't it take a 200 avg quant to equal 1% quality? But of course its not possible to get an avg quant of 200. So I thought if 2 avg quant = 100% and 31 avg quant = 0% why not make the equation like this instead?:

(avg quant)(-100 / 29) + (3100/29) = (quality)

What do you think?


Hm..It would be logical,but whats the meaning of quality 0%.And there would be an influence by the credits( I personaly encode them at quant 31)..With this the influence is not too much..But there is anohter solution..I can internaly calculate the avg quant of the movie..

MoonWalker

[Marc FD]

a quant of 31 is not 0% quality :
You can have it worse !! use a 10 kbps bitrate, 1 fps, 1400x960 and sharpen it to death : you'll get the uglier result you seen in your life ! (even if the quantizers are not all 31)

[MoonWalker]

Originally posted by Marc FD
a quant of 31 is not 0% quality :
You can have it worse !! use a 10 kbps bitrate, 1 fps, 1400x960 and sharpen it to death : you'll get the uglier result you seen in your life ! (even if the quantizers are not all 31)

That was I thining..That 0% quality is relative..But it could be a blind quality(ie 100% quant:2, 0% quant:31)

BTW I looked at the code and saw tha average quantizer is derived form the quant of the movie only and not the credtis...

MoonWalker

[Aktan]

Originally posted by Marc FD
a quant of 31 is not 0% quality :
You can have it worse !! use a 10 kbps bitrate, 1 fps, 1400x960 and sharpen it to death : you'll get the uglier result you seen in your life ! (even if the quantizers are not all 31)

But doesn't that apply to the 200/ avg quant formula too? Yes I know its all blind look at quality and does not really tell quality but i thought it would be more logical thats all :)

[Marc FD]

yep, exactly :)
just trust you eyes !
(but it can be usefull to compare two very similar encodes to keep the best)