Log in

View Full Version : Interpreting the debugview output


Bulletproof
25th June 2002, 09:35
Could somebody please post an explination as to what each variable that debugview outputs represents? I remember doom9 once did this for Nandub a while ago explaining what each set of numbers represented.

canadian_fbi
25th June 2002, 15:08
well i'm no expert on this, but i'm pretty sure i can tell you what the second pass variables are:


00210386 7982.07292882 [2892] 2nd-pass: quant:3 H.263 inter stats1:4781 scaled:2833 actual:3768 overflow:144 movie
--------------------------------------------------------------------------------------------------------------------
A B C D E F G H I J K L

A, B: debugview stuff telling you the sequential number of the debug output and the time it was recorded.
C: process id of the process creating the output, i think. i never used this number anyway.
D: first or second pass of the encode.
E: the quantizer assigned to this frame.
F: the quantization type used in this frame.
G: the type of frame (inter = p-frame, intra = i-frame).
H: the first-pass size of this frame.
I: the precise scaled size of the frame as determined by the curve compression.
J: the actual size of the frame as encoded at the quantizer listed previously.
K: the total overflow so far (i believe a positive number here means that the movie is that many bytes lower than the total scaled value so far says it should be.)
L: whether or not this frame is in the movie or the credits.

i hope this helps (and i hope this is right). :)

MoonWalker
25th June 2002, 15:56
Yes that it's correct...If you want some usefull(in my opinion) statistics about the logfile download my proggie :p

Bulletproof
26th June 2002, 03:49
What does a line like this mean:

[3020] P-frame quantizer prevented from rising too steeply

And could you explain the overflow a bit more, I'm still sorta confused about it.

gldblade
26th June 2002, 04:00
That means that the codec prevented quantizers from changing too quickly. For example, XviD will prevent the quantizer from jumping from 2 to 5, because it's more than 2 quantizer away and thus too big a jump.

Bulletproof
26th June 2002, 04:40
I see, thanks. Can you please elaborate a bit more on the overflow?

Bulletproof
26th June 2002, 05:20
And these too please:

[3020] avg scaled framesize:3565

I assume that is self-explanitory, but is that just the scale, or after the quantizer used also?

[3020] bias bonus:499 bytes

bias bonus is adding 499 bytes per frame?

[3020] quant:2 threshold at 331 : -90 percent

what is the threshold representing? and what is the percentage representing and why is it a negative value?

gldblade
26th June 2002, 06:08
Hmm... Never really seen those before.

Bulletproof
26th June 2002, 06:15
They actually pop up as the very first lines when debugview starts grabbing the input. Scroll up to the very beginning in debugview.

Bulletproof
27th June 2002, 07:03
Could -h or moonwalker please help out in explaining those please.

MoonWalker
29th June 2002, 13:04
Sometime ago -h has explained to me what these means..I don't remember very much but I'll try my best :)



avg scaled framesize:3565

Well this is the average scalled size of the whole movie. Remember the grapha about alt-cc??Well this is the middle line..An so when wheput at At-cc High Distance from average % : 100% We specify as a big frame all the frames tha are 3565+(100%*(3565))=7130bytes. And it's the scaled size(because no quantization has occured)


bias bonus:499 bytes

I can't quite remember sorry :( (Koepi,Nic,-h ??? :p :p )

quant:2 threshold at 331 : -90 percent

Well I did one test today and I got this..
[3612] quant:2 threshold at 469 : -86 percent
[3612] quant:3 threshold at 1401 : -59 percent
[3612] quant:4 threshold at 7049 : 101 percent

Well as far as I can remember it's how far from quant2 will be the avg quantizer. The threshold is the size of the frame..(Again anyone to explain this better? )

MoonWalker

-h
3rd July 2002, 08:04
Could somebody please post an explination as to what each variable that debugview outputs represents? I remember doom9 once did this for Nandub a while ago explaining what each set of numbers represented.

I know you were interested in the thresholding output from the PM (not sure what else there is) - can't tell you unfortunately because I can neither use xvid.dll nor browse the source on this computer (doh). Perhaps in a couple weeks when my travelling slows down.

-h