[XviD-devel] Re: [Ffmpeg-devel] Trellis quant (fwd)
Christoph Lampert
chl at math.uni-bonn.de
Fri Apr 18 23:27:36 CEST 2003
Hi,
as I assumed: ffmpeg quant does coefficient reduction for Pframes
by default. With trellis this is replaced by R-D-optimization, so I-frames
get smaller with Trellis, P-frames get larger.
gruel
----------------------------------------------------------
To: ffmpeg-devel at lists.sourceforge.net
Subject: Re: [Ffmpeg-devel] Trellis quant
Hi
On Friday 18 April 2003 19:04, Christoph Lampert wrote:
> Hi,
>
> I'm confused about Trellis-Quantization (or just too stupid to understand
> the code):
>
> My impression was that trellis quant sometimes lowers values of quantized
> DCT coefficients in order to safe bits (using a lagrangian).
> Therefore fixed quant files with trellis quant should become smaller at
> a reduced PSNR compared to encoding at the same quant without trellis.
>
> But the opposite is true: With trellis, my test files got _bigger_ at a
> _higher_ PSNR, but still at fixed quant. From what I see, I-VOPs indeed
> do get smaller, but all P-VOPs get larger.
>
> How is that possible? Or is there some other less efficient
> thresholding mechanism enabled by default which I overlooked?
the default quantization code adds a bias (-1/4 for P and 0 for I) like some
h263 doc recommands, u can change the bias with the ibias & pbias options
(for mencoder)
[...]
--
Michael
screen[y][x] ^= cursor[cy][cx]; (violates patent #4,197,590)
median(mv[y-1][x], mv[y][x-1], mv[y+1][x+1]); (violates patent #5,905,535)
buf[i]= qp - buf[i-1]; (violates patent #?)
for more examples, see http://mplayerhq.hu/~michael/patent.html
stop it, see http://petition.eurolinux.org & http://petition.ffii.org/eubsa/en
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Ffmpeg-devel mailing list
Ffmpeg-devel at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ffmpeg-devel
More information about the XviD-devel
mailing list