[Ffmpeg-devel] Differences between mencoder and ffmpeg whenencoding using x264

Robert Swain robert.swain
Tue Feb 21 12:09:29 CET 2006


On Tuesday 21 February 2006 10:31, Erik Slagter wrote:
> On Mon, 2006-02-20 at 22:02 +0100, Erik Slagter wrote:
> > > diff -u, please
> >
> > FWIW, here the diff -u:
>
> Now we're on it, I think this reads even better (I hope this aligns using
> monospace font)
>
> ffmpeg                   mencoder
>
> cabac=0                  cabac=1
> ref=1                    ref=1
> deblock=0:0:0            deblock=1:0:0
> analyse=0x1:0            analyse=0x1:0x111
> me=dia                   me=hex
> subme=6                  subme=5
> brdo=0                   brdo=0
> mixed_ref=0              mixed_ref=0
> me_range=4               me_range=16
> chroma_me=0              chroma_me=1
> trellis=0                trellis=1
> 8x8dct=0                 8x8dct=0
> cqm=0                    cqm=0
> chroma_qp_offset=0       chroma_qp_offset=0
> slices=1                 slices=1
> nr=0                     nr=0
> bframes=0                bframes=0
> keyint=12                keyint=250
> keyint_min=7             keyint_min=25
> scenecut=0               scenecut=40
> rc=cbr                   rc=cbr
> bitrate=3000             bitrate=3000
> ratetol=1.3              ratetol=1.0
> rceq='tex^qComp'         rceq='blurCplx^(1-qComp)'
> qcomp=0.50               qcomp=0.60
> qpmin=2                  qpmin=10
> qpmax=31                 qpmax=51
> qpstep=3                 qpstep=2
> vbv_maxrate=9000         vbv_maxrate=9000
> vbv_bufsize=65536        vbv_bufsize=8000
> ip_ratio=1.25            ip_ratio=1.40

So...

-ac -loop -partitions parti4x4+partp8x8+partb8x8 -me_method 7 -subq 5 
-me_range 16 -chroma -trellis 1 -gop_size 250 -keyint_min 25 -sc_threshold 40 
-bit_rate_tolerance 3000 -rc_eq 'blurCplx^(1-qComp)' -qcompress 0.60 -qmin 10 
-qmax 51 -max_qdiff 2 -rc_buffer_size 8000000 -i_quant_factor 0.71

...should bring them into line. me_method 7 corresponds to hex. I think qpstep 
is 4 in the x264 cli. ip_ratio = 1/abs(i_quant_factor). The rest of the 
differences you should be able to see. I think these command line options are 
correct but I may be wrong. Also, I don't know if the mencoder options are 
synced to the x264 cli precisely. They're pretty close though.

Rob





More information about the ffmpeg-devel mailing list