yuvscaler.1

Motion JPEG编解码器源代码

.TH "yuvscaler" "1" "26 Septembre 2001" "MJPEG Linux Square" "MJPEG tools manual"

.SH NAME
yuvscaler \- UPscales or DOWNscales a YUV4MPEG2 stream to a specified format

.SH SYNOPSIS
.B yuvscaler 
.RB [ -I
.IR input_keyword ]
.RB [ -M 
.IR mode_keyword ]
.RB [ -O
.IR output_keyword ]
.RB [ -n\ p | s | n ]
.RB [ -v\ 0 - 2 ]
.RB [ -h ]

.SH DESCRIPTION
\fByuvscaler\fP is a general (UP or DOWN) scaling utility that is provided to
allow conversion between different video formats. It reads in a
YUV4MPEG2 format stream from stdin (such as that produced
by \fBlav2yuv\fP) and outputs such stream to stdout (e.g. to be
consumed by \fBmpeg2enc\fP).
The most common use of yuvscaler is to generate input suitable
for MPEG encoding to VCD or SVCD formats using \fBmpeg2enc\fP(1).

\fByuvscaler\fP will automatically take into account PAL/SECAM and
NTSC format, as well as widescreen (16:9) format and interlacing. Use
of yuvscaler was designed to be straightforward.

.SH EXAMPLES

\fBVCD encoding:\fP
.br
lav2yuv my-file.avi | yuvscaler -O VCD | mpeg2enc -f 1 -o vcd.%d.m1v

\fBSVCD encoding:\fP
.br
lav2yuv my-file.qt | yuvscaler -O SVCD | mpeg2enc -f 4 -o svcd.%d.m2v

\fBEncoding anamorphic 16:9 as letterboxed VCD:\fP
.br
lav2yuv my-ana16-9.eli | yuvscaler -M WIDE2VCD | mpeg2enc -f 1 -o letterbox_vcd.%d.m1v

\fBEncoding anamorphic 16:9 as letterboxed SVCD:\fP
.br
lav2yuv my-ana16-9.eli | yuvscaler -M WIDE2STD -O SVCD | mpeg2enc -f 4 -o letterbox_svcd.%d.m2v

\fBEncoding anamorphic 16:9 as letterboxed DVD:\fP
.br
lav2yuv my-ana16-9.eli | yuvscaler -M WIDE2STD -O DVD | mpeg2enc -f 8 -o letterbox_dvd.%d.m2v

.SH OPTIONS

.TP 
.B -I \fIinput_keyword\fP
.IP
Specifies input processing options.  Input options may be combined by using
multiple \fB-I\fP flags. Possible \fIinput_keyword\fPs are:

.IP
\fBUSE_WidthxHeight+WidthOffset+HeightOffset\fP to select a useful area of
the input frame (all multiple of 2, Height and HeightOffset multiple of 4 if interlaced), the rest
of the image being discarded
.IP
\fBACTIVE_WidthxHeight+WidthOffset+HeightOffset\fP to select an active area
of the input frame (all multiple of 2, Height and HeightOffset multiple of 4 if interlaced), the rest
of the image being made black

.TP 
.B -M \fImode_keyword\fP
.IP
Sets the scaling mode options.  Scaling mode options can be combined by giving several
\fB-M\fP flags.
Possible  are:

.IP
\fBBICUBIC\fP - to use the (Mitchell-Netravalli) high-quality bicubic upsacling
and/or downscaling algorithm
.IP
\fBRESAMPLE\fP - to use a classical resampling algorithm -only for
downscaling- that goes much faster than bicubic. For coherence reason,
\fByuvscaler\fP will use RESAMPLE if only downscaling is necessary,
BICUBIC if not.
.IP
\fBWIDE2STD\fP - convert widescreen anamorphic (16:9) input video
into letterboxed (4:3), generating necessary black lines.  Note that
for MPEG applications if your \fIplayer\fP can properly decode
anamorphic video you may wish to leave the video unscaled and simply
specify the approprate format during MPEG encoding (see \fBmpeg2enc\fP(1)).
.IP
\fBRATIO_WidthIn_WidthOut_HeightIn_HeightOut\fP to specified
scaling ratios of WidthIn/WidthOut for width and
HeightIN/HeightOut for height to be applied to the useful area. The
output frames that result from scaling can have a different format
than the output format specified using \fB-O\fP options(s).  In this
case, \fByuvscaler\fP automatically generate necessary black lines and
columns and/or skips necessary lines and columns to center the scaled
image in the output frame.
.IP
\fBWIDE2VCD\fP to scale anamorphic widescreen (16:9) frames into letterboxed 
VCD format (equivalent to -M WIDE2STD -O VCD).  
.IP
\fBFASTVCD\fP to transcode full sized frames to VCD 
(equivalent to -M RATIO_2_1_2_1 -O VCD, see below).
.IP
\fBFAST_WIDE2VCD\fP to transcode full-sized anamorphic (16:9) frames to 
letter-boxed VCD (equivalent to -M WIDE2STD -M RATIO_2_1_2_1 -O VCD, see below).
.IP
\fBNO_HEADER\fP to suppress stream header generation on output
.IP
\fBMMX\fP to use MMX functions for BICUBIC scaling (experimental feature!!)
.IP

.TP 
.B -O \fIoutput_keyword\fP

.IP
Sets the output frame format options.  Output format
options can be combined by giving several \fB-O\fP flags.
By default, output frames size will be the same as input frame sizes, as
well as output frames interlacing if coherent with output keyword
specification which prevails, top_first if not. 

.IP
\fBMONOCHROME\fP to generate monochrome frames on output.  Chrominance
information for the scaled pixels is set neutral (white).  MPEG
encoding of black and white video will be both inefficient \fIand\fP
show annonying chrominance artifacts unless this is done.
.IP
\fBVCD\fP to generate VCD compliant frames on output (taking care of
PAL and NTSC standards). Non-interlaced output.
.IP
\fBSVCD\fP to generate SVCD compliant frames on output (taking care of
PAL and NTSC standards). Top-field-first interlaced frames.
.IP
\fBDVD\fP to generate DVD compliant frames on output (taking care of
PAL and NTSC standards). Top-field-first interlaced frames.
.IP
\fBSIZE_WidthxHeight\fP to generate frames of size WidthxHeight on
output (multiple of 4)
.IP
If VCD, SVCD or DVD keywords are not used, output interlacing is taken of the
same type as input. By default, output frames size will be the same as input
frame size.
.IP

.TP 
.B -n p|s|n
(usually not necessary) if norm could not be determined from data flux, specifies the OUTPUT norm
for VCD/SVCD (p=pal,s=secam,n=ntsc).

.TP
.B -v 0|1|2
Specifies the degree of verbosity: 0=quiet, 1=normal, 2=verbose/debug

.TP
.B -h
Prints a summary of command line syntax.



.PP
\fBNote:\fP The most useful downscaling ratios are handled using dedicated optimised
scaling routines. They are:
.IP
RATIO_WidthIn_WidthOut_2_1 => Full-sized frame to VCD downscaling
.IP
RATIO_3_2_1_1 => Full-sized frame to SVCD downscaling
.IP
RATIO_1_1_4_3 => WIDE2STD downscaling mode
.IP
RATIO_WidthIn_WidthOut_1_1 => downscaling only concerns width, not height
.IP
RATIO_1_1_HeightIn_HeightOut => downscaling only concerns height, not width
.IP
RATIO_2_1_2_1 => FASTVCD, slightly width distorted (real ratio 45 to 22) but faster
full-sized frame to VCD downscaling (-M RATIO_2_1_2_1 -O VCD)
.IP
RATIO_WidthIn_WidthOut_8_3 => specific to WIDE2VCD downscaling (-M WIDE2STD -O VCD)
.IP
RATIO_2_1_8_3 => specific to (slighly distorted) FAST_WIDE2VCD downscaling (-M WIDE2STD -M RATIO_2_1_2_1 -O VCD)
.IP
RATIO_1_1_1_1 => copy useful input part of possible several files into output frames


.SH AUTHOR
This man page was written by Xavier Biquard et al.
.br
If you have questions, remarks, problems or you just want to contact
the developers, the main mailing list for the MJPEG\-tools is:
  \fImjpeg\[e-mail]\fP

.TP
For more info, see our website at
.I http://mjpeg.sourceforge.net/

.SH SEE ALSO
.BR mjpegtools (1),
.BR lavpipe (1),
.BR lav2yuv (1)
.BR mpeg2enc (1)