MJPEG Tools Svn

Fixed buffer size (tempfile[256]) could overflow if a long pathname was
passed. strdup() was used for the filename itself so malloc the tempfile
name as well!

is_MJPG was very carefully set and cleared - but then NEVER used.

Only 1 use of lav_filetype() left - all the others had long since been replaced
by lav_file->format.

Added the -z/-Z options, to allow the zero-motion pass to have a
different error threshold than the motion-detection pass.

Added -v, to control verbosity. "-v 1" outputs the pixel-detection
stats per frame, like it did before. "-v 0" shuts that up.

Added -h/-?, to print a short command summary. Bad options also print
the command summary.

Updated man page to reflect these changes.

Added the new version of the script from: Nicolas Boos

Changelog:

* Added two passes encoding mode, for both ffmpeg and mpeg2enc.
* Added new "unsharp" filter type, based on the y4munsharp tool.
* Now custom quant. matrices are also used with the ffmpeg encoder.
* The decoder tool can be specified.
* Audio pass-through is now enabled via the new -c flag.
* Tweaked a bit the mvcd output profile. It now uses its own "mvcd"
matrices and the audio bitrate is 128 instead of 224 kbps.

Add @PROGRAM_NOPIC@ to the AM_CXXFLAGS. On PPC systems (OS/X at least) this
expands to -mdynamic-no-pic which generates slightly faster code.

SAD11 patch from Scott Smith

Ok - the quantization SIMD disable/enable logic added. If no typos were made
then the IA32 stuff is done. Altivec - not yet.

Oops - checked the passed name rather than the environ variable for 'all'.

let folks know that 'all' can be used.