-o
string
-- Output file. Similarly, only
named files, no standard output.
-soft
integer
-- Soften images by zeroing N
high-frequency levels. For example, "soft 3" will get rid of blocks at
levels 1, 2 and 3 (see the diagram), i.e. blocks
from 0 to 20.
-sharp
integer
-- Sharpen images by doubling N
high-frequency levels.
Doubling the values is done by simply doubling the bin widths. Note
that more control can be exersized by using the
-sf option
Example: "sharp 2" will double the values in
blocks of level 1 and 2, i.e. blocks from 0 to 11.
-edge
integer
-- Edge detection by preserving N
levels of
high-frequency luminance blocks. Example: "edge 3" will zero all
blocks except luminance blocks at levels 1, 2 and 3, i.e. blocks 0,3,6
and 9. At the same, the subbands which remain are amplified by the
factor of 2.
-gray -- Convert to
gray scale (kill colors). This will zero all chrominance blocks.
-fl2
-- Double the fields. In each frame field
2 (upper field) is thrown away, and field 1 with modified header is
placed in the bitstream instead.
-fdecim
integer
-- Decimate frames (implies field
doubling). Example: "fdecim 3" means that instead of frames
1,2,3,4,5,6,7,8,9,... the sequence will be 1,1,1,4,4,4,7,7,7,....,
i.e. out of each 3 frames, the first one will be repeated 3 times,
while the other two will be thrown away.
-nframes
integer
-- Number of frames to use. If this
argument is not specified, all complete frames are processed. The tool
assumes that the last frame is always incomplete since it doesn't
verify the consistency of data, only the headers.
-prog
integer
-- Progressive transmission, N
frames per update. In this mode each frame is transmitted in 14
updates, 3 blocks -- one for each component -- per update. Each update
happens after the specified number of frames. Note that "prog N"
implies the frame decimation with the factor of 14*N.
-sf
string
-- Scale the bin widths by scales
from named
file. The file must contain 42 floating point number, one for each
block starting with block 0. The tool ignores
the rest of the file, so it can filled with comments or other information.
There is no formatting: blanks, tabs and new lines are all treated as
white space. A file containt 42 1's will cause no change in the
bistream. As an example, one can implement "edge 2" using "sf foo"
where file "foo" contains "1 0 0 1 0 0 1 0 0 1 0 0 0 0 0 0 ...." (the
rest are all zeroes).
-ber
float
-- Flip random bits (bits per
million).
The tool simply determines the blocks in which errors would happen. It
then zeroes corresponding blocks as would happen in a system with
simple error detection.
-berh
float
-- Same as "ber", but preserve 3
low-pass blocks. This simulates systems where extra care is take to
recover from errors in those small blocks.
-bseed
integer
-- Random number generator's seed.
The tool uses the same seed (for reproducibility of the results) when
generating
random bit errors, but this option can be used to try different seeds.
-repl
integer
-- Replicate file for better
viewing.
Some clips (usually one second long or shorter) are too short for convenient
viewing. The tool creates a longer clip in which the original clip is
simply repetead the specified number of times.
-erf
string
-- File name for error log.
In badly corrupted bit streams there might be too many errors to be
readable on a computer screen. Using the "erf" option, one can put
them all in a log file for later viewing.
-pal
-- PAL format (default NTSC). This only
effects the bit rate reporting.
-bwrep
string
-- Bin Width Report File. If this
file is specifed, the tool will print the binwidth of each block into
the file.
-crep
string
-- Compression Report File. If this
file is specifed, the tool will log the compression ratios and bit
rates for each of the fields in the clip.
-rbsrep
string
-- Relative Block Size REPort. File name to which the report about
block size (as percent of the total fields size) is written.
-blue
float
-- scale blue (Cb) component by the factor specified.
-red
float
-- scale red (Cr) component by the factor specified.
Note: Specifying "-blue 0 -red -0" is equivalent to "gray"
-fade
int
-- gradually dissolve the picture into solid gray. The numeric
argument must specify the number of frames during which the fading is
done. The frames after that number are discarded, i.e. "-fade 120"
implies "-nframes 120".
Example:
"prog 5" means that only one out of each 70 frames will be displayed,
and it will be modified every 5 frames, i.e. each 1/6 of a second.
First only level 6 blocks of frame 1 will be displayed for 1/6 of a
second, then
levels 6 and 5 for the next 1/6 of a second, etc. After 7/3 seconds,
when the whole frame 1 is displayed in full, the video will switch to
level 6 blocks of frame 71, and so on.
Block level designation in the stylized Mallat diagram:
