opj_jp3d_compress(1) Works with JPEG2000 files


List of parameters for the JPEG2000 Part 10 encoder: ------------

Required Parameters (except with -h):

-i : source file (-i source.bin or source*.pgx)

-m : source characteristics file (-m imgfile.img)

-o : destination file (-o dest.jp3d)

Optional Parameters:

-h : display the help information

-n : number of resolutions (-n 3,3,3)

-I : use the irreversible transforms: ICT + DWT 9-7 (-I)

-C : coding algorithm (-C 2EB) [2EB, 3EB]

-r : different compression ratios for successive layers (-r 20,10,5)

- The rate specified for each quality level is the desired compression factor. - Rate 1 means lossless compression
(options -r and -q cannot be used together)

-q : different psnr for successive layers (-q 30,40,50)

(options -r and -q cannot be used together)

-b : size of code block (-b 32,32,32)

-c : size of precinct (-c 128,128,128)

-t : size of tile (-t 512,512,512)

-p : progression order (-p LRCP) [LRCP, RLCP, RPCL, PCRL, CPRL]

-s : subsampling factor (-s 2,2,2) [-s X,Y,Z]

- Remark: subsampling bigger than 2 can produce error

-SOP : write SOP marker before each packet

-EPH : write EPH marker after each header packet

-M : code-block style (-M 0) [1=BYPASS(LAZY) 2=RESET 4=RESTART(TERMALL)

8=VSC 16=PTERM 32=SEGSYM 64=3DCTXT] Indicate multiple modes by adding their values. ex: RESTART(4) + RESET(2) + SEGMARK(32) = -M 38

-D : define DC offset (-D 12)

-x : create an index file *.Idx (-x index_name.Idx)

-ROI : c=%d,U=%d : quantization indices upshifted

for component c=%d [%d = 0,1,2] with a value of U=%d [0 <= %d <= 37] (i.e. -ROI:c=0,U=25)

-d : offset of the origin of the volume (-d 150,300,100)

-l : offset of the origin of the tiles (-l 100,75,25)

DEFAULT CODING: ------------

* Lossless * 1 tile * Size of precinct : 2^15 x 2^15 x 2^15 (means 1 precinct) * Size of code-block : 64 x 64 x 64 * Number of resolutions in x, y and z axis: 3 * No SOP marker in the codestream * No EPH marker in the codestream * No sub-sampling in x, y or z direction * No mode switch activated * Progression order: LRCP * No index file * No ROI upshifted * No offset of the origin of the volume * No offset of the origin of the tiles * Reversible DWT 5-3 on each 2D slice * Coding algorithm: 2D-EBCOT

REMARKS: ---------

- The markers written to the main_header are : SOC SIZ COD QCD COM. - COD and QCD markers will never appear in the tile_header.

- You need enough disk space memory (twice the original) to encode the volume,i.e. for a 1.5 GB volume you need a minimum of 3GB of disk memory)

- When loading *.pgx files, a relative path to directory is needed for input argument

followed by the common prefix of the slices and a '*' character representing sequential numeration.

( -i relativepath/slices*.pgx )

- The index file has the structure below:
Image_height Image_width Image_depth Progression order: 0 (LRCP) Tiles_size_X Tiles_size_Y Tiles_size_Z Components_nb Layers_nb Decomposition_levels [Precincts_size_X_res_Nr Precincts_size_Y_res_Nr Precincts_size_Z_res_Nr]
[Precincts_size_X_res_0 Precincts_size_Y_res_0 Precincts_size_Z_res_0] Main_header_end_position Codestream_size Tile_0 [start_pos end_header end_pos TotalDisto NumPix MaxMSE] ... Tile_Nt [ '' '' '' '' '' '' ]
Tpacket_0 [Tile layer res. comp. prec. start_pos end_pos disto] ... Tpacket_Np ['' '' '' '' '' '' '' '' ] MaxDisto TotalDisto


This manual page was written by Mathieu Malaterre <[email protected]> for the Debian GNU/Linux system, but may be used by others.