xinxinlx / openjpeg

Automatically exported from code.google.com/p/openjpeg
Other
0 stars 0 forks source link

e2_colr/a4_colr/b2_mono/a6_mono_colr is failing the test suite #289

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
e2_colr is failing the test suite as can be seen:

http://my.cdash.org/testDetails.php?test=14097705&build=580755

******Parameters********* 
 base_filename = /home/voxxl/Dashboards/MyTests/data/baseline/conformance/e2_colr.ppm
 test_filename = /home/voxxl/Dashboards/MyTests/openjpeg-continuous-linux_gcc47-trunk-debug/tests/conformance/Temporary/e2_colr.j2c.ppm
 nb of Components = 1
 Non regression test = 0
 separator Base = 
 separator Test = 
 MSE values = [ 1.000000 ]
 PEAK values = [ 4.000000 ]
 Non-regression test = 0
 NbFilename to generate from base filename = 0
 NbFilename to generate from test filename = 0
************************* 
Step 1 -> Header comparison
ERROR: height mismatch [comp 0] (594><149)

Original issue reported on code.google.com by mathieu.malaterre on 13 Mar 2014 at 3:47

GoogleCodeExporter commented 9 years ago
The baseline is odd-shaped:

$ head -4 data/baseline/conformance/e2_colr.ppm
P3
# Copyright (c) 2000-2003 Pegasus Digital Imaging
1023 594
255

However, the codestream is declared as:

[...]
2       : New marker: SIZ (Image and tile size)

  Required Capabilities          : JPEG2000 full standard
  Reference Grid Size            : 1025x601
  Image Offset                   : 2x7
  Reference Tile Size            : 511x793
  Reference Tile Offset          : 2x7
  Components                     : 3
  Component #0 Depth             : 8
  Component #0 Signed            : no
  Component #0 Sample Separation : 4x4
  Component #1 Depth             : 8
  Component #1 Signed            : no
  Component #1 Sample Separation : 4x4
  Component #2 Depth             : 8
  Component #2 Signed            : no
  Component #2 Sample Separation : 4x4
[...]

Original comment by mathieu.malaterre on 13 Mar 2014 at 3:48

GoogleCodeExporter commented 9 years ago

Original comment by mathieu.malaterre on 14 Mar 2014 at 11:29

GoogleCodeExporter commented 9 years ago

Original comment by mathieu.malaterre on 14 Mar 2014 at 11:45

GoogleCodeExporter commented 9 years ago
This issue was updated by revision r2737.

Original comment by mathieu.malaterre on 14 Mar 2014 at 12:12

GoogleCodeExporter commented 9 years ago
Issue 279 has been merged into this issue.

Original comment by mathieu.malaterre on 14 Mar 2014 at 1:04

GoogleCodeExporter commented 9 years ago

Original comment by mathieu.malaterre on 14 Mar 2014 at 2:05

GoogleCodeExporter commented 9 years ago

Original comment by antonin on 24 Mar 2014 at 3:57

GoogleCodeExporter commented 9 years ago

Original comment by mathieu.malaterre on 25 Mar 2014 at 10:38

GoogleCodeExporter commented 9 years ago
@mathieu, @antonin,

an option to compensate for dx/dy in opj_decompress would probably be the best 
way to pass these tests.

Any thoughts ?

Original comment by m.darb...@gmail.com on 22 Nov 2014 at 12:54

GoogleCodeExporter commented 9 years ago
This issue was closed by revision r2942.

Original comment by m.darb...@gmail.com on 22 Nov 2014 at 2:09

GoogleCodeExporter commented 9 years ago
wrong issue number in commit, sorry

Original comment by m.darb...@gmail.com on 22 Nov 2014 at 2:10

GoogleCodeExporter commented 9 years ago

Original comment by m.darb...@gmail.com on 22 Nov 2014 at 2:29

GoogleCodeExporter commented 9 years ago
This issue was updated by revision r2943.

Original comment by m.darb...@gmail.com on 22 Nov 2014 at 5:55

GoogleCodeExporter commented 9 years ago
This issue was updated by revision r2944.

Issue is left open since there's still one image failing the tests.

Original comment by m.darb...@gmail.com on 22 Nov 2014 at 5:59

GoogleCodeExporter commented 9 years ago
commit r2944 deals with up sampling image components to image size.

"  -upsample\n"
"    Downsampled components will be upsampled to image size\n"

This gives a good example of how image/component offsets to reference grid work 
in case of sub-sampling (this was not intuitive for me).

Now that all components are of the same size with all dx/dy == 1, a6_mono_colr 
is output as a single ppm image (imagetopnm assumes rgba) so tests are still 
failing.

Original comment by m.darb...@gmail.com on 22 Nov 2014 at 6:15

GoogleCodeExporter commented 9 years ago
@mathieu,

I think we can add yet another option to opj_decompress to force component 
splitting in imagetopnm.
Any thoughts ?

Original comment by m.darb...@gmail.com on 3 Dec 2014 at 8:28