JPEG XS essence support

[nishabfhg]

Hi,

Here are my changes to include JPEG XS essence support into bmx. It works pretty much like JPEG 2000. Only PARSE_FRAME_START_SIZE has to be larger in RawEssenceReader. Let me know if I need to modify something.

Thank you. Best regards, Nisha

[philipnbbc]

Thanks! I don't have time available today to review but will try make a start tomorrow or Wednesday. Is there a sample available somewhere online I can use to test?

[nishabfhg]

Great! I don't know if there are any jxs samples available online. I have added a short RGB test sequence here: https://github.com/nishabfhg/bmx-jxs/tree/main/Blender_FullHD_Testsequence. In raw2bmx, you will need the jxs_rgba command line option in this case.

[philipnbbc]

Here is the list of warnings from the Linux build:

/home/runner/work/bmx/bmx/src/essence_parser/FilePatternEssenceSource.cpp: In member function ‘int64_t bmx::FilePatternEssenceSource::ReadFileSize(uint32_t)’:
/home/runner/work/bmx/bmx/src/essence_parser/FilePatternEssenceSource.cpp:208:14: warning: unused variable ‘res’ [-Wunused-variable]
  208 |         bool res = NextFile();
      |              ^~~

In file included from /home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:33:
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp: In member function ‘virtual void bmx::JXSEssenceParser::ParseFrameInfo(const unsigned char*, uint32_t)’:
/home/runner/work/bmx/bmx/include/bmx/essence_parser/JXSEssenceParser.h:49:34: warning: comparison of unsigned expression in ‘< 0’ is always false [-Wtype-limits]
   49 |         #define SUCCESS(v) (((v) < 0) ? 0 : 1)
      |                              ~~~~^~~
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:144:29: note: in expansion of macro ‘SUCCESS’
  144 |         while (p < end_p && SUCCESS(result))
      |                             ^~~~~~~
/home/runner/work/bmx/bmx/include/bmx/essence_parser/JXSEssenceParser.h:50:34: warning: comparison of unsigned expression in ‘< 0’ is always false [-Wtype-limits]
   50 |         #define FAILURE(v) (((v) < 0) ? 1 : 0)
      |                              ~~~~^~~
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:148:21: note: in expansion of macro ‘FAILURE’
  148 |                 if (FAILURE(result))
      |                     ^~~~~~~
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:211:66: warning: comparison is always false due to limited range of data type [-Wtype-limits]
  211 |                                 if (PIH_.Hsl() < 1 || PIH_.Hsl() > 0xffff) // This includes the EOF check
      |                                                       ~~~~~~~~~~~^~~~~~~~
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:154:24: warning: enumeration value ‘MRK_NIL’ not handled in switch [-Wswitch]
  154 |                 switch (NextMarker.m_Type)
      |                        ^
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:154:24: warning: enumeration value ‘MRK_WGT’ not handled in switch [-Wswitch]
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:154:24: warning: enumeration value ‘MRK_COM’ not handled in switch [-Wswitch]
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:154:24: warning: enumeration value ‘MRK_NLT’ not handled in switch [-Wswitch]
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:154:24: warning: enumeration value ‘MRK_CWD’ not handled in switch [-Wswitch]
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:154:24: warning: enumeration value ‘MRK_CTS’ not handled in switch [-Wswitch]
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:154:24: warning: enumeration value ‘MRK_CRG’ not handled in switch [-Wswitch]
/home/runner/work/bmx/bmx/src/essence_parser/JXSEssenceParser.cpp:154:24: warning: enumeration value ‘MRK_CAP’ not handled in switch [-Wswitch]

/home/runner/work/bmx/bmx/apps/bmxtranswrap/bmxtranswrap.cpp: In function ‘int main(int, const char**)’:
/home/runner/work/bmx/bmx/apps/bmxtranswrap/bmxtranswrap.cpp:3967:20: warning: enumeration value ‘JPEGXS_CDCI’ not handled in switch [-Wswitch]
 3967 |             switch (output_essence_type)
      |                    ^
/home/runner/work/bmx/bmx/apps/bmxtranswrap/bmxtranswrap.cpp:3967:20: warning: enumeration value ‘JPEGXS_RGBA’ not handled in switch [-Wswitch]

[philipnbbc]

As promised, here is the process for creating regression tests for JXS.

More info on testing can be found here. This process assumes you only have an RGBA example. If you also have a CDCI example file then ignore the steps below that delete that particular part of the tests; you will have 4 test data files instead of 2.

• Copy tests/jpeg2000 to tests/jpegxs
• Remove the test inputs and data: .j2c .bin and .md5 files
• Copy in file 00000 and 00003 from Blender_FullHD_Testsequence/ (i.e. leave a gap which will be filled in using the --fill-pattern-gaps option in the tests)
• Modify tests/CMakeLists.txt
  • Add a add_subdirectory(jpegxs) after the jpeg2000 inclusion
• Modify .gitattributes
  • to have only 2 lines for _1 and _2. These test data files will be generated later
• Modify tests/jpegxs/CMakeLists.txt
  • Remove cdci from the tests list
  • Replace jpeg2000 with jpegxs
• Modify tests/jpegxs/test_rgba.cmake
  • Change the _3 and _4 in names to _1 and _2
  • Change the --j2c_rgba arg to --jxs_rgba "${TEST_SOURCE_DIR}/Blender_Sequence_FullHD_1920x1080_8b_444_%d.jxs"
  • Remove --aspect-ratio and --frame-layout (which will be parsed from the bitstream) and other args like --transfer-ch as required that don't match the content
• Rebuild using cmake --build .
• Generate sample files for the test using cmake --build . --target bmx_jpegxs_rgba_data
  • the files will be output to the test_samples/jpegxs/ sub-directory in the build directory
  • Check that the sample files look ok. The .bin files are the XML output from mxf2raw and the .mxf files are generated by raw2bmx and bmxtranswrap.
• Generate the test data (.bin and .md5) files using cmake --build . --target bmx_jpegxs_rgba_data
• Run the tests using ctest to check the new JXS tests are run and pass
• Remove the Blender_FullHD_Testsequence directory

[nishabfhg]

Hi Philip, thanks for the review! I will try to make all the modifications next week and get back to you.

[nishabfhg]

Hi Philip, I have modified the code according to your review points. I also added a regression test for jpegxs cdci. Let me know if there's anything else. (I will only be able to get back to this after the New Year holidays)

[philipnbbc]

I've had a quick look through and looks like you covered everything, thanks for the updates! I'll probably take a proper look sometime next week as I'm away from work.

[nishabfhg]

Hi Philip. I made the fixes. I hope I didn't miss anything. I synced my version with your latest commits, so the merge conflicts should not appear anymore.

[philipnbbc]

All done now except for some unexpected changes to some non-jxs test checksums - see below the diff showing what needs to be changed. Do you know what caused this? Just asking as it might be an issue that needs fixing.

If you run cmake --build . -t bmx_test_data and cmake --build . -t libMXF_test_data it should regenerate the checksums as bbelow. cmake --build . -t test should then pass.

Diff showing what needs to be changed back:

diff --git a/deps/libMXF/test/test_indextable.md5 b/deps/libMXF/test/test_indextable.md5
index e781305e..6dc5d47a 100644
--- a/deps/libMXF/test/test_indextable.md5
+++ b/deps/libMXF/test/test_indextable.md5
@@ -1 +1 @@
-d41d8cd98f00b204e9800998ecf8427e
\ No newline at end of file
+5ff6163c7004300edb5e0c6670e547d3
\ No newline at end of file
diff --git a/test/as02/iecdv25.md5s b/test/as02/iecdv25.md5s
index ec6da267..8e831eb3 100644
--- a/test/as02/iecdv25.md5s
+++ b/test/as02/iecdv25.md5s
@@ -1 +1 @@
-abce8e563ec3408a0dbf10ae34dfe30c;ca114c9f2707afde489d6e6066f8833c;e4a2bdff4418802e3d127654ee906dc6;99538bb806a99596b95236ac8a965b52
\ No newline at end of file
+227a7e8a6b34571cfb3a9b70642c0183;4e4605266cb9d916830ea1eef722321d;d8b021e1b8bc3bd3796676b0b5e8f594;16ff89dda611b560eda0739fd90d0214
\ No newline at end of file
diff --git a/test/avid_mxf/mpeg2lg_422p_hl_1080i.md5s b/test/avid_mxf/mpeg2lg_422p_hl_1080i.md5s
index 606895d7..8e1e6d3e 100644
--- a/test/avid_mxf/mpeg2lg_422p_hl_1080i.md5s
+++ b/test/avid_mxf/mpeg2lg_422p_hl_1080i.md5s
@@ -1 +1 @@
-f95eae8d685f5c222c7e3e9fb84b9f9a;419a18c68d4b87dfb10408653a6bc58e;805d6fd84c1ed39585053c49016f52f7
\ No newline at end of file
+f189845c35fbe0a69d36b181d85937f9;0ece8877ab256657df705de6a01a8a48;f323a74f1505abc8bf201eeef07e0089
\ No newline at end of file
diff --git a/test/avid_mxf/mpeg2lg_mp_hl_1920_1080i.md5s b/test/avid_mxf/mpeg2lg_mp_hl_1920_1080i.md5s
index 2b2e3f30..793a0a50 100644
--- a/test/avid_mxf/mpeg2lg_mp_hl_1920_1080i.md5s
+++ b/test/avid_mxf/mpeg2lg_mp_hl_1920_1080i.md5s
@@ -1 +1 @@
-93d9e5755cdcdd13000b71bcad7d55d1;a31f0491c9255156d07ec78ac265311c;f3259501cd9d48ce94f5f1aa0006d697
\ No newline at end of file
+fbd8e5365fbd58c4a4908cd2e78319f7;0ece8877ab256657df705de6a01a8a48;f323a74f1505abc8bf201eeef07e0089
\ No newline at end of file
diff --git a/test/d10_mxf/d10_40_2997.md5 b/test/d10_mxf/d10_40_2997.md5
index 20d6158e..8563ad8d 100644
--- a/test/d10_mxf/d10_40_2997.md5
+++ b/test/d10_mxf/d10_40_2997.md5
@@ -1 +1 @@
-8a256f169a5cbaa6145f95a9d008df0a
\ No newline at end of file
+6e62f6adec11065622203c7e75cc2b38
\ No newline at end of file

[nishabfhg]

The md5s might have changed when I rebuilt the project using the visual studio solution. I am not sure why this happened.

[nishabfhg]

That's great! Thanks for the review.