JXL: Initial support#2463
Conversation
| if (FromHeif) { | ||
| int32u Size; | ||
| Get_B4 (Size, "Exif header length"); | ||
| Get_B4 (Size, "Tiff header offset"); |
There was a problem hiding this comment.
That part was previously added in 8a2c57c then was modified by me to parse more files that have zero offset. Looks like that identifier is not part of specs? Maybe some other files may have offset other than 0 or 6?
There was a problem hiding this comment.
Great to have a source, let's go with that.
The specs seems to tolerate something else that Exif\0\0, no strong preference but I would keep the check on the value until we find any other value. too strict?
666aa16 to
66e0f78
Compare
Application name has |
Argh, everyone does as it wants, not easy to have a good rule... |
I simply picked one from the link posted in JPEG XL issue -> https://people.csail.mit.edu/ericchan/hdr/jxl_images/20140606_102418_IMGP0297.jxl |
|
The XML is brotli compressed in this one. |
1b43ba9 to
0af26ec
Compare
|
@JeromeMartinez I am not sure what you plan to do with the Brotli compressed data in JXL. I just did a quick implementation for MSVC only just for testing to see what's in there. |
I am very interested, but I would like something similar to zlib, with using system libs on Linux and (patched, with specific build projects if needed) on Win & Mac. |
c3cf290 to
feb2fcb
Compare
|
@JeromeMartinez I think I got MSVC project done. Cannot figure out CMake and no idea about MacOS or Linux makefiles. |
|
@cjee21 thank you for your work. We'll try on our side to manage the other makefiles. |
443950a to
bacc8d5
Compare
|
Not sure if I have done it the right way but CMake also works now. |
thank you! |
141cda8 to
de97b00
Compare
|
Not sure if the Configure and CMake is done properly or the right way (the CMake looks non-ideal due to brotli upstream CMake seems to be missing something) but it does work in the CI including in MediaInfo CLI/GUI CI. If this is fine then we can merge brotli PR first then edit this PR (since it is now using my branch for testing) then repeat for MediaInfo CLI/GUI PR. |
|
and #2463 (comment) is not fixed yet |
|
@JeromeMartinez Trace offset is strange in little endian bitstream? UPDATE: Fixed. Hope I did not break anything or cause regressions. |
a686d6f to
43d7928
Compare
|
No idea how to organize properly the JPEG XL Parser structure and also how to concat the jxlp boxes. Anyway, the current state of output: |
32637a3 to
48dfa76
Compare
9e041cf to
79cc9cc
Compare
cbfee7e to
936b4a3
Compare

No description provided.