Xanashi/Icaros

jp2 support

Closed this issue · 13 comments

I have a bunch of jp2 image files so I manually added jp2 extension to the list. However, I get black thumbnails for some of them. I can view them OK in XnViewMP. I was wondering if jpeg 2000 is fully supported.

JPEG2000 would be nice in the future, it also goes hand in hand with PDF thumbnailing since Adobe Acrobat supports that as a image codec option.

Hi 03302024,
Could you send me some the files that produces black thumbnails?

Could you send me some the files that produces black thumbnails?

Here you are: https://archive.org/download/dr_pratts-high-test-plan-of-the-eastern-counties-and-the-midlands-12471000/12471000.jp2

Thank you! I'll look into it, and get back to you.

Here is another one. It is a scan of a white page, but the thumbnail is shown in black.
test file.zip

Just a follow up for this feature request.

I looked into this and these files fail to decode with FFmpeg.
So the FFmpeg jpeg2000 decoder is where this needs to be fixed.

Thank you for looking into this issue. How we could inform FFmpeg people about this?

It's possible to report issues to the FFmpeg developers using their issue tracker:
FFmpeg Issue Tracker

They do have a few requirements for reporting bugs/requesting features, that must be read before creating a ticket:
https://ffmpeg.org/bugreports.html

OK, thank you. I created an account there to report the issue. Hope it will be fixed soon.
https://trac.ffmpeg.org/ticket/11207

I see there are bug reports over there from 14 years ago. Now, I am wondering if any other jp2 decoder could be used meanwhile?

The FFmpeg team fixes issues from that list every single day, so just because some tickets haven't been picked up by anyone in a long time, doesn't mean it's a common occurrence. Most tickets usually gets attention.

I'm not that familiar with the jp2 format, and what alternative decoders exists, but I'm hoping FFmpeg will be able to decode these files in the near future.

There is of course also a chance that these files were just incorrectly encoded, and may need to be re-encoded.

OK, thank you for looking into this.