K19 for rpi3

  • The deinterlace changes need to be backported onto the kernel and ffmpeg branches that are being used in LE10 releases, and then we can think about spinning a new version. The lack of HEVC support, which is not on the Pi Foundation to-do list, means we are still 50/50 on what the user reaction will be like. It probably means we will choose timing; i.e. avoid Christmas and New Year periods when we'd like a break and some downtime with families. It's not the time to invite lots of user demand for something that can't be fixed.

  • The deinterlace changes need to be backported onto the kernel and ffmpeg branches that are being used in LE10 releases, and then we can think about spinning a new version. The lack of HEVC support, which is not on the Pi Foundation to-do list, means we are still 50/50 on what the user reaction will be like. It probably means we will choose timing; i.e. avoid Christmas and New Year periods when we'd like a break and some downtime with families. It's not the time to invite lots of user demand for something that can't be fixed.

    Just one more question please.

    Are these problems addressed in LE11?

    And you guys certainly need to take a break for the holidays to be with your family and friends.

    Thank you all for your hard work.

  • The three missing things (holy trinity?) are/were:

    • Hardware deinterlace
    • 3D support
    • Optimised software HEVC support

    Deinterlace is done. 3D is possible and not on the Foundations to-do list, but one of their developers is likes 3D so it will probably happen from personal interest at some point. Software optimisation for HEVC is not likely to be redone because it's a large effort and requires a level of hackery in ffmpeg that will never be accepted upstream, which is the design requirement for all code now. Tests show the non-optimised HEVC support in LE11 images is sufficient for SD media, but not 720p/1080p. So if you really need HD media, either stick with LE9.2 or bump up to an RPi4 with native HEVC hardware decode.

  • The three missing things (holy trinity?) are/were:

    • Hardware deinterlace
    • 3D support
    • Optimised software HEVC support

    Deinterlace is done. 3D is possible and not on the Foundations to-do list, but one of their developers is likes 3D so it will probably happen from personal interest at some point. Software optimisation for HEVC is not likely to be redone because it's a large effort and requires a level of hackery in ffmpeg that will never be accepted upstream, which is the design requirement for all code now. Tests show the non-optimised HEVC support in LE11 images is sufficient for SD media, but not 720p/1080p. So if you really need HD media, either stick with LE9.2 or bump up to an RPi4 with native HEVC hardware decode.

    Thanks for bearing with me and explaining the current problem.

    I can see what the development team is dealing with.

    Thanks again for the info and I hope you guys have a great holiday.