[flyingrat]

I am still showing 0bytes cached playing a HDS/HLS stream. Is that number being updated by ffdmpeg or is the caching happening internal to ffmpeg and not displayed on the debug. I did notice it took a little longer to start then normal so it may have been buffering.

I can play 720p 2,5 Mbps HLS without problem. Can you elaborate the issue with some more details, please.

[elupus]

I would very much like to see us getting the 1.2 release. That is the final release before the new refcounted frames stuff. After that will be a messier update.

[elupus]

What is the current status here btw? Do we have an updated list of dropped or not dropped patches?

[vicbitter]

I would very much like to see us getting the 1.2 release. That is the final release before the new refcounted frames stuff. After that will be a messier update.

I ported the XBMC patches (from ffmpeg 1.1.3) to ffmpeg 1.2 today and have ffmpeg 1.2 running with XBMC. Early days but it seems to be running okay.

---

What is the current status here btw? Do we have an updated list of dropped or not dropped patches?

These are the XBMC patches needed for ffmpeg 1.2

Code:

0001-Support-raw-dvdsub-palette-as-stored-on-normal-dvd-s.patch
0002-Change-fallthrough-logic-for-read_seek-to-be-based-o.patch
0003-matroska-Check-return-value-of-avio_seek-and-avoid-m.patch
0004-asf-hacks.patch
0005-if-av_read_packet-returns-AVERROR_IO-we-are-done.-ff.patch
0006-added-Ticket-7187-TV-Teletext-support-for-DVB-EBU-Te.patch
0007-Don-t-accept-mpegts-PMT-that-isn-t-current.patch
0008-Don-t-reparse-PMT-unless-it-s-version-has-changed.patch
0009-fixed-compile-with-VDPAU-header-versions-without-MPE.patch
0010-Read-PID-timestamps-as-well-as-PCR-timestamps-to-fin.patch
0011-Get-stream-durations-using-read_timestamp.patch
0012-include-stdint.h.patch
0013-aacenc-add-recognized-profiles-array.patch
0014-changed-allow-4-second-skew-between-streams-in-mov-b.patch
0015-fixed-memleak-in-mpegts-demuxer-on-some-malformed-mp.patch
0016-Speed-up-mpegts-av_find_stream_info.patch
0017-allow-customizing-shared-library-soname-name-with-ma.patch
0018-dxva-mpeg2-Allocate-slices-array-dynamically-fixes-v.patch
0019-dxva-mpeg2-speed-up-slice-allocation.patch
0020-dxva-vc1-Take-BI-into-account-for-forward-and-backwa.patch
0021-dxva-vc1-Pass-overlapping-transforms-hint.patch
0022-dxva-h264-Fix-dxva-playback-of-streams-that-don-t-st.patch
0023-Changed-format-string-td-not-supported-by-our-MingW-.patch
0024-add-public-version-of-ff_read_frame_flush.patch
0026-Handle-return-value-of-BeginFrame-better.patch
0029-stop-forcing-gas-preprocessor-for-darwin.patch

[juggie]

I am still showing 0bytes cached playing a HDS/HLS stream. Is that number being updated by ffdmpeg or is the caching happening internal to ffmpeg and not displayed on the debug. I did notice it took a little longer to start then normal so it may have been buffering.

I can play 720p 2,5 Mbps HLS without problem. Can you elaborate the issue with some more details, please.

Hi Flyingrat,

I can play 720p HLS without a problem too (most of the time), even on Frodo. When I tested your version on win32 (my htpc is linux, but testing on win32 was just quicker). It was not obvious if any buffering was occurring. When I press O while playing a 720P HLS stream, it shows 'cache 0B' and the percent fluctuates, usually 88/100% when the stream is working good. Thats the same behaviour as frodo which is the subject of this thread: http://forum.xbmc.org/showthread.php?tid=154016 which is the thread that led me here. If ffmpeg is indeed buffering, it would be awesome to be able to control the size of the buffer, visual display in the debug would be a bonus as well.

Thanks!

[flyingrat]

I am still showing 0bytes cached playing a HDS/HLS stream. Is that number being updated by ffdmpeg or is the caching happening internal to ffmpeg and not displayed on the debug. I did notice it took a little longer to start then normal so it may have been buffering.

I can play 720p 2,5 Mbps HLS without problem. Can you elaborate the issue with some more details, please.

I can play 720p HLS without a problem too (most of the time), even on Frodo. When I tested your version on win32 (my htpc is linux, but testing on win32 was just quicker). It was not obvious if any buffering was occurring. When I press O while playing a 720P HLS stream, it shows 'cache 0B' and the percent fluctuates, usually 88/100% when the stream is working good. Thats the same behaviour as frodo which is the subject of this thread: http://forum.xbmc.org/showthread.php?tid=154016 which is the thread that led me here. If ffmpeg is indeed buffering, it would be awesome to be able to control the size of the buffer, visual display in the debug would be a bonus as well.

Yeah, to display/control the internal ffmpeg cache buffer would be a great feature but unfortunately it will have to wait until we have "shoehorned" the new ffmpeg version into xbmc for all platforms. /Regards, Lars.

[flyingrat]

What is the current status here btw? Do we have an updated list of dropped or not dropped patches?

During the work to verify and in some cases refactor the old n1.1.3 patches to n1.1 I took the opportunity to check the status of some lingered upstream patches. It seems that some of the upstream patches (originated from xbmc) still didn't make it into release 1.2, possibly because these patches weren't marked as tested. It's also worth noting that even though it's currently not a huge pain to refactor the xbmc custom patches, it's getting harder for each new release of ffmpeg thus it's probably worth putting some effort to pull them upstreams. I'll start working with this as soon as the new ffmpeg libs are pulled into xbmc.

The big effort right now seems to be testing all platforms. It's still unclear to me whether the platform testing should be coordinated and carried out by the xbmc team (after a pull request?) or if it's up to the "pull requester" to gather a test crew for this purpose. How is it supposed to work?

Anyhow, there are now two new branches:

1. The XBMC branch "ffmpeg-n1.2", fork git://github.com/FlyingRat/xbmc-ffmpeg-v1.1-bump.git which includes the new ffmpeg "n1.2-xmbc" branch (from below) on top of a rebased master.

2. The FFmpeg branch "n1.2-xmbc", fork git://github.com/FlyingRat/FFmpeg-with-XBMC-custom-patches.git. This branch includes ffmpeg release 1.2 with the refactored xbmc add on patches.

These are some obsolete xbmc add on patches I found so far:

Code:

Obsolete (N/A): 0012-include-stdint.h.patch
Obsolete (N/A): 0023-Changed-format-string-td-not-supported-by-our-MingW-.patch
Obsolete (N/A): 0024-add-public-version-of-ff_read_frame_flush.patch
Obsolete (N/A): 0029-stop-forcing-gas-preprocessor-for-darwin.patch
Obsolete (1.2): 0035-ffmpeg-fixed-missing-S_DVBSUB-subtitles.patch

These are the xbmc add on patches that are included in the "n1.2-xmbc" branch:

Code:

0001-Support-raw-dvdsub-palette-as-stored-on-normal-dvd-s.patch
0002-Change-fallthrough-logic-for-read_seek-to-be-based-o.patch
0003-matroska-Check-return-value-of-avio_seek-and-avoid-m.patch
0004-asf-hacks.patch
0005-if-av_read_packet-returns-AVERROR_IO-we-are-done.-ff.patch
0006-added-Ticket-7187-TV-Teletext-support-for-DVB-EBU-Te.patch
0007-Don-t-accept-mpegts-PMT-that-isn-t-current.patch
0008-Don-t-reparse-PMT-unless-it-s-version-has-changed.patch
0009-fixed-compile-with-VDPAU-header-versions-without-MPE.patch
0010-Read-PID-timestamps-as-well-as-PCR-timestamps-to-fin.patch
0011-Get-stream-durations-using-read_timestamp.patch
0013-aacenc-add-recognized-profiles-array.patch
0014-changed-allow-4-second-skew-between-streams-in-mov-b.patch
0015-fixed-memleak-in-mpegts-demuxer-on-some-malformed-mp.patch
0016-Speed-up-mpegts-av_find_stream_info.patch
0017-allow-customizing-shared-library-soname-name-with-ma.patch
0018-dxva-mpeg2-Allocate-slices-array-dynamically-fixes-v.patch
0019-dxva-mpeg2-speed-up-slice-allocation.patch
0020-dxva-vc1-Take-BI-into-account-for-forward-and-backwa.patch
0021-dxva-vc1-Pass-overlapping-transforms-hint.patch
0022-dxva-h264-Fix-dxva-playback-of-streams-that-don-t-st.patch
0026-Handle-return-value-of-BeginFrame-better.patch

---

Platforms builders, please notice:
Work is in progress to unify the dependency build for linux, android, rbpi and osx. Until that work is finished start with commit 8f219c5 (from the 10th) if you want to make a successful build on any of these platforms! Ref: "http://forum.xbmc.org/showthread.php?tid=107126&pid=1368880#pid1368880".
--

[DonJ]

The big effort right now seems to be testing all platforms. It's still unclear to me whether the platform testing should be coordinated and carried out by the xbmc team (after a pull request?) or if it's up to the "pull requester" to gather a test crew for this purpose. How is it supposed to work?

I suggest you do a pull-request once you feel comfortable with the patch - platform devs will perform tests from there and you'll receive feedback from the team.

[flyingrat]

The big effort right now seems to be testing all platforms. It's still unclear to me whether the platform testing should be coordinated and carried out by the xbmc team (after a pull request?) or if it's up to the "pull requester" to gather a test crew for this purpose. How is it supposed to work?

I suggest you do a pull-request once you feel comfortable with the patch - platform devs will perform tests from there and you'll receive feedback from the team.

Ok, thanks! I'm just gonna run a couple of test-builds with the updated build dependencies and then I'll issue a pull request.

[flyingrat]

New "XBMC-FFmpeg-n1.2-binaries" for win32, osx and ios uploaded. Please test and provide feedback! For the first time ever I'm now able to watch hires (720p) HLS stream from SVT play on my mac and ipad without stuttering, oh yeah!

[vicbitter]

New "XBMC-FFmpeg-n1.2-binaries" for win32, osx and ios uploaded. Please test and provide feedback! For the first time ever I'm now able to watch hires (720p) HLS stream from SVT play on my mac and ipad without stuttering, oh yeah!

An RPi build, if anyone is interested...

OpenELEC RC5 for RPi with ffmpeg 1.2

[bjober]

Guys, I am a dedicated follower of this thread. I must say you are doing a tremendous work and I just wish I've been a programmer to help you out.
One thing I could help with would be to test the builds on the atv2 but I can't find the iOS binaries Flyingrat have uploaded.

[Ed76]

thanks flyingrat for you work! i will test the windows version at the next weekend. one question : i think that the most important problem of xbmc is the bad picture quality of the windows xbmc version (compared to the ds-players with lav codec). have you noticed an improved picture quality with updated ffmpeg? thanks for you answers!

[flyingrat]

Guys, I am a dedicated follower of this thread. I must say you are doing a tremendous work and I just wish I've been a programmer to help you out.
One thing I could help with would be to test the builds on the atv2 but I can't find the iOS binaries Flyingrat have uploaded.

My fault! I probably closed the lid before the uploading was finished but it's on its way as we speak. There is also a new raspbmc package, many thanks to mr Bberski!

---

thanks flyingrat for you work! i will test the windows version at the next weekend. one question : i think that the most important problem of xbmc is the bad picture quality of the windows xbmc version (compared to the ds-players with lav codec). have you noticed an improved picture quality with updated ffmpeg? thanks for you answers!

I believe the quality has improved, especially the HLS streams are much better! It should be "state of art" as it's based on the latest ffmpeg libraries

[flyingrat]

I took the liberty to upload Vicbitters OpenELEC version as well, thank you!