* Moved writing of configure options from Controller to Model. Fixes#191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a9.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
First is some more debugging lines useful for figuring out why automatica palette selection didn't work.
The other is a comment about how the code will segfault if there are more than 64 formats. Currently there are not.
Lastly witch a variable from int to unsigned int. There is no way for it to go negative and logically should always be positive, so by making it an unsigned in we don't have to cast it later.
* ZoneMinder compiles with ffmpeg 0.5..master
* Make VideoStream class able to stream through ffmpeg better.
* Ability to use fixed quality instead of fixed bitrate (by specifying bitrate 0...100)
* Format url parameter supports <format>/<encoder> syntax
Details:
- Redefine av_err2str to a to avoid compiler warning on newer g++'s
- When using rtp format; if ffmpeg does not have a default codec, use CODEC_ID_MPEG4. (because this is what the default is as of 2.2 when this code was written)
- Specify ofc->packet_size if rtp format is requested
- Video generated in a thread to guarantee a constant fps.
- Move _AVCODECID definition into zm_ffmpeg.h (and use instead of (AV)CodecID
- Call avformat_network_init because ffmpeg warns about it beeing required soon.
- increase VideoStream::video_outbuf_size to 4MiB to be able to encode larger pictures
1) Removed the V4L2_CAPTURE_FIELDS option.
2) V4L2 Interlacing control options added to the new per-monitor deinterlacing option.
3) Fixed the 32bit version of Linear.
4) Added Blend (50%) and Blend (25%).
5) Completed the grayscale version of the Four Field Motion Adaptive deinterlacing algorithm.
1) Added a workaround against SWS_CPU_CAPS_SSE2 not being defined.
2) Fixed enableDisableAlarms javascript error.
3) swscale unsupported format errors now include the unsupported format in the error.
4) Fixed a small error affecting local cameras with rotation enabled.
1) Fixed wrong size calculation in Image::AssignDirect.
2) Improved libjpeg subpixel order selection.
3) Finished adding support in the simple Overlay function.
4) Completed Flip and Rotate 32bit RGB support.
5) Allow ZM to compile on 32bit with omit frame pointer disabled by defining _DEBUG
To use one less register in the SSE algorithms.
6) Removed the counter variable in some loops to reduce loop overhead.
7) Modified crop query error handling.
8) Most of the shared data now declared as volatile.
9) Small improvements to the AlarmedPixels motion detection.
10) Changed the default blend percent from 7% to 12%.
11) Fixed an earlier bug created by me: motion detection checking the wrong pixels.
2) ZoneMinder should now compile cleanly wihout swscale
3) Fixed multiple monitors capturing from the same device and channel.
Current code allows for multiple monitors sharing the same device, each on a different channel
Or, multiple monitors sharing the same device, all on the same channel.
In both cases, capture method, width, height and palette must be identical on all monitors.
However, target colorspace can be different because each monitor handles the format conversion separately.
2) Allow ZM to compile without swscale - hopefully
3) Saved few cpu cycles during image blending
4) Fixed wrong zone colours showing up
5) Unrelated fix: Allow ZM to be compiled with the latest ffmpeg revisions
2) Changed image class buffer related functions
3) Improved format selection for local cameras
4) Removed imagesize check for swscale
5) Fixed mmap unexpected memory size when changing capture options
6) Fixed unrelated bug in zmdc.pl: sockets directory doesn't exist
2) SSE2 aligned memcpy now functional
3) Fixed fill functions bug from an earlier commit
4) Many changes to zm_local_camera.cpp format selection
5) Fixed memory leak in AssignDirect()
6) Fixed Target Colorspace webinterface issues
7) Small other changes
1) Delta and blend SSE and standard alogirthms inside
2) Changed event streaming to use sendfile() syscall if available
3) Direct memory capture (Less memcpys required)
4) Added macros for BGR24 and RGB32(and its variants) formats
5) Modified Monitors database table to add a target colorspace option
6) Changed most image buffer allocations to be on 16 byte boundary, work still in progress
7) Added Target colorspace instead of the old colours selection. this one also visible to local cameras
8) Replacing Y deltas option with ZM_CPU_EXTENSIONS
9) Added SSE detection code
10) Changed some code to make linking success with inlining enabled.
11) Some other changes and work still in progress for some listed above
Isaac Connor
Isaac Connor
Isaac Connor
APHW2 MFGENG
abishai
Michael Schupikov
Andy Bauer
Steve Gilvarry
Isaac Connor
SteveGilvarry
Dmitry Smirnov
Emmanuel Papin
Isaac Connor
Andrew Bauer
Sune1337
Kfir Itzhak
Isaac Connor
nextime
stan