mrv2 v1.1.4

v1.1.4

This is a major release with big speed improvements and fixes to the color pipeline.

• On the second re-release of v1.1.4 annotations with blit viewports are possible.
• On a re-release of v1.1.4 improved OpenGL performance as soon as a clip is
  loaded and starts playing, when:
  Window->Preferences->Playback->Auto Playback
  and
  Window->Preferences->Playback->Auto Hide Pixel Bar
  are active.
• Simplified build scripts.
• Fixed macOS build issues (fixed in v1.1.3 re-release actually).
• Prevented a macOS python plug-in issue with Python 3.11.
• Fixed Wayland support on Ubuntu 22.04.3 LTS and later.
• Improved performance of OpenGL drawing between 20% and 150%.
  macOS has seen the most benefit going from playing videos at 4K at 15FPS to 60FPS.
• Added Preferences->OpenGL->Color Buffer Accuracy to select between automatic,
  float or 8-bit buffers for speeding up movies or getting more accurate color
  precision.
• Added option to select blitting of main viewports for potentially
  faster OpenGL drawing.
• Added Windows' stack traces with line numbers for easy debugging.
• Added Windows' stack traces to be saved into tmppath(). That is:
  %TMP%/mrv2.crash.log.
• Allowed pyFLTK be able to compile in Debug.
• Fixed floating values not respecting float values below 0 or higher than 1.
• Improved the performance of OpenGL by avoiding OpenGL context switching when
  not needed (you don't have a text input widget, nor a text annotations
  showing).
• Added Linux's stack traces with line numbers by relying on libbacktrace.
• Improved performance of OpenGL dramatically, particularly on macOS.
• Improved performance of drawing OpenGL annotations (drawings).
• Renamed skipped frames as drop frames.
• Fixed Preview thumbnails above the Timeline crashing the viewer on all
  platforms.
• Hide Preview thumbnails when going fullscreen or presentation.
• Fixed Playlist creation showing an empty view for the first clip loaded.
  (regression of v1.1.2).
• Several fixes to Presentation Mode.

mrv2 v1.1.3

• pyFLTK build improvements (removal of warnings).
• Made Windows .zip file smaller by removing unit tests.
• Added Linux Desktop name to About Window.
• Fixed update-mrv2.py plug-in for Windows. The install location is now retrieved from the registry. Only if not found it looks in hard-coded paths.
• Install update-mrv2.py always, instead of removing it when python demos was not installed.
• Added instructions on Windows on how to retrieve the Explorer thumbnails.
• Darby added automatic scrolling of the timeline when playing and ability to toggle track and clip info.
• Added Darby's new options to Network connections.
• New action icons, which should display properly at high resolutions.
• New custom file requester icons, which should display properly at high resolutions.
• Sped up comparison wipes with Alt and Shift dragging.
• Sped up color corrections from the Color Panel.
• Fixed libdecor incompatibility with older libdecor plug-ins.
• Fixed + and - signs rotating the image when using one of the draw tools and
  not just the pencil tool. With all of them now the pencil size is changed.
• Fixed Python installation on macOS on a re-release two days later.

mrv2 v1.1.2

v1.1.2

• Windows version of mrv2 is now compiled with MSVC 2022.
• Fixed opening directories from within the command-line.
• Fixed stopping of playback when playing and selecting an area with
  Shift + Left Mouse Button drag or panning with Ctrl + Drag.
• Made status bar report "Everything OK." after 10 seconds without further
  errors.
• Removed -debug flag from command-line as it only works on debug builds.
• Made Metadata tab not appear if all metadata is shown in video/audio
• Fixed a compilation issue between FLTK's JPEG libraries and those of
  libjpeg-turbo on Linux.
• Fixed Right Mouse Button toggling off the audio when there had been a drag
  action with some other button before.
• Fixed a precision problem when shifting annotations at the beginning of a
  clip due to a following clip being dragged to the same place as the
  annotation.
• Improved performance of timeline drawing when annotations are present.
• Fixed refreshing of image sequences
  (ie. Files Panel->Select Image->RMB->File/Refresh Cache) when there were
  missing frames and later they were added. It used to work in v1.0.0 but a
  change in tlRender broke it.
• Updated mrv2 to Python 3.11 to conform to the VFX 2024 Platform.
• Python 3.11 on Windows is now built with optimizations on.
• Updated mrv2's glfw3 to 3.4.
• Updated mrv2's FLTK build to use the system libdecor
  (needs v0.2.2 at least).

mrv2 v1.1.1

v1.1.1

• Added keyboard shortcuts for Render/Video Levels and Render/Alpha Blend.
• Added support for multiple audio tracks. You can switch from one to the
  other by selecting the Audio button (the speaker icon) with the
  Right Mouse Button and selecting a track.
• Fixed audio metadata in audio section of Media Info Panel for each audio
  stream.
• Added missing video metadata in video section of Media Info Panel.
• Improved speed of refreshing of a clip (when switching audio tracks or using
  View->Ignore Display Window).
• Made movies longer than 30 minutes with multiple audio tracks, use a short
  read ahead / read behind or switching audio would be too slow.
• Fixed saving movies with annotations and with an alpha channel.
• Fixed saving movies with annotations at half or quarter resolution.
• Fixed an ugly bug on Linux when saving movies with annotations repeating
  every other frame (ie. saving on two's).
• Fixed line error reporting in Python's Output (pybind11 bug).
• Added Editor/Jump to Error to Python Panel. You can select the line of the
  error in the Python output and the Editor will jump to the offending line.
  If you don't select anything in the Python output, it will jump to the last
  error counting from the bottom.
  It will skip errors from files that are <frozen ...>, like importlib.
  If the error is on a file and not from the Python Editor, mrv2 will try to
  open an external editor and jump to the line of it.
  The editor to use is taken from the EDITOR environment variable first.
  If not set, mrv2 will try to use one of the popular editors. Its defaults
  supports the syntax for emacs, gvim, vim, VS Code, notepad++ and nano.
  The preferred editor is checked at start up based on what you have installed
  on your PATH and whether you are running interactively or also have a
  terminal.
  Visual editors are given preference over terminal editors.
  Of course, you can edit the editor with Python Panel->Editor/External Editor.
• Fixed Python error reporting in the Python Output of the Python Panel for
  mrv2 python plug-ins.
• Fixed message that said it was saving a movie without audio when saving
  a sequence of frames.
• Fixed reading of FPS from OpenEXR images as they are now rational instead of
  a double.
• Fixed thumbnail of image sequences in mrv2's custom file requester when the
  sequence did not start at 0.
• Fixed a minor bug using snprintf of int64_t with an int specifier.
• Improved the main web page for mrv2.
• Fixed translation of "Color" in Panels. Of course, since English and
  Spanish use the same word, this still makes no difference :D!
• Fixed the name of the video stream in the Media Info Panel to be like the
  layer name listed in the Panels and in the Color Channel pull down.
• Added back truehd decoder and encoder which were removed from
  TLRENDER_FFMPEG_MINIMAL compilations.
• Sped up redrawing when the Media Info Panel was open, as we know refresh
  only the tabs that change.
• Fixed changing the first and last frames from the Media Info Panel.
• Modified int sliders to use HorSlider instead to make it consistant.
• Fixed (very slow) changing speed from the Media Info Panel.
• Fixed HorSlider resizing improperly when values were bigger than 10000 or
  smaller than -10000.
• Improved precision when accepting or editing text with a zoom factor or
  pixels_per_unit different than 1.

mrv2 v1.1.0

v1.1.0

• Build fixes.
• Made mrv2's custom file requester list presumed sequences of a single frame
  as a file instead of a custom sequence name display.
• Matched NDI colors.
• Made NDI streams not display thumbnails in any of the panels, as it would
  look like mrv2 had hanged.
• Made installed size and installer smaller, by removing some unneeded python
  libraries used for documenting mrv2.
• Made installed size and installer smaller, by removing Python's unittests.
• Made default Windows' installer not install Python's Tk libraries by default.
• Made default Windows' installer not install IDLE nor turtledemo.
• Overall, the Windows installed size is now almost 150Mb smaller, now at 325Mb.
• Made Windows installer create a registry entry backup for the file
  associations, instead of using an .ini file which was worse. On uninstall,
  the original file association is restored or the registry entry is deleted.
• Added a friendlier name to Windows' Open With RMB menu. Now the version is
  listed, instead of just "mrv2.exe". Also the name in English is provided.
• Added a latest version to Windows' Open With RMB menu. If used, it will
  associate the latest version installed with the file. Upgrading won't
  require you to reset file associations like before.
• Fixed rotations not framing the image even when View->Auto Frame was on.
• Added symbolic links on Windows .exe installer, so that it works more like
  Unix.
  You can use mrv2-v1.1.0.exe to call a specific version of the viewer for
  example.
• Split Timeline into "Timeline" and "Edit" Preferences in Preferences Window.
• Added Default View to "Edit" to select what gets displayed when you hit the
  Edit button ("Video Only" or "Video and Audio").
• Fixed Edit Viewport size when transitions were present.
• Fixed a potential crash when switching to nuke's ocio config.
• Made Python's setOcioConfig accept ocio::// configurations.
• Fixed colors on Python commands being orange instead of dark yellow as before.
• Fixed Python's Panel output, which was not redrawing or scrolling when in Window mode.
• Made View pull-down menu display the full name of the display / view.
• Made UI topbar resize a tad nicer.
• Fixed Python Output in Python Panel which was not refreshing at all due to
  my misuse of the FLTK API since v0.9.3.
• Fixed a stray menu bar on Python Panel on closing it.
• Improved performance of package creation using multithreading when possible
  with cpack's generators.
• Added support for NDI's BT601 and BT2020.
• Added support for NDI's alpha channel.
• Added support for saving PNG and TIFF with alpha channel when saving a half
  or float OpenEXR image. TIFFs will get saved in float. PNG in 16-bits.
• Eliminated duplicated error and warning messages from the logs.
• Improved start up time and sync issues with NDI streams.
• Simplified all parameters in NDI stream to just selecting the source.
• Allowed NDI streams to change size, pixel aspect or color space.
• You currently cannot mix two NDI streams, one with audio and one without.
  Only the first one information will be used.
• You also cannot stream two clips through NDI with different audio
  frequencies or channels.
• Streaming two clips through the same connection after stopping for a while
  can lead to audio stuttering. Either stream them one after the other or
  open a new connection.
• Fixed duplicated FLTK libraries on Linux distribution.
• Fixed framing view while playing (quick re-release of v1.1.0).
• Fixed auto frame view when switching clips of different resolutions (second re-release of v1.1.0)

mrv2 v1.0.9

v1.0.9

• Added a preference setting on Preferences->Errors to control what to do on FFmpeg errors.
• Made the panel drag bar a tad more grey, for a nicer look and more clear distinction when several panels are stacked on the dockbar.
• Fixed a thumbnail refresh issue on mrv2's custom file requester which would show when we save over a previous file. tlRender would cache the file and would display the previous thumbnail before saving. Now it is fixed.
• Removed mrvSequence.cpp which was empty.
• Added Global USD Render Option parameters which were missing.
• Fixed overlapping of audio labels in Edit window.
• Fixed rotation inverting the coordinates of the annotations.
• Fixed the pixel coordinate information when flipped or rotated.
• Fixed the pixel color bar display and information when rotated.
• Fixed text tool incorrect positioning when image was rotated.
• Made arrow head size be proportional to the line size.
• Made drawing cursor be drawn with float coordinates.
• Made circle's center also be drawn with float coordinates.
• Added back .wmv/a decoders and asf demuxer in TLRENDER_FFMPEG_MINIMAL.
• Added Video Color Primaries, Color TRC and Color Space FFmpeg information to Media Info Panel's video.
• Fixed FFmpeg reader to take color coefficients. Now the gray ramp of Sam Richards is the same.
• Fixed a bug in session loading when there was no Input Color Space specified
  in the image, which got introduced in v1.0.8.
• Made pen size be able to be 1 pixel making the cursor one line only.
• Made prores_ks (FFmpeg's native encoder) match Apple's ProRes one at 4444p10le. It is now possible to take ProRes4444 encoded files to video editors. THIS IS A BIG IMPROVEMENT!!!
• Upgraded to FFmpeg v6.1.1.
• Improved rotation speed of movies from cell phones, by doing it in OpenGL with metadata instead of with a C++ function.
• Added Creation, Modified Date and Disk Space to Media Info Panel.
• Removed Frame information from Start/End Time in Image Tab of Media Info Panel.
• Made Start/End Time in Image Tab of Media Info Panel not use scientific notation for seconds.
• Fixed changing frame rate from the Timeline toolbar not updating the FPS indicator in the Image Tab of the Media Info Panel.
• Improved the performance of Media Info Panel when playing a sequence and the Image tab was closed.
• Removed Default Speed information from Video Tab and moved it to Image Tab.
• Improved colors of deactivated enums showing too dark to read the black text by default.
• It is now possible to use OpenColorIO's built-in configs. That is, those stating with ocio://.
• Added a Choice menu in Preferences->OCIO to select any of the three built-in OCIO configs.
• Set ocio default config to the built-in:
  ocio://default
• Fixed a focus issue on Linux when using a draw tool, then typing into one input widget in the toolbar and going back to the view window not ending the draw tool, but continue its trace.
• Fixed Text annotations when image was rotated jumping to the wrong location on confirmation.
• Entering a Text annotation on the viewport can be confirmed by doing SHIFT + Enter.
• Added FFmpeg's actual version information, not just each library.
• Fixed listing of codecs in the About Window.
• Added missing .gif decoder and encoder.
• Added mrv::TextBrowser to allow copying the text in them with a RMB context menu. This is useful, for example, in the About Window for the GPU or codec information, as the full lines get copied, unlike a Fl_Text_Display in which a line can be cut in half.
• Fixed gcc version information in mrv2's About Window. It was showing the GLIBCXX instead.
• Darby simplified and fixed the A/B comparison sometimes getting out of sync.
• Improved the look of HorSlider widget. Now the X to reset to defaults is a tad to the right.
• Fixed selection of Wipe mode which would not work after scrolling in the dock (FLTK bug?).
• Made Compare options automatically frame the view if View->Auto Frame is on.
• Improved the playback performance of the viewer a tad.
• Fixed an OpenGL flickering of the timeline and sometimes of the main viewport
  when the secondary window was opened and then closed.
• Fixed cursor not showing in panels when drawing was enabled.
• Improved the performance of adding clips to an EDL Playlist. There's no more flickering.
• Fixed thumbnails in the Files and Stereo panels resetting to their start time, instead of showing their current time.

mrv2 v1.0.8

v1.0.8

• Added rotation of the images +90 and -90 with automatic framing if
  "View->Auto Frame" is on.
• I have enabled TLRENDER_FFMPEG_MINIMAL as a default to compile FFmpeg with
  less muxers, demuxers, parsers and protocols, for smaller file sizes and a faster
  compilation and program start up.
• Made Help->Documentation point to the online docs if the local docs are not installed.
• Made H264 entry in Save Movie Options panel not show when FFmpeg was built as LGPL.
• Updated OpenUSD to v24.03.
• Fixed <| (go to start) and |> (go to end) buttons not refreshing the button status on playback.
• Added -ics (-ocioInput), -od (ocioDisplay), -ov (-ocioView) and
  -ol (-ocioLook) command-line flags to override the default settings.
  Note that -ocioDisplay and -ocioView must be used together.
• Made OCIO Input Color Space be stored with the image, except when doing
  comparisons in which the first image's Input Color Space is used.
• Made OCIO Input Color Space for each image be stored in session files.
• Made FFmpeg's repeated warnings and errors only show up once in the status
  bar and the logs.
• Added a Resolution parameter to the Save Movie options. You can save:
  • Same Size
  • Half Size
  • Quarter Size

mrv2 v1.0.7

v1.0.7

• Slow code clean up.

• Faster windows minimal compilations for quickly checking on GitHub.

• Added optional component installation on Windows's .exe installer. Currently,
  you can install:

  * mrv2 application (obviously)
  * mrv2 documentation
  * mrv2 Python demos
  * mrv2 Python libraries
  

• Changed name of install directory of mrv2 on Windows from "mrv2 vX.X.X" to "mrv2-vX.X.X" to avoid spaces for easier scripting.

• Made the installer and mrv2's intalled size smaller, shaving close to 120Mb.

• Fixed image sequence detection that could happen randomly depending on how
  the filesystem returned the order of files.

• Added listing the OS and distro versions at the start and in About of mrv2
  for debugging purposes.

• Made panels be listed alphabetically, regardless of Natural Language.

• Automated version update in docs.

• Removed view sources from html docs to save space on disk.

• Cleaned up docs directory before building docs.

• Sped up creation of docs in build system.

• Updated Python API with FFmpeg and Background changes.

• Fixed reading permissions on files as they were broken! I did not notice
  as I was using an NTFS drive which sets umask 0022 by default.

• Made reading session more robust to handle missing files or wrong settings.

• Made checking for readable files faster.

• Fixed saving large images/movies with annotations on work properly.

• Improved memory consumption of saving movies, particularly large ones like
  4K.

• Improved logging system logging the tlRender warnings.

• Improved logging system adding a Status mode to report information right
  away.

• FFmpeg's logging now prints out the codec/module where the error was
  generated.

• Improved movie save reporting of parameters.

• Fixed saving a movie with annotations when the movie is bigger than the
  viewport.

• Added GBR8/9/10/12 reading support for VPX.

• Fixed start and end timeline buttons not refreshing thumbnails in the Panels.

• Fixed image panel not refreshing its information when changing images and
  the playback was stopped.

• Fixed Preferences->Positioning->Position/Size when both were used.

• Saving of .otio files as movies when the first clip did not start at 0 now
  works properly.

• Upgraded to RtAudio v5.2.0 on all platforms.

• Added a "View/Auto Frame" to turn off auto framing of the view when changing
  from one clip to another.

• Added User Interface->View->Auto Frame to preferences.

• Made saving a VP9 or AV1 with the wrong extension not fail. Instead, they
  are renamed to .mp4.

• Made mrv2's file requester favorites listing automatically remove
  non-existent directories.

• When saving a sequence of OpenEXR, we encode the speed in the actual
  image file, as taken from the playback tool bar. This value will take
  precedence over the Sequence Default speed as set in Preferences->Playback.

• Made warnings also show up in the status bar, but with an orange background.

• Improved HUD Attributes. They are now listed alphabetically and they are not
  repeated. Also, they refresh properly.

• Fixed Media Information panel not refreshing properly when changing images.

• mrv2's tlRender library now reads the video and audio stream metadata.

• When there's no audio metadata there's no longer the titles of
  Attribute/Value used at the end of the Metadata tab in the Media Info Panel.

• Metadata in Media Info Panel is now sorted and stripped of repeated data.

• Media Info Panel's tabs now remember whether they were opened and closed.

• Added GoPro Cineform codec to the list of profiles you can use to encode.
  The biggest benefit is that it can encode at GBRP_12LE and GBRAP_12LE (ie.
  RGB and RGBA at 12 bits with alpha) for a replacement to Apple's proprietary
  ProRes4444 codec.
  It is supported by Adobe Suite (Premiere, After Effects, Media Encoder),
  Apple's Final Cut Pro, Blackmagic Design Davinci Resolve, Avid Media
  Composer, Nuke, Flame and Inferno.

• Removed the GoPro prefix name from Cineform when saving.

• Fixed saving of EXR movies to movie files being just RGBA_U8 instead of
  RGBA_U16.

mrv2 v1.0.6

v1.0.6

• Darby fixed Render->Alpha Blend modes which were partially broken.
• Darby fixed macOS checkers drawing.
• Darby added gradient background and moved it to draw video.
• Made Background colors use Flmm's color chooser with alpha (alpha is current discarded, thou).
• Added printing out of audio and video codec names when saving with FFmpeg.
• Fixed presets names. Instead of using '_' we now use '-'.
• Added last successful saved movie to recent files for easy checking it back.
• Fixed pixel format selection in VP9 codec not working properly due to profile being set to 0, which would prevent saving with 10 or 12 bits.
• Fixed Prores color shifting due to bad unneeded use of libswscale API.
• Added options for color_range, colorspace, color_primaries and color_trc in tlRender's source code. mrv2 can set them if you open the window called Advanced Settings in the Save Movie dialog.
• Made mrv2 remember the save movie and image settings for easy saving of multiple movie or image files.
• Made FFmpeg errors be reported to the console immediately.
• Fixed macOS VideoToolbox's hardware encoding not working.
• Fixed Windows' presets not being read due to spaces in the installed directory.
• Fixed a long standing bug of gamma (when changed in the top bar slider) not being reapplied when changing from one clip to another.

mrv2 v1.0.5

v1.0.5

• Code clean up.
• Build system clean up and consolidation with environment variables.
• Fixed building pyFLTK on Linux which could fail if LD_LIBRARY_PATH was not
  set.
• Improved building speed on all platforms.
• Fixed Windows' Python compilation screwing up if some other Python version
  was installed.
• Fixed pyFLTK compilation on macOS and Linux having swig not use the right
  path includes.
• Fixed NDI compilation which had gotten broken.
• Fixed listing of movie files when they were named as sequences.
• Allowed saving movie files with the speed (FPS) as set in the playback
  toolbar as long as you are not saving audio.
• Updated version in web page docs.
• Added FFmpeg presets for saving codecs. Currently we ship mjpeg (none),
  h264, vp9 and av1 presets, but you are free to create your own.
• Switched to building with gcc-12 on Rocky Linux 8.9 (not in Dockerfile or
  Github builds).
• Fixed encoding of movie files' YUV conversion. Now the movie file is much
  more accurate.
• Added pixel formats currently supported for each codec.
• Added saving alpha channel in ProRes_4444 and ProRes_XQ when YUVA_4444P_16LE
  pixel format is selected.
• Added saving alpha channel in VP9 when YUVA_420P is selected and the container
  is a Matroska file.
• Added HISTORY.md file to the Web docs.