Release v0.1.21

• New:
  • Option --camera_position to provide coordinates where the camera clip should be positioned in the video providing very granular custom layout options. Using this option results in ignoring the layout option.
  • Option --camera_order to define the order the cameras should be processed. This allows to overlay one camera over another one and define which one should be on top.
  • Left and right-pillar cameras have been included as options and within the different layouts.
  • If an event does not have files for a certain camera (i.e. pillar cameras) then resulting layout will be as if those cameras were excluded.
  • Added street name from events.json, this became available with Tesla 2025.38.6
  • Added option --ffmpeg_debug so that when running with loglevel debug, FFMPEG output is not provided unless this is enabled as well.
  • New layout MOSAIC replacing WIDESCREEN as WIDESCREEN does not make sense with all cameras.
  • Docker: Added pre-release docker image making it easier to use pre-release versions.
• Changed:
  • Many internal changes, including how layouts are calculated, typing, etc.
  • start_timestamp and end_timestamp will now also restrict within the clip and not just folder.
  • The MacOS executable is now compiled for Apple Silicon.
  • Docker: Reduces Docker image sizes by 24-66% by switching from jrottenberg/ffmpeg to Debian's FFmpeg package on python:3-slim base. Contributed by @magicalyak
  • Docker: Developer QoL improvements for Windows-based contributors. Contributed by @ozskywalker
  • Docker: Now will use latest tesla_dashcam from pypi instead of requiring downloading it first.
• Fixed:
  • Issue with GPU type check of qsv for Linux. Contributed by cjwang18
  • ffmpeg error when swapping front/rear and excluding front or rear
  • ffmpeg error when swapping left/right and excluding left or right
  • Traceback if duration in video clip from Tesla is N/A, now clip will be excluded. Issue #212 <https://github.com/ehendrix23/tesla_dashcam/issues/212>_
  • When using front-layout, will now use ratio of clip instead of fixed 4/3 Issue #214 <https://github.com/ehendrix23/tesla_dashcam/issues/214>_
  • Will now force SAR to be 1:1 same as Tesla clips.
  • Camera width/height scale calculation to prevent zero multiplication when scale is not set
  • Alignment (halign/valign) defaults not being applied when not explicitly specified
  • Typo in qsv arg check. Contributed by @cjwang18
  • Docker: Add tesla_dashcam to non-accelerated container image. Contributed by @croadfeldt

Release v0.1.21.b4

• New:
  • Option --camera_position to provide coordinates where the camera clip should be positioned in the video providing very granular custom layout options. Using this option results in ignoring the layout option.
  • Option --camera_order to define the order the cameras should be processed. This allows to overlay one camera over another one and define which one should be on top.
  • Left and right-pillar cameras have been included as options and within the different layouts.
  • If an event does not have files for a certain camera (i.e. pillar cameras) then resulting layout will be as if those cameras were excluded.
  • Added street name from events.json, this became available with Tesla 2025.38.6
  • Added option --ffmpeg_debug so that when running with loglevel debug, FFMPEG output is not provided unless this is enabled as well.
  • Docker: Added pre-release docker image making it easier to use pre-release versions.
• Changed:
  • Many internal changes, including how layouts are calculated, typing, etc.
  • Docker: Reduces Docker image sizes by 24-66% by switching from jrottenberg/ffmpeg to Debian's FFmpeg package on python:3-slim base. Contributed by @magicalyak
  • Docker: Now will use latest tesla_dashcam from pypi instead of requiring downloading it first.
• Fixed:
  • Issue with GPU type check of qsv for Linux. Contributed by cjwang18
  • ffmpeg error when swapping front/rear and excluding front or rear
  • ffmpeg error when swapping left/right and excluding left or right
  • Issue with GPU type check of qsv for Linux. Contributed by @cjwang18
  • ffmpeg error when swapping front/rear and excluding front or rear
  • ffmpeg error when swapping left/right and excluding left or right
  • Traceback if duration in video clip from Tesla is N/A, now clip will be excluded. Issue #212 <https://github.com/ehendrix23/tesla_dashcam/issues/212>_
  • When using front-layout, will now use ratio of clip instead of fixed 4/3 Issue #214 <https://github.com/ehendrix23/tesla_dashcam/issues/214>_
  • Will now force SAR to be 1:1 same as Tesla clips.
  • Typo in qsv arg check. Contributed by @cjwang18
  • Docker: Add tesla_dashcam to non-accelerated container image. Contributed by @croadfeldt

Release v0.1.21.b3

• New:
  • Option --camera_position to provide coordinates where the camera clip should be positioned in the video providing very granular custom layout options. Using this option results in ignoring the layout option.
  • Option --camera_order to define the order the cameras should be processed. This allows to overlay one camera over another one and define which one should be on top.
  • Left and right-pillar cameras have been included as options and within the different layouts.
• Fixed:
  • Issue with GPU type check of qsv for Linux. Contributed by cjwang18
  • ffmpeg error when swapping front/rear and excluding front or rear
  • ffmpeg error when swapping left/right and excluding left or right

Release v0.1.21b2

• New:
  • Option --camera_position to provide coordinates where the camera clip should be positioned in the video providing very granular custom layout options. Using this option results in ignoring the layout option.
  • Option --camera_order to define the order the cameras should be processed. This allows to overlay one camera over another one and define which one should be on top.
• Fixed:
  • Issue with GPU type check of qsv for Linux. Contributed by cjwang18
  • ffmpeg error when swapping front/rear and excluding front or rear
  • ffmpeg error when swapping left/right and excluding left or right

Release v0.1.20

• BREAKING CHANGE:

  • How offsets are calculated has been changed and can impact result if negative values were being provided for start_offset and/or end_offset!

• New:

  • Option --sentry_start_offset to set the starting offset specifically for sentry based events
  • Option --sentry_end_offset to set the ending offset specifically for sentry based events
  • Added support for Intel VAAPI GPU acceleration (https://trac.ffmpeg.org/wiki/Hardware/VAAPI). Contributed by timrettop.
  • Two new dockerfiles have been added. One to support NVIDIA and other one to support VAAPI. Contributed by magicalyak

• Changed:

  • Updated Docketfile. Contributed by magicalyak

• Fixed:

  • Resolved issue on Linux with default device initialization failure when using Intel GPU acceleration and multiple usable devices by adding new qsv gpu_type (https://trac.ffmpeg.org/ticket/7649). Contributed by timrettop.

Release v0.1.19

• Fixed:
  • Tracebacks when there is no events.json file issue #168
  • Exit going through preferences within RunPreferences if an execution of tesla_dashcam failed

Release v0.1.18

• Fixed:
  • Traceback when running executable without providing --ffmpeg option.

Release v0.1.17

Release 0.0.17

• New:

  • Option --text_overlay_fmt to set the overlay text for the video. Contributed by JakeShirley
  • Option --timestamp_format for formatting timestamps.
  • Option --sentry_offset to set the start and end offset based on Sentry event timestamp.
  • Option --merge_template to allow merging of video files grouped based on this. Resulting movie filename will be based on this template.
  • Option --set_moviefile_timestamp to set the video file timestamp on the OS to start, end, or time of Sentry event.
  • Option --keep-events to keep event video files after they have been merged into a video.
  • Option --display_ts to show timestamps in the text output. This does not impact video output but is handy when using a monitor option. Contributed by croadfeldt
  • Option --title_screen_map, which generates a map of the event location and displays it for the first 3 seconds of the movie.
  • Added support for event information file and ability to display it in the overlay text. Contributed by JakeShirley
  • Support for FreeBSD 11. Contributed by busbyjon
  • Source can now include wildcards, shell variables, and will do user expansion (i.e. ~ on Unix, ~user on Windows).
  • Output and temp_dir can now include shell variables and will do user expansion (i.e. ~ on Unix, ~user on Windows).
  • Metadata tag title in video file is now set to reason for event (if exist) and timestamp or start/end timestamp
  • Metadata tag creation_time in video files created is now set to start timestamp of that particular video.
  • When scanning folders a message will be printed after every 10 folders scanned to show progress.
  • --gpu and --no-gpu are now valid arguments irrespective of platform.
  • event and final movie files will now include GPS coordinates with the location for an event. For movies this will be for the 1st event within the movie.

• Changed:

  • Improvement for Docker file size and stability. Contributed by magicalyak
  • Choice values for parameters (i.e. FULLSCREEN, intel, black) are now case-insensitive.
  • Updated supporting libraries to latest available.

• Fixed:

  • When providing an invalid start or end timestamp will now result in a error instead of a traceback.
  • Added x265 compatibility tag for QuickTime. Contributed by dburkland
  • Event file will now be removed when providing an output file and only 1 event is processed, leaving only 1 movie file.
  • Providing a mount as a source resulted in it no files found. Now when a mount is provided it will be handled same as folders.

Release v0.1.17b6

• Added debug information to show what platform and processor were identified, can help in troubleshooting.
• When system information returned is MacOS Intel then check actual processor with sysctl to ensure it is Intel and not Apple Silicon
• When running the binary on Apple Silicon provide a message to download an Apple Silicon ffmpeg instead and use --ffmpeg parameter.

Unless something comes up this will be turned into final release v0.1.17

Release v0.1.17b5

• --no-gpu and --gpu is now available irrespective of platform. GPU is still default on Intel Macs, for all others it has to be enabled (see README).
• Will now detect when running on Apple Silicon if Python installed is a Universal2 binary (or tesla_dashcam compiled with Apple Silicon Python binary) and disable GPU by default due to issues with ffmpeg on Apple Silicon.