AXedocs
Install

Screenshots & Video

Capture PNGs, record H.264 MP4s, stream frames to ffmpeg or your own reader, stream native H.264 from a device, and record one movie through iPhone Duo simulator folds and rotations.

Media commands print the file path on stdout. record-video and stream-video stop cleanly on Ctrl+C or SIGTERM.

Screenshots#

shell
axe screenshot --udid "$UDID" --output screen.png   # a file
axe screenshot --udid "$UDID" --output ./shots/     # an existing directory, auto-named
axe screenshot --udid "$UDID"                       # auto-named in the current directory

screenshot prints the PNG's path. With --output json the result is {path, pixelWidth, pixelHeight}:

shell
axe --output json screenshot --udid "$UDID" --output screen.png

Screenshots are in pixels. Divide by the screen scale that axe displays reports to get points for taps.

Video recording#

shell
axe record-video --udid "$UDID" --fps 15 --output recording.mp4
axe record-video --udid "$UDID" --fps 10 --quality 60 --scale 0.5 --output small.mp4
OptionRangeDefault
--fps1 to 30, an upper limit10
--quality1 to 100; higher uses a higher bitrate80
--scale0.1 to 11.0
--outputMP4 path or existing directoryaxe-video-<timestamp>.mp4

Press Ctrl+C to stop. AXe finishes the MP4 and prints its path. Stop with SIGINT or SIGTERM, not SIGKILL.

Video streaming#

shell
axe stream-video --udid "$UDID" --fps 10 --format mjpeg > stream.mjpeg
axe stream-video --udid "$UDID" --fps 30 --format ffmpeg | \
  ffmpeg -f image2pipe -framerate 30 -i - -c:v libx264 -preset ultrafast output.mp4
FormatOutput
mjpegMultipart JPEG (the default).
rawFor each frame, a 4-byte big-endian length, then a JPEG.
ffmpegConcatenated JPEGs for ffmpeg -f image2pipe.
bgraPacked BGRA pixels. A selected-screen stream prints its width and height on stderr.
h264Native H.264 Annex B, physical devices only.

Stream bytes go to stdout and diagnostics to stderr; keep them separate. --fps is an upper limit. stream-video refuses --output json, because stdout carries frames.

stdout is binary

Always redirect or pipe stream-video. Printing frames to a terminal will garble it.

Physical devices#

shell
axe screenshot --udid "$DEVICE_UDID" --output device.png
axe record-video --udid "$DEVICE_UDID" --fps 3 --scale 0.5 --output device.mp4
axe stream-video --udid "$DEVICE_UDID" --format raw --fps 3 --scale 0.5 > frames.bin
axe stream-video --udid "$DEVICE_UDID" --format h264 > device.h264

The screenshot stream supports mjpeg, raw, ffmpeg and bgra. Recording writes an H.264 MP4 after a clean stop; JSON is {path, pixelWidth, pixelHeight, frames}.

Native h264 uses the device's own encoder settings, so leave out --fps, --quality and --scale. Its frames keep the panel's native orientation; when they need turning, as on an iPad in portrait, stderr says how many degrees clockwise. Stopping waits for the session to end on the device, which can take about 20 seconds.

The device sends one native display stream at a time. Native h264 and Xcode Device Hub's screen view replace each other's stream, and AXe then exits with deviceStreamReplaced. Screenshots, the screenshot stream and recording work alongside Device Hub.

One screen, or following the active screen#

On any simulator, and especially the iPhone Duo simulator, you can pin a capture to one screen or follow the active one:

shell
axe displays --udid "$UDID"   # read the active screen's id and uuid

axe screenshot --udid "$UDID" --screen-id 3 --screen-uuid "$SCREEN_UUID" --output inner.png
axe record-video --udid "$UDID" --screen-id 3 --screen-uuid "$SCREEN_UUID" --output inner.mp4
axe stream-video --udid "$UDID" --screen-id 3 --screen-uuid "$SCREEN_UUID" --format raw > inner.bin
axe record-video --udid "$UDID" --follow-active-display --output folds.mp4

With --screen-id and --screen-uuid, a capture fails if that screen stops being active, instead of switching to another one. A selected screenshot never replaces an existing file.

--follow-active-display records one fixed-size movie that follows the active screen through folds and rotations. The first frame sets the size; later frames fit inside it upright, with black bars where they do not fill it. It is low-rate capture: it does not encode every frame of a transition animation. It cannot be combined with the screen pair.