Watch Face Format Always On Display Assessor

Command-line tool for estimating AOD compliance of Wear OS Watch Face Format screenshots or mockups.

Prerequisites

• A reasonably recent Python installation.

• Python's pillow extension (install with pip install pillow).

• To capture screenshots using adb, the adb executable needs to be accessible via the PATH environment variable, or via the ANDROID_HOME environment variable (in subdirectory platform-tools).

Usage

aod.py aod.py [-h] [-s] [-n] [-b B] [-c] [-v] [source]

source      filename (.png); if unspecified, capture from adb
-d          display diagnostic information
-s          watchface is square (default: round)
-n          consider all non-black pixels to be fully used
-b B        bleed factor
-c          calculate bleed factor from white circle
-h, --help  show this help message and exit
-v          show program's version number and exit

If source is unspecified, aod.py will attempt to capture a screenshot from a connected device/emulator using adb. If it succeeds, the screenshot will be saved as adb.png. You can inspect that image to verify that the correct screenshot was obtained. adb.png will be overwritten when required; if you want to keep any such image, move/copy/rename it before rerunning aod.py.

Android Studio Integration

If you use Android Studio for your project, you can optionally configure Android Studio to run the script. To do so:

• In Android Studio, click the Run/Debug Configurations drop-down and select Edit Configurations....

• Click the Add New Configuration button and select Shell Script.

• Configure the script like this (but adapted to your folder locations):

You should now be able to check AOD compliance using Android Studio's Run button. This won't rebuild or rerun your project, but will display aod.py's output in Android Studio's Terminal tool window.

Note

When used in this way, the script relies on adb. If adb can't be found, see here.

Assumptions

Up to 15% of pixels within the display may be used.

The height and width of the watch face (specified in watchface.xml) are both 450px. (The size of the image passed to aod.py need not be that size.)

Variable Intensity vs. Non-Black

By default, aod.py assumes that partially-illuminated pixels make a partial contribution to illumination. For example, if all pixels displayed #808080, overall illumination would be about 50%. Transparency (opacity, alpha) is also taken into account; eg, if all pixels displayed #80ffffff, overall illumination would be about 50%.

If you specify the -n command line option, an alternative algorithm is employed which considers that non-black pixels are fully used; eg, if all pixels displayed #01000001, overall illumination would be 100%.

Bleed

The actual area of circular images generated by a watch or AVD is not exactly what you'd expect by calculating the area based on the watch face diameter. This may be because the image is discrete (pixels) rather than a perfect smooth circle.

Presumably, Google assesses pixel usage against the actual area rather than the calculated area. aod.py attempts to account for this by multiplying the calculated area by a 'bleed' factor. For variable-intensity calculations (the default), a bleed factor of 0.995 is normally used. For non-black calculations (command-line -n), a bleed factor of 1.0015 is normally used. Those built-in bleed factors are only approximate; actual bleed factors depend on screen size, watch type, and probably other considerations.

If you want to specify your own bleed factor, use the -b command-line option.

aod.py can calculate what bleed factor should be used for a particular image source. To do this, capture an all-white circular image from the AVD or watch (and don't expect those to be identical!). Pass that image to aod.py and specify the -c command-line option. Subsequently, use the provided bleed factor with the -b command-line option for images from the same source. To test the process, rerun aod.py with the same all-white image but with the calculated bleed factor; the image's average intensity or non-black usage should be almost exactly 100%.

Limitations

• Only provides approximate results because the exact details of Google's testing aren't known.

• Doesn't attempt to deduce shape of watch face (so use -s command-line option if necessary).

• Doesn't support rectangular images; eg, from "Wear OS Rectangular" AVD.

• Doesn't support greyscale-formatted images.

Acknowledgements

• Concept and initial code is based on Grégoire Sage's Fitbit AOD utility.

• Android Studio integration is based on Turtlepaw's Clockwork.