About Dynaface
What is Dynaface?
Dynaface is a free, open-source research tool for measuring facial symmetry and facial movement from photographs and video using AI-based facial landmark detection. It has been used in peer-reviewed studies of idiopathic facial paralysis, oral-ocular synkinesis, and facial weakness in facioscapulohumeral muscular dystrophy; see Research.
Which platforms does it run on?
Mac, iPhone, and iPad (one app on the App Store), Windows 10 and 11 (the Microsoft Store), and Android 8.0 or later (a direct download). Every version is free and computes the same measurements. For scripting and batch work there is also a Python library and a .NET library.
Does it analyze video?
Yes. Dynaface analyzes every frame of an MP4, MOV, or M4V video (up to 5,000 frames), charts each measurement over the length of the clip, and exports one CSV row per frame. On Android, video analysis requires Android 9 or later. Video is imported from a file, so record the clip with the device's camera app first.
What does it measure?
Facial symmetry measures including the Facial Asymmetry Index (FAI), oral commissure excursion, brow height, eye area (palpebral fissure), and dental display; facial distances such as intercanthal distance and mouth length; head pose; and, from profile photos, the nasofrontal angle, nasolabial angle, and nasal tip projection. The Measurements page defines each one.
How many landmarks does it detect?
98 on a frontal view, using the WFLW landmark scheme. On a profile view it traces the outline of the face and marks six anatomical points along it.
Can measurements be reported in millimeters?
Yes. Dynaface converts pixels to millimeters from the distance between the pupils, assuming the population-average pupillary distance of 63 mm. For accurate absolute values, enter the subject's measured pupillary distance in Settings. Ratios and angles do not depend on this setting.
Is Dynaface a registered medical device?
No. Dynaface is not a registered medical device and is not intended to diagnose, treat, cure, prevent, or monitor any disease or medical condition; it provides informational facial symmetry measurements only. AI-based measurements can contain errors and are affected by image quality, lighting, pose, facial expression, camera angle, and landmark detection accuracy, so verify results independently before relying on them. See Legal.
Data and Privacy
Are images or videos uploaded to a server or cloud service?
No. Dynaface is not designed to transmit data to external servers. All image and video processing happens locally on the device running the app.
Is anything stored outside the local device?
No. Dynaface keeps its library of analyzed photos and videos, the measurements, and its settings on the device, and no feature writes data to a remote location. There is no account to create.
Does analysis require internet access?
No. Analysis runs entirely offline; the neural network models are installed with the app. The only network activity Dynaface starts is opening the online user guide in your web browser when you choose it from the menu, and anything the browser itself does is outside Dynaface.
What output does Dynaface produce?
CSV files of the measurements, annotated images, annotated MP4 video (on Mac, Windows, iPhone, and iPad), and plain text copied to the clipboard. All output is written where you choose on your own device; none of it is transmitted by the app.
Research Use
Are there licensing or data-use considerations for clinical research?
Dynaface is released under the Apache License 2.0, which governs permitted use, redistribution, and disclaimers of warranty; the project imposes no additional licensing. Institutional requirements such as IRB approval and HIPAA review are the responsibility of the researcher's institution and are not addressed by the license.
How do I cite Dynaface?
Cite Renne, Heaton, and Boahene (2026) in Laryngoscope. The full reference and a BibTeX entry are on the Research page, along with the neural networks whose authors should also be cited.
Can I process many files automatically?
Yes. The Python library
(pip install dynaface) computes the same measurements from code, and
the .NET library does the
same for C# applications. The
source code is on GitHub.