FaceWinUnlock-Tauri: 2D Face Unlock for Windows 10 and 11

What FaceWinUnlock-Tauri Changes

FaceWinUnlock-Tauri addresses a specific gap in Windows hardware support. Some Windows 10 and 11 computers have cameras that do not meet the requirements for Windows Hello face recognition, so the system cannot offer the built-in unlock experience even when the device is capable of capturing ordinary video. The project adds an alternative path by placing a custom Credential Provider DLL into the Windows login flow.

The intended users are individual Windows users who want convenience on a personal home computer or development machine. The README explicitly places the software in low-security situations where convenient login is more important than the stronger protections of hardware-backed biometrics. It is not presented as a replacement for Windows Hello in environments holding sensitive data.

The design carries a major security tradeoff. The project uses 2D facial recognition rather than Windows Hello's 3D structured-light or infrared methods. The README warns that a photograph or video may bypass this kind of recognition. Liveness detection is included, and version 0.2.2 introduced a model identified in the README as the RGB face liveness detection model from the Tongyi laboratory. Later update notes say that liveness detection was improved, but they also record that its performance still could not match version 0.2.2. That admission makes liveness detection a configurable defense rather than a guarantee.

The project also distinguishes account passwords from Windows PINs. Users must enter the password belonging to the Windows account. A PIN used to log in through the normal Windows interface is not accepted by the software.

Credential Provider and Recognition Data Flow

The repository layout separates the main application areas into `UI/`, `Server/`, and `Unlock/`. The README describes a Vue 3 and Element Plus interface, an OpenCV-based face-recognition backend, and a custom Credential Provider DLL. The visible description also refers to Rust code, although the project changed its source-distribution policy after 2026-03-01. The available repository listing does not provide enough implementation detail to describe the complete IPC protocol or credential submission sequence, so the exact internal calls cannot be confirmed here.

The user-facing flow is clear from the installation guide. Initialization checks camera permissions and the registry environment. The software then locks an account, waits five seconds, and expects face recognition to complete initialization. After that, preferences select a camera, the face-management interface enrolls one or more faces, and account association records an alias, Windows account type, username, and account password.

Recognition is invoked when the Windows screen is locked. The README gives the keyboard shortcut below.

Win + L

After the screen locks, moving the mouse or pressing a keyboard key starts recognition, unless the configured delay has not yet elapsed. The software compares camera input with enrolled faces and submits the associated account credentials when recognition succeeds. The release notes describe fixes to a case where an incorrect username or password could cause credentials to be sent repeatedly to LSA, as well as fixes for DLL unloading while recognition was still running.

The changelog also records support for multiple faces, multiple accounts, local and online accounts, multi-domain accounts, and graded recognition contexts including startup, lock screen, UAC, and the user layer. Those are documented features, but the README does not state that every Windows security layer behaves identically or provide a compatibility matrix.

Installation Through the Initialization Wizard

There is no package-manager installation command in the README. The project directs users to downloadable Windows builds through external file-sharing links, then describes a graphical initialization process. The repository's top-level source entry `UI/` is a Vue 3 application, while the README says the Rust core is no longer published for current versions. A new user should therefore not assume that cloning the repository and running a documented development command will produce the same installer as a release.

Before starting, the README says to read the risk warning. Initialization checks camera permissions and the registry environment, and it strongly recommends photographing the program's recovery notices during step two. The software locks the account, waits five seconds, and attempts an automatic unlock. The user should not manually unlock during this interval.

The documented first-use sequence is:

1. Run the application and complete system initialization. 2. Open Preferences and select a camera device. 3. Open `面容管理`, then choose `添加新面容`. 4. Associate the enrolled face with an account. Enter the alias, Windows account type, username, and password. 5. Lock Windows and test recognition by moving the mouse or pressing a keyboard key.

The README does not document a command-line installation, a supported camera API list, or a precise rollback command if the initialization fails. The absence of a recovery command is especially important because the program modifies registry state and interacts with `Winlogon`.

Windows Security and Login Constraints

The most important limitation is the documented possibility of a failed DLL or incorrect path configuration preventing the Windows login interface from displaying normally. In an extreme failure, the user may be unable to reach the desktop. The project does not publish a text-based rollback procedure in the README, so recovery depends on the program's warnings, photographs, and normal Windows recovery options available before initialization.

Password handling is another constraint. Version 0.4.0 added password encryption storage, and the README says that the software accepts an account password rather than a PIN. It also warns that repeated password errors can cause Microsoft to lock the account and advises uninstalling the software if that occurs. A user who does not know the underlying account password should not configure automatic face-to-account association.

The software also supports an application password and timeout-based reauthentication, according to the changelog. Those features protect access to the application itself, but they do not change the underlying weakness of 2D face matching. The application password should not be confused with the Windows account password.

The README's recommended scope is narrow: Windows 10 and Windows 11 64-bit systems, with Windows 7 64-bit explicitly described as untested. The project is designed for personal computers and development machines where security requirements are lower. Its restriction against high-confidentiality offices and servers is a functional boundary, not merely a warning about installation complexity.

FaceWinUnlock-Tauri Compared with Windows Hello

Windows Hello is the more secure default when the hardware supports it. Its documented distinction is hardware-backed 3D structured-light or infrared liveness checking, while FaceWinUnlock-Tauri uses 2D facial recognition and an optional software liveness model. Windows Hello therefore offers a stronger security basis for systems that meet its hardware requirements.

FaceWinUnlock-Tauri's practical distinction is compatibility. It targets Windows devices with ordinary cameras that do not support the native Windows Hello face-unlock path. The software adds enrollment, account association, delayed recognition, retries, and a custom login interface outside the native Windows Hello flow. That flexibility comes with additional components: a DLL, registry changes, `Winlogon` interaction, camera configuration, and application-managed account credentials.

The comparison is therefore not simply feature versus no feature. A device supported by Windows Hello avoids the need for FaceWinUnlock-Tauri's 2D recognition and login-interface modification. FaceWinUnlock-Tauri is more relevant when the hardware limitation is real and the user accepts the weaker authentication model. Its custom integration may also create more failure modes than the native path, especially during initialization, DLL loading, camera capture, and shutdown.

Release History and Upgrade Cost

The latest listed release is 0.4.0 on 2026-10-01. The last push was also on 2026-10-01, so the repository was not archived and had recent development activity at that point. Earlier releases include 0.3.7 on 2026-08-27 and 0.3.6 on 2026-08-07.

Upgrades can require more than replacing an application executable. The release notes for versions including 0.2.0, 0.2.2, 0.3.0, 0.3.5, 0.3.6, 0.3.7, and 0.4.0 state that reinitialization is required. Version 0.4.0 specifically says that users should uninstall the previous software before updating and initialize again. Other changes include moving unlock settings under the DLL, removing the boot recognition and unlock service switches, and removing service logs. A user must account for configuration migration, re-enrollment, account association, and testing after such changes.

The repository is licensed under AGPL-3.0. That matters for anyone examining or redistributing the code, although the README does not provide legal interpretation of the license. The project also became closed-source after 2026-03-01 because of reports that copies were being sold under other names. The repository says that the remaining source is version 0.3.2 with the core Rust code removed, while later versions continue to be delivered through releases and repository updates. That makes source-based review of the current core impossible from the disclosed repository state.

Editorial conclusion

FaceWinUnlock-Tauri may suit a personal Windows 10 or 11 computer where ordinary webcams cannot provide Windows Hello and convenience matters more than strong authentication. It should not be adopted for high-confidentiality, server, or security-sensitive workstations. Verify first whether your account password is available, whether the selected camera and Windows version match the documented environment, and whether you understand the recovery procedure for a failed Credential Provider before initialization.

Frequently asked questions

Official sources

1. Issues
2. License: AGPL-3.0
3. README
4. Releases
5. zs1083339604/FaceWinUnlock-Tauri on GitHub