Contract us
HarmonyOS Integrated NetEase Cloud Audio & Video: Bulk User Beauty Status Management of Beauty SDK for Entertainment Live Streaming
Updated：2026-05-19
Bulk User Beauty Status Management for Multi-User Mic Linking Live Streaming on HarmonyOS
In multi-user mic linking interactive scenarios of entertainment live streaming, efficiently and accurately managing beauty status for dozens or even hundreds of connected audiences is a core challenge directly affecting live streaming performance and user experience. The traditional one-to-one management mode will rapidly cause memory surge and CPU overload when coping with massive simultaneous mic linking interactions between hosts and audiences, further leading to severe live stream stuttering and frame drops. This article elaborates on how to design and implement a bulk user beauty status management solution oriented to massive concurrent scenarios on HarmonyOS, ensuring stable and smooth LetMagic beauty effects for all participants in large-scale interactive live broadcasts.
Core Challenge: Complexity Upgrade from Individual Management to Group Management
When live rooms shift from host solo broadcasting to real-time multi-user mic linking, the complexity of beauty management rises exponentially:
Multiplier Effect of Resource Consumption
Each connected user requires independent beauty processor instances, running buffers, as well as face detection and rendering pipelines. Creating separate instances for every user leads to multiplied resource overhead. Hundreds of connected users will exhaust mobile device memory and GPU resources instantly.
Massive Concurrency for Status Synchronization
Frequent user access and exit, plus real-time beauty parameter adjustments such as unified filter switching for all audiences by hosts, generate massive status synchronization requests. Traditional polling and event broadcast mechanisms will trigger severe message storms and system latency.
Fluctuating Computing Load
Intensive beauty processing is required for all user video frames during group singing or collective performances, while only a few participants need effect processing during host monologues. The system must intelligently perceive and adapt to such fluctuating computing pressure.
Difficulty in Balancing Experience Consistency
Under limited hardware resources, it is tricky to maintain acceptable beauty effects for all connected users. Developers need to formulate refined strategies to decide whether to prioritize image quality for hosts and core guests or allocate resources evenly to all users.
Architecture Design: Shared Resource Pool & Status Distribution Center
This solution abandons the traditional mode of maintaining independent full-function beauty instances for individual users, and adopts a new architecture featuring centralized resource scheduling and differentiated status application, consisting of two core modules.
1. Shared Beauty Resource Pool
Instead of building heavyweight independent beauty processors for each user, the system initializes controllable high-performance beauty computing core pools during system startup or live room initialization, which act as exclusive beauty computing clusters for live rooms.
- Intelligent Scheduling: When users join mic linking, the system marks their video streams and assigns them to computing cores in the pool for time-sharing processing. Efficient context switching enables simultaneous frame processing for multiple users.
- Dynamic Scaling: The scale of resource pools is adjustable. The system dynamically changes the number of active computing cores according to the number of online users, video resolution and device load including CPU, GPU usage and temperature. Spare cores are activated under high load and switched to low-power standby mode when pressure eases off.
2. User Status Center & Differentiated Renderer
Serving as the core hub to realize personalized beauty effects for all users, this centralized manager maintains all user beauty configurations.
- Unified Storage: It stores lightweight key-value formatted beauty files of all connected users in memory, recording beauty switch status, skin smoothing intensity, face slimming level, filter selection and other exclusive parameters of each user.
- Bulk Status Distribution: When global configuration changes occur such as applying unified star-glow filters to all audiences, the manager broadcasts unified instructions to all computing cores in the resource pool instead of sending separate commands to each user. All eligible video streams apply new effects synchronously to achieve atomic bulk status update.
- Differentiated Rendering: When processing video frames, each computing core first queries complete valid parameters from the user status center, then loads corresponding effect modules and intensity values for rendering. The same computing core can efficiently render personalized beauty effects for different users in turn.
Key Technical Implementation: In-depth Utilization of HarmonyOS Features
- Efficient Memory Sharing Mechanism: Leverage HarmonyOS optimized distributed soft bus and high-efficiency IPC to minimize memory copies during video frame collection, pool computing and encoding streaming, realizing zero-copy or single-copy high-efficiency data transmission.
- Priority-based Dynamic Task Scheduling: Set high priority for core beauty computing threads to guarantee real-time performance; assign lower priority to user configuration management and UI interaction tasks to ensure overall system responsiveness. Host video streams are granted the highest priority to maintain ultra-smooth core viewing experience.
- GPU Batch Rendering Optimization: Merge post-processing requests including filter overlay and sharpening for multiple user video frames at the GPU rendering stage. Complete multi-frame processing via one single rendering command to greatly reduce GPU instruction overhead and status switching frequency.
Performance & Experience Guarantee Strategies
- Hierarchical Degradation Mechanism: The system triggers automatic performance degradation under high pressure such as frame rate drop and overheating. It firstly lowers processing resolution for non-host users, then disables non-core special effects like background blurring temporarily, and finally retains basic skin beautification effects for marginal users to prioritize image quality of hosts and core guests.
- Resource Preheating & Caching: Preload and cache commonly used filters and special effect resources during live room initialization to avoid initial stuttering caused by resource loading when users join mic linking.
- Status Recovery & Synchronization: Instantly restore historical beauty configurations from the user status center for users who reconnect after temporary network disconnection, achieving seamless consistent visual experience.
Conclusion
Realizing bulk user beauty status management on HarmonyOS entertainment live streaming apps represents a technical upgrade from extensive resource stacking to refined service scheduling. By building shared computing resource pools and centralized status managers, developers can converge massive personalized beauty demands into an efficient, controllable and dynamically scalable processing framework.
