AR Face Prop is Meishe’s all-in-one interactive solution: face beautification, sticker props, background division, and other interactive play types in a single SDK. Instead of assembling a beauty engine from one vendor, stickers from another, and segmentation from a third, your product gets the full effect set from one engine, one integration, and one content pipeline.
1. Product Positioning
AR Face Prop runs on the cross-platform audio/video processing engine used across the product family, with real-time rendering without pre-processing. It supports face beautification, sticker props, background division, and other types of features, providing a variety of interactive play. The platform set covers iOS, Android, Web, Desktop (Windows), and smart hardware; React Native and Flutter join via the short-video solution. It is a layer for your own app — the camera UI, permissions, and upload flow remain yours.
2. What is in the box
| Capability | What it does in your app |
|---|---|
| Face beautification | Skin smoothing, face reshaping, and makeup applied live in the camera preview |
| Sticker props | 2D and 3D stickers anchored to detected facial features — glasses sit on the nose, crowns sit on the head, 3D props follow head turns |
| Background division | Real-time separation of subject and background, so users change scenes without a green screen |
| Interactive play | Gesture-triggered effects and responsive props — a raised hand starts an effect, a wave switches the sticker |
| Filters | Colour-grading filters with adjustable intensity alongside the AR features |
All capabilities read from one detection result per frame. A user wearing beautification, a 3D sticker, and a replaced background generates one face analysis — the frame budget stays close to flat as the visible effect set grows.
3. Why one SDK instead of three
Assembling effects from multiple vendors works on paper and hurts in production:
- Three integrations become three maintenance tracks — each update, each platform port, each device quirk handled separately.
- Effects from different engines never quite match — detection timing and rendering paths differ, and users notice on close inspection.
- Content pipelines fork — three authoring flows, three delivery mechanisms, three release rhythms.
One SDK turns that into one of each: one integration on one engine, one consistent render, one content pipeline in the Asset Creation Platform. Campaigns ship as packages that every capability picks up simultaneously.
The snippet below is schematic. Names and signatures are illustrative, not the shipped API.
// Illustrative: beautification, a 3D prop, and background division together.
import { createEffectsRuntime } from "@meishe/ar-effects"; // placeholder name
const rt = await createEffectsRuntime({ platform: "ios" });
await rt.looks.apply("look.natural.01"); // beautification
await rt.props.attach("prop.crown.party.3d"); // sticker prop
await rt.segments.setBackground("scene.confetti"); // background division4. Interactive play that keeps users in the camera
The interactive layer is what turns an effects feature into a product loop. Gesture detection gives hands the controls — raise to start, wave to switch, clench to advance a template step. Props respond to the face they are attached to, and because gesture detection shares the same per-frame analysis as the other effects, adding interaction adds no second detection pass.
For product teams, this is the difference between a camera users visit and a camera users stay in: effects that respond, in real time, on every platform the product ships on.
5. Why teams choose Meishe
| Advantage | What it means for your project |
|---|---|
| All-in-one coverage | Beautification, stickers, background division, and interactive play from one SDK — no vendor assembly |
| Self-developed engine | Meishe’s own cross-platform audio/video processing engine, developed and refined in-house across the product family |
| Real-time, no pre-processing | Effects the moment the camera opens |
| Small footprint | A small core; optional modules on demand |
| Content pipeline included | Asset Creation Platform for in-house authoring; Special Effects Customization for produced collections |
6. What shipping teams ask first
- Which parts of AR Face Prop do we enable first? Whichever matches your launch feature set — the rest are modules you switch on later from the same library.
- Does the interactive layer need its own integration? No — gesture detection is part of the same stack and shares the same detection pass.
- Who builds our sticker library? Your designers in the Asset Creation Platform, or Meishe’s Special Effects Customization service for produced collections.
| Your product | What AR Face Prop covers |
|---|---|
| Social camera app | Beautification and sticker props as the daily core; seasonal campaigns as packages |
| Commerce and try-on flows | Beautification plus segmentation for clean product focus |
| Education and community apps | Interactive play that keeps sessions in the camera |
| Enterprise capture tools | Background division for professional framing without studio setup |
Each row uses the same SDK and the same content pipeline — the difference is which modules you enable first and which packages your designers author.
However the product plan starts, the SDK investment compounds in one direction: more packages, more platforms, more play — same engine, same pipeline.
The same compounding applies to campaigns: each seasonal package builds on the assets and conventions of the last, so the library gets richer while the effort per campaign stays flat.
Between launch and the second year, the pattern holds: every capability you switch on later — segmentation, gestures, a new sticker family — joins a stack your team already operates.
The practical starting move is small: enable beautification and one sticker family, publish a first campaign through your own channel, and read the engagement numbers. Everything after that — background division for a commerce push, gesture triggers for a community feature, a 3D prop line for a seasonal event — is a module switch and a package publish on a stack the team already knows.
That is the quiet advantage of the all-in-one arrangement: growth never requires a re-platforming decision, because the capabilities you grow into were present from the first integration. The SDK you linked on day one is the SDK running your second-year flagship campaign.
And because every campaign lands as data, the operating cost of staying ambitious stays low: a bigger library is more packages to curate, not more code to maintain.
FAQ
Can we start with just beautification and stickers?
Yes. Background division and interactive play are modules you enable when the roadmap calls for them — same SDK, same packages.
Do 3D props run on every platform?
Yes — 2D and 3D sticker sets are part of the shared capability set, resolving on every covered platform from the same packages.
How does background division work without a green screen?
Segmentation runs in real time on the camera frame, separating subject from background as part of the same rendering stack — no physical setup, no manual masking.
Can effects respond to more than gestures?
Props anchored to facial features respond to head movement, and gesture events trigger actions — the interactive vocabulary covers the common camera-app patterns.
What does an update to our sticker library involve?
Packages are data: publish through your own channel and every platform picks up the new content — the app binary stays untouched.