Official Kuaishou Account
This article explains the principles of remote photoplethysmography (rPPG), how facial skin color changes caused by heartbeats can be captured by a camera to measure heart rate, respiration, SpO₂ and other physiological signals, and reviews traditional and data‑driven algorithms for robust signal extraction.
‘All‑Things AR’ (万物AR) is a Kuaishou Y‑tech solution that lets users capture any real‑world object with a phone, automatically segments it using a custom AI model, and renders an animated 3D avatar via a lightweight SLAM‑based pipeline, enabling low‑cost, high‑quality AR experiences.
This article presents a method for fine‑grained control of highlight effects in 3D material makeup, analyzing the impact of material properties, lighting, and geometry, and proposes using model‑space normal maps to ensure stable, adjustable highlights across different facial models and poses.
This article explains how various noise algorithms—value, gradient, simplex, cellular, and FBM—are applied to short‑video visual effects, showcases shader implementations and image examples, and compares GPU random‑number generators to help developers choose the right balance of performance and visual quality.
SnowflakeNet introduces a novel Snowflake Point Deconvolution architecture combined with a Skip‑Transformer to explicitly split and refine points, enabling high‑quality reconstruction of fine local geometry in incomplete point clouds and outperforming prior methods on both dense and sparse benchmarks.
This article explains the challenges of background distortion caused by portrait beautification effects, describes Kuaishou Y‑Tech’s line‑segment‑based optimization framework that preserves line slopes and triangle shapes, and demonstrates the algorithm’s effectiveness through before‑and‑after visual results.
This article explains how Kuaishou's Y‑Tech team designs a server‑side video‑effects platform that tackles high computational load, real‑time constraints, and resource limits by adopting asynchronous workflows, task‑queue scheduling, and cloud‑native serverless frameworks such as Netflix Conductor and Knative.
QuanTaichi introduces a new language abstraction and compiler system that quantizes simulation data, dramatically reducing memory and bandwidth usage so that high‑precision physical effects—once requiring multiple GPUs—can now run on a single GPU, even on mobile devices.
The article details Kuaishou Y‑Tech’s end‑to‑end solution for converting a single RGB image into an interactive 3D photo on mobile devices, covering depth estimation, image‑inpainting, custom KwaiNN inference, and real‑time 3D rendering techniques that run on all smartphone models without depth sensors.
The article details Kuaishou Y‑tech's YKit AI SDK architecture, its unified interface, modular design, performance optimizations, and three real‑world case studies that illustrate how the SDK enables large‑scale, high‑quality short‑video effects across diverse devices while addressing challenges of effect variety, performance, and cost.
