This article explains how a simple CSS mask‑image technique—using a single mask picture and a few style rules—can replicate Bilibili’s subtitle effect that avoids covering on‑screen characters, includes a complete demo, performance notes, and links to related CSS properties.
This article explains how to design a live‑streaming danmaku (bullet‑screen) feature by using Redis Zset as the underlying data structure, limiting the list to the latest ten messages, and handling retrieval via polling APIs or WebSocket, with Kafka for message buffering and a distributed lock to avoid ordering issues.
This article details the design and optimization of a high‑throughput live‑streaming danmaku system, covering background requirements, bandwidth challenges, short‑polling versus WebSocket delivery, compression and frequency controls, service splitting with caching and lock‑free ring buffers, and reports successful handling of 700 k concurrent users during a major event.
Liu Jun explains how Bilibili’s engineers replaced traditional pre‑processed SVG masks with a real‑time, browser‑based machine‑learning pipeline—using MediaPipe SelfieSegmentation, OffscreenCanvas, and Web Workers—to extract human contours and block faces in danmaku, achieving roughly 5 % CPU load on a 2020 M1 MacBook.
This article details the challenges and solutions for building a scalable, low‑latency danmaku service for live streaming, covering background requirements, bandwidth constraints, protocol choices, short‑polling implementation, reliability measures, and performance results that supported 700,000 concurrent users during a major event.
This article details the design, bandwidth optimization, and reliability engineering behind a custom live‑streaming danmaku service that supports up to one million simultaneous users, covering problem analysis, compression techniques, polling strategies, service splitting, and performance results from a major traffic event.
To support Southeast Asian live-streaming, we designed a custom danmaku system capable of handling up to a million concurrent users per room, tackling bandwidth pressure, weak-network latency, and reliability by employing HTTP compression, response simplification, short-polling, local caching, and lock-free ring buffers.
This article describes the challenges of delivering real‑time danmaku for a Southeast Asian live‑streaming service, analyzes bandwidth pressure, network instability, and reliability issues, and presents a series of backend optimizations—including HTTP compression, response simplification, short‑polling, service splitting, caching, and a custom ring‑buffer—to reliably support up to one million concurrent users.
This article describes the design, challenges, and optimization strategies of a custom live‑streaming danmaku system for Southeast Asian markets, covering bandwidth constraints, latency issues, long‑polling versus WebSocket approaches, service splitting, caching, and a ring‑buffer implementation that supported 700 k concurrent users during a major sales event.
To support Southeast Asian live streaming with up to a million concurrent users per room, we replaced Tencent Cloud’s inadequate danmaku service with a custom solution that optimizes bandwidth via HTTP compression, response simplification, ordering, frequency control, and a short‑polling delivery mechanism, while ensuring reliability through service splitting, caching, and lock‑free ring buffers.
The paper describes how Bilibili transformed its danmaku service from a simple, limited‑recall pipeline into a ten‑fold larger, KV‑store‑backed recommendation architecture that unifies engineering and AI layers, uses dynamic sharding and Redis locks, and ultimately boosts recall pool size, exposure, and experiment speed while reducing downgrade rates.
This article reveals the technical design of Youku's high‑performance danmaku rendering engine, covering its background, industry landscape, GPU‑based rendering pipeline, multi‑layer architecture, performance‑boosting techniques, real‑world benchmarks, and the innovative effects it enables.
This article explains the design and implementation of a flexible danmaku SDK that uses AI‑based image segmentation to let comments flow around people in videos, covering architecture, algorithmic processing, server deployment, client rendering options, performance challenges, and future extensions.
This tutorial explains how to locate Bilibili video danmaku (bullet‑comment) APIs, extract the CID, and use Python libraries such as requests, BeautifulSoup, and pandas to download, clean, and save the comment data to CSV files, with an optional API‑based shortcut.
This article explains the concept of danmaku (bullet comments), why they improve user experience, and provides a detailed guide on implementing a danmaku system using HTML + CSS or Canvas, including stage design, track management, collision handling, and reusable code examples.
This article explains the design and implementation of a flexible, high‑performance bullet‑comment (danmaku) system for long‑form video on iOS, covering canvas rendering, track management, polling, view reuse, animation control, gradient effects, data handling, and integration with ads and user interactions.
iQiyi’s community strategy defines a core of users, diverse content formats, and strict platform mechanisms, evolving danmaku features and anti‑spoiler tools while avoiding low‑value shortcuts, and leverages IP‑driven campaigns such as meme festivals and fan‑selected performances to boost engagement without compromising the core viewing experience.
