@blendviewer/core

Developer-Built Data Operating System - Enable anyone to build their own storage, compute, and visualization services

简体中文 | Website | Documentation

What Are We? What Problems Do We Solve?

💡 Core Philosophy

Enable every developer to become a cloud service provider, not just a consumer.

Traditional cloud platforms (AWS, Alibaba Cloud):

• ❌ You can only use their provided services
• ❌ Want to customize? You need to build the entire infrastructure yourself
• ❌ Database + Storage + Compute + Queue = 6-12 months of development

BlendViewer Data Operating System:

• ✅ Provides infrastructure, you deploy your own services
• ✅ Upload code → Automatically deploy as cloud service
• ✅ Unified SDK, call your services in 3 lines of code
• ✅ Other developers can use the services you deploy

🎯 Core Problems We Solve

Lower the barrier to cloud service development, let developers focus on business logic.

// Traditional approach (6-12 months)
Build database + object storage + container orchestration + task queue + message push...

// BlendViewer (1-3 days)
import { OpenBase, FusionFlow, VisionScope } from '@blendviewer/core';

What Features Do We Have?

🗄️ OpenBase - Data Infrastructure (Co-build Storage Services)

You can:

• Use our database + object storage (out of the box)
• Or deploy your own data services

const db = OpenBase.createClient('your-project-id');

// Database operations
await db.users.create({ data: { name: 'Alice' } });
await db.users.findMany({ where: { age: { gt: 18 } } });

// Object storage (automatic chunking for large files)
await db.storage.upload({
  objects: [{ objectName: 'video.mp4', filePath: './8GB-video.mp4' }],
});

// Format conversion
await db.storage.convert({
  objects: [{ objectName: 'video.mov' }],
  options: { to: 'mp4', quality: 'high' },
});

// Data analysis
await db.olap.aggregate({
  table: 'sales',
  measures: ['COUNT(*)', 'SUM(revenue)'],
  dimensions: ['date', 'region'],
});

⚡ FusionFlow - Multi-Runtime Compute Platform (Co-build Compute Services)

You can deploy any compute service:

• AI inference (Stable Diffusion, LLM, speech recognition)
• Video processing (transcoding, editing, effects)
• Data processing (crawlers, ETL, reports)
• Scientific computing (simulation, modeling, optimization)

Deploy Your Service (3 Steps)

# 1. Create bv.yaml
runtime: container # or v8 | wasm | python
modes: [sync, async, stream]
dockerfile: ./Dockerfile

# 2. Upload and deploy
bv deploy

// 3. Anyone can call your service
const compute = FusionFlow.createClient({ projectId: 'your-service-id' });

// Synchronous call (< 5s)
const result = await compute.run({
  mode: 'sync',
  input: { text: 'Hello', operation: 'translate' },
});

// Async task (minutes/hours)
const task = await compute.run({
  mode: 'async',
  input: { videoUrl: 'https://...', operation: 'transcode' },
});
task.watch(({ progress }) => console.log(`Progress: ${progress}%`));

// Stream output (real-time)
const stream = await compute.run({
  mode: 'stream',
  input: { prompt: 'Tell me a story' },
});
for await (const chunk of stream) {
  process.stdout.write(chunk.text);
}

Platform automatically provides:

• ✅ Load balancing + auto-scaling
• ✅ Async task queue + auto-retry
• ✅ Real-time progress push
• ✅ Monitoring + logging + billing

🎨 VisionScope - Universal Visualization Engine (Co-build Renderers)

You can:

• Use our rendering engine (supports 80+ formats)
• Or register your own renderer

const viewer = new VisionScope({ container: '#viewer' });

// Automatic format detection
viewer.load('https://example.com/model.glb'); // 3D model
viewer.load('https://example.com/video.mp4'); // Video
viewer.load('https://example.com/report.pdf'); // PDF
viewer.load('https://example.com/photo.jpg'); // Image
viewer.load('https://example.com/code.py'); // Code (syntax highlighting)
viewer.load('https://example.com/data.zip'); // Archive preview

Supported formats:

• 📄 Documents: PDF, DOCX, XLSX, PPTX, TXT, MD
• 🖼️ Images: JPG, PNG, SVG, WebP, GIF
• 🎬 Video: MP4, WebM, OGG
• 🎵 Audio: MP3, WAV, AAC
• 💻 Code: JS, Python, Java, C++
• 📦 Archives: ZIP, RAR, 7Z
• 🎨 3D Models: GLB, GLTF, FBX, OBJ, STL

View complete format list →

Why Choose Us?

🆚 Comparison with Traditional Approach

| Scenario | Traditional Approach | BlendViewer | | ---------------------- | -------------------------------------------------------- | ------------------------------ | | Deploy Compute | Build container orchestration + registry + load balancer | Upload code, auto-deploy | | Async Tasks | Build task queue + Worker + retry logic | Built-in queue + auto-retry | | Real-time Progress | Build message push service | Platform auto-push | | Data Storage | Build database + object storage + CDN | One line db.storage.upload() | | Large File Upload | Write chunking + resume yourself | Platform auto-handles | | Visualization | Integrate multiple rendering libraries | One API supports 80+ formats | | Development Cycle | 6-12 months | 1-3 days | | Maintenance Cost | Multiple servers + ops team | Serverless, zero maintenance |

🌟 Core Advantages

1. Ultra-Low Barrier

// Complete full-stack in 3 lines of code
const db = OpenBase.createClient('project-id');
const compute = FusionFlow.createClient({ projectId: 'pod-id' });
const viewer = new VisionScope({ container: '#viewer' });

2. Developer Co-Building

• Your deployed services → Automatically become platform capabilities
• Other developers can call your services
• You can provide APIs externally and earn revenue

3. Unified SDK

• Same API for frontend/backend/mobile
• Node.js / Browser / React / Vue all supported
• TypeScript type-safe

4. Production-Grade Reliability

• Auto load balancing + failover
• Distributed task queue
• Global CDN acceleration
• Monitoring + logging + alerts

🚀 5-Minute Quick Start

Installation

npm install @blendviewer/core

Use Built-in Services

import { OpenBase, Auth } from '@blendviewer/core';

// 1. Authentication
await Auth.loginWithApiKey({
  projectId: 'your-project-id',
  apiKey: process.env.API_KEY,
});

// 2. Use database + storage
const db = OpenBase.createClient('your-project-id');
await db.users.create({ data: { name: 'Alice' } });
await db.storage.upload({
  objects: [{ objectName: 'photo.jpg', filePath: './photo.jpg' }],
});

Deploy Your Compute Service

# bv.yaml
runtime: container
modes: [sync, async]
dockerfile: ./Dockerfile

bv deploy  # Upload and auto-deploy

// Others can call your service
const compute = FusionFlow.createClient({ projectId: 'your-service-id' });
const result = await compute.run({
  mode: 'sync',
  input: { text: 'Hello' },
});

🎯 Use Cases

Any application that needs data + compute + visualization:

• 🎬 Video Platforms (transcoding, editing, AI subtitles)
• 🤖 AI Applications (image recognition, speech synthesis, LLM)
• 🏢 SaaS Applications (CRM, ERP, collaboration tools)
• 🛒 E-commerce Platforms (product management, image/video processing)
• 📊 Data Analysis (BI reports, visualization)
• 🎨 Creative Tools (design collaboration, 3D modeling, rendering)

📖 More Examples

Example 1: Build a Video Transcoding Service

// 1. Upload video
const db = OpenBase.createClient('video-platform');
const upload = await db.storage.upload({
  objects: [{ objectName: 'video.mov', filePath: './input.mov' }],
});

// 2. Call your deployed transcoding service
const transcoder = FusionFlow.createClient({ projectId: 'ffmpeg-transcoder' });
const task = await transcoder.run({
  mode: 'async',
  input: { videoUrl: upload[0].url, format: 'mp4' },
});

// 3. Monitor progress
task.watch(({ status, progress, result }) => {
  console.log(`${status}: ${progress}%`);
  if (status === 'completed') {
    // 4. Play online
    const viewer = new VisionScope({ container: '#player' });
    viewer.load(result.outputUrl);
  }
});

Example 2: Build an AI Image Processing Service

// 1. Upload image
const db = OpenBase.createClient('image-ai');
const image = await db.storage.upload({
  objects: [{ objectName: 'photo.jpg', filePath: './photo.jpg' }],
});

// 2. Call AI service (background removal, super-resolution)
const aiPod = FusionFlow.createClient({ projectId: 'stable-diffusion' });
const result = await aiPod.run({
  mode: 'sync',
  input: {
    imageUrl: image[0].url,
    operations: ['remove-background', 'upscale-4x'],
  },
});

// 3. Preview result
const viewer = new VisionScope({ container: '#result' });
viewer.load(result.outputUrl);

🌍 Cross-Platform Support

// Node.js (CommonJS)
const { OpenBase } = require('@blendviewer/core');

// ES Modules (Node.js / Browser / Vite / Next.js)
import { OpenBase } from '@blendviewer/core';

// React / Vue / Svelte
import { VisionScope } from '@blendviewer/core';

Supported Environments:

• ✅ Node.js 14+ (CommonJS & ESM)
• ✅ All modern browsers
• ✅ React, Vue, Angular, Svelte
• ✅ Next.js, Nuxt.js, SvelteKit
• ✅ Webpack, Vite, Rollup, esbuild

🔗 Resources

• 📚 Full Documentation
• 🎮 Examples: ModelVerse Navigation
• 🏠 Official Website
• 💬 Community Forum
• 🐛 Open Source

📄 License

MIT License

🤝 Join Us

We are building a developer-co-built data operating system.

• 💡 Share Your Ideas
• 👥 Join Community

Made with ❤️ by BlendViewer Community