Command Palette

Search for a command to run...

UnylyUnyly
Весь каталог

Lumina

БесплатноНе проверен

Lumina: An AI-powered Active Capture system for high-fidelity 3D Head Avatars. Uses real-time pose guidance and 3D Gaussian Splatting to transform standard webc

GitHubEmbed

Описание

Lumina: An AI-powered Active Capture system for high-fidelity 3D Head Avatars. Uses real-time pose guidance and 3D Gaussian Splatting to transform standard webcam feeds into animatable digital twins.

README

Lumina logo

Lumina

Lumina is a desktop-first active-capture system for fast 3D head reconstruction.
V1 is focused on a practical local workflow: capture a single neutral head from a laptop webcam, fit a coarse full-head model with rough hair coverage, and export a GLB on macOS.

V1 Goal

  • Single-person only
  • Neutral head only
  • Laptop webcam only
  • Full head with rough hair
  • GLB export only
  • Local reconstruction on macOS
  • Target turnaround under 5 minutes

Current Architecture

  • apps/director-desktop Electron shell for local capture, session control, and desktop runtime integration
  • apps/director-web React + Vite + Three.js renderer embedded inside Electron
  • services/recon-api Optional local orchestration service for manifests, pipeline control, and future decoupling
  • services/recon-worker Local reconstruction worker for FLAME fitting and GLB export
  • packages/protocol Shared TypeScript protocol and event contracts

Product Direction

Lumina v1 is intentionally not chasing maximum fidelity. The current target is a robust and fast local capture loop that:

  • guides the user through a fixed neutral pose sequence
  • auto-accepts stable frames
  • stores a local JSON session manifest
  • prepares a local reconstruction handoff

The long-term direction can still include higher-end Gaussian Splatting pipelines, but that is not the default production path for the first macOS release.

Local Development

Requirements:

  • Node.js 20+
  • pnpm
  • Python 3.11+
  • uv
  • macOS with webcam access

Install and run:

corepack enable
corepack prepare [email protected] --activate
pnpm install
uv sync
pnpm dev

Important pnpm Note

On some machines, pnpm blocks install/build scripts by default.
If Electron fails with Electron failed to install correctly, approve build scripts and rebuild:

pnpm approve-builds
pnpm rebuild
pnpm --filter director-desktop exec electron --version
pnpm dev

Approve at least:

  • electron
  • esbuild

Repository Notes

  • Logo asset: apps/director-web/public/assets/lumina-logo.jpeg
  • Local capture sessions are written under ~/Documents/LuminaSessions
  • The desktop app currently includes:
    • webcam preview
    • pose sequence state
    • automatic frame acceptance scaffold
    • manual retry
    • local manifest writes through Electron IPC

Status

This repository currently contains the desktop-first scaffold, capture-loop foundation, local IPC bridge, and reconstruction boundaries for Lumina v1.

from github.com/oemercemtabar/Lumina

Установить Lumina в Claude Desktop, Claude Code, Cursor

Рекомендуется · одна команда, все IDE
unyly install lumina

Ставит в Claude Desktop, Claude Code, Cursor и VS Code — сам разбирается с npx, uvx и сборкой из исходников.

Впервые? Поставь CLI: curl -fsSL https://unyly.org/install | sh

Или настроить вручную

Выполни в терминале:

claude mcp add lumina -- uvx lumina

Пошаговые гайды: как установить Lumina

FAQ

Lumina MCP бесплатный?

Да, Lumina MCP бесплатный — установка в пару кликов через Unyly без оплаты.

Нужен ли API-ключ для Lumina?

Нет, Lumina работает без API-ключей и переменных окружения.

Lumina — hosted или self-hosted?

Self-hosted: сервер запускается локально на твоей машине командой из раздела установки.

Как установить Lumina в Claude Desktop, Claude Code или Cursor?

Открой Lumina на unyly.org, выбери вкладку своего клиента (Claude Desktop, Claude Code, Cursor) и нажми Install — конфиг сгенерируется автоматически, без правки JSON.

Похожие MCP

Compare Lumina with

Не уверен что выбрать?

Найди свой стек за 60 секунд

Автор?

Embed-бейдж для README

Похожее

Все в категории ai