NVR Command Center (Native Rust + GPU + Cloud P2P)

<0.9sParallel 12-Cam Startup<80msioctl(FIONREAD) Zero-Lag SyncCloud P2PZero-Config Remote Tunnel0% CPUOpenGL UV Shader Zoom (1x-5x)

System Architecture Flow

01Auto LAN Probe -> Cloud P2P PTCP Tunnel

0212+ Native Rust OS Threads (-fpsprobesize 0)

03Kernel Pipe ioctl(FIONREAD) Burst Drain

04eframe/egui OpenGL Texture Pool + UV Zoom

Quick Start (Ubuntu / Linux)

$ git clone https://github.com/pulkitmundra/nvr-command-center.git && cd nvr-command-center && ./install.sh

01 / Engineering & Implementation

Re-architected a 12+ channel NVR surveillance wall from Python/Tkinter into a standalone Native Rust binary (eframe/egui + OpenGL glow), cutting memory usage from ~650 MB to ~70 MB for 24/7 operation.
Engineered a native Rust Dahua / CP Plus / Amcrest Cloud P2P authenticated PTCP tunnel (AES-256-OFB + PBKDF2-HMAC-SHA256 + sequence-window deduplication) with automatic LAN -> Cloud P2P fallback to stream all 12 cameras from any internet connection with zero port forwarding.
Eliminated 12-20s sequential startup delays and multi-second TCP socket lag by spawning 12+ parallel native OS threads with zero-probe FFmpeg demuxing (-fpsprobesize 0) and Linux kernel pipe ioctl(FIONREAD) burst draining at 100% native wire rate (<80ms sync).
Implemented 0.0% CPU digital zoom & pan (1x-5x) via OpenGL UV texture coordinates, independent per-tile SD <-> HD and Single <-> Grid controls, a 3-tab multi-brand NVR configurator with 1-click channel auto-discovery, and composite 12-camera wall snapshots.

<0.9s parallel startup, <80ms synchronized glass-to-glass latency, zero-config Cloud P2P remote tunneling, 0.0% CPU UV shader zoom, and ~70 MB RAM across 12 concurrent streams