Accelerating Headless Video Renders with FFmpeg NVENC and Docker on Linux

1. CPU vs. NVENC Hardware Architecture

In software-only rendering, Chromium and FFmpeg compete for the same host CPU cycles. While Chromium uses threads to compute CSS transformations, layout boundaries, and WebGL frames, the CPU must also handle macroblock motion estimation, discrete cosine transforms (DCT), and entropy encoding.

NVIDIA NVENC is a standalone hardware subsystem integrated into NVIDIA graphics cards. When enabled, raw frame buffers stream directly into the GPU's fixed-function silicon encoder, leaving CPU cores dedicated exclusively to Chromium's React runtime:

2. Host Prerequisites & NVIDIA Container Toolkit

Before Docker containers can access the host GPU, the host Linux kernel requires the proprietary NVIDIA driver and the NVIDIA Container Toolkit runtime hook:

# 1. Add NVIDIA Container Toolkit repository
curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg
curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list | \
  sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
  sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list

# 2. Install and configure container runtime
sudo apt-get update
sudo apt-get install -y nvidia-container-toolkit
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker

# 3. Verify hardware passthrough inside basic container
docker run --rm --gpus all nvidia/cuda:12.4.1-base-ubuntu22.04 nvidia-smi

3. Dockerfile Configuration for GPU-Accelerated Nodes

Standard distribution packages of FFmpeg on Debian often omit the non-free NVENC headers. The production image builds upon an NVIDIA CUDA runtime layer, ensuring binary driver access and hardware codec passthrough:

FROM nvidia/cuda:12.4.1-runtime-ubuntu22.04 AS base

# Install Node.js 20, system fonts, and required graphics libraries
RUN apt-get update && apt-get install -y --no-install-recommends \
    curl \
    ca-certificates \
    ffmpeg \
    libnss3 \
    libatk-bridge2.0-0 \
    libdrm2 \
    libxkbcommon0 \
    libxcomposite1 \
    libxdamage1 \
    libgbm1 \
    fonts-liberation \
    fonts-noto-color-emoji \
    chromium-browser \
    && curl -fsSL https://deb.nodesource.com/setup_20.x | bash - \
    && apt-get install -y nodejs \
    && rm -rf /var/lib/apt/lists/*

# Expose all GPU devices and capabilities to container runtime
ENV NVIDIA_VISIBLE_DEVICES=all
ENV NVIDIA_DRIVER_CAPABILITIES=compute,video,utility
ENV PUPPETEER_EXECUTABLE_PATH=/usr/bin/chromium-browser

WORKDIR /app
COPY package*.json ./
RUN npm ci --omit=dev
COPY . .

ENTRYPOINT ["npx", "remotion", "render"]

4. Configuring Remotion CLI for Hardware NVENC Encoders

To route Remotion's internal frame pipe directly into the GPU encoder, pass the h264_nvenc codec flag alongside calibrated hardware preset switches:

# Production GPU render execution string
npx remotion render RootComposition out/reel-nvenc.mp4 \
  --codec=h264-nvenc \
  --pixel-format=yuv420p \
  --concurrency=6 \
  --scale=1 \
  --gl=angle

5. Production Performance & Throughput Benchmarks

Empirical benchmarks measured on a 60-second 1080x1920 60 FPS motion graphics composition (3,600 total frames) illustrate substantial scalability gains: