FFmpeg Static Build with NVENC Support

Build FFmpeg with maximum static linking for all codecs, while keeping NVIDIA libraries dynamically linked (as required). This creates a highly portable FFmpeg binary with NVIDIA GPU acceleration support.

Features

✅ Multi-Distro Support: "Works" on Debian, Ubuntu, RHEL, Fedora, CentOS, Arch Linux, openSUSE
✅ Static Linking: All codec libraries statically linked with -fPIC for PIE compatibility
✅ NVIDIA Support: CUDA, NVENC, CUVID hardware acceleration (10x+ realtime encoding)
✅ VAAPI Support: Intel/AMD hardware acceleration
✅ Latest Versions: FFmpeg 8.0.1, NASM 2.16.03, AOM 3.13.1, and latest stable codecs
✅ Production Ready: Tested with H.264 and HEVC NVENC encoding

Supported Codecs

Video Codecs

H.264/AVC: libx264 (software), h264_nvenc (NVIDIA), h264_vaapi (VAAPI)
H.265/HEVC: libx265 (software), hevc_nvenc (NVIDIA), hevc_vaapi (VAAPI)
VP8/VP9: libvpx
AV1: libaom (software encoder/decoder), av1_nvenc (NVIDIA - RTX 40 series+)

Audio Codecs

MP3: libmp3lame
AAC: libfdk-aac
Opus: libopus
Vorbis: libvorbis
Theora: libtheora

Requirements

Linux distribution with one of these package managers: apt, dnf, yum, pacman, or zypper
NVIDIA GPU (optional, for NVENC support)
NVIDIA drivers installed (optional, for GPU acceleration)
Root/sudo access for installing dependencies

Quick Start

Build Everything

Run the master build script to build all components:

cd scripts
./build-all.sh

This will:

Install system dependencies
Install NVIDIA CUDA toolkit
Build all codec libraries
Build FFmpeg with all features
Run validation tests

Individual Component Build

You can also build components individually:

cd scripts

# Install dependencies
./dependencies.sh

# Install NVIDIA CUDA
./nv-cuda-sdk.sh

# Build assemblers
./nasm.sh
./yasm.sh

# Install NVIDIA codec headers
./nv-codec-header.sh

# Build video codecs
./x264.sh
./x265.sh
./libvpx.sh
./aom.sh

# Build audio codecs
./opus.sh
./lame.sh
./libfdk-aac.sh

# Build FFmpeg
./ffmpeg.sh

# Validate build
./validation.sh

Configuration

Edit vars.sh to customize:

# Library versions
NASM_VERSION="3.01"
X264_BRANCH="master"
X265_VERSION="4.1"
VPX_VERSION="v1.15.2"
FFMPEG_VERSION="8.0.1"
# ... and more

# Build directories
WORK_DIR="/tmp/sources"    # Where source files are downloaded
DEST_DIR="/usr/local"      # Where FFmpeg will be installed

# CUDA location
CUDA_DIR="/usr/local/cuda"

# Enable CUDA validation tests
VALIDATE_CUDA="false"  # Set to "true" to test CUDA encoding

# Library versions
NASM_VERSION="2.16.03"      # Updated for AOM compatibility
X265_VERSION="4.1"
VPX_VERSION="v1.15.2"
AOM_VERSION="v3.13.1"
FFMPEG_VERSION="8.0.1"

Supported Distributions

Debian/Ubuntu

# Uses apt package manager
# Tested on Debian 12 (Bookworm) and Ubuntu 22.04/24.04

RHEL/Fedora/CentOS

# Uses dnf/yum package manager
# Tested on Fedora 39+ and RHEL 8/9

Build Troubleshooting

NVENC Not Working

If NVENC encoders are not available:

Check NVIDIA drivers:
```
nvidia-smi
```

Verify CUDA libraries:

ls -la /usr/local/cuda/lib64/libnvcuvid.so*

Check FFmpeg NVENC support:

/usr/local/bin/ffmpeg -encoders | grep nvenc

Library Path Issues

If you get library errors:

# Update library cache
sudo ldconfig

# Check library paths
echo $LD_LIBRARY_PATH
ldd /usr/local/bin/ffmpeg

AOM/NASM Build Failures

If AOM build fails with NASM errors:

NASM 2.16.03+ is required for AOM v3.13+
The build scripts use the correct version
Rebuild NASM: sudo ./scripts/nasm.sh

Position Independent Code (PIC) Errors

If you get relocation errors during FFmpeg build:

All static libraries must be built with -fPIC
The scripts automatically add these flags
Rebuild the failing library with: sudo ./scripts/<library>.sh

Arch Linux

# Uses pacman package manager
# Tested on Arch Linux (rolling release)

openSUSE/SLES

# Uses zypper package manager
# Tested on openSUSE Leap/Tumbleweed

Output

The FFmpeg binary will be located at:

/usr/local/bin/ffmpeg

Check Static/Dynamic Linking

To verify which libraries are statically vs dynamically linked:

ldd /usr/local/bin/ffmpeg

Expected dynamic dependencies:

linux-vdso.so.1 (kernel interface)
libcuda.so.1 (NVIDIA CUDA)
libnvcuvid.so.1 (NVIDIA Video Decoder)
libnvidia-encode.so.1 (NVIDIA Encoder)
Standard system libraries (libc, libm, libpthread, libdl)

All codec libraries (x264, x265, vpx, opus, etc.) should NOT appear in the output.

Performance

NVENC Hardware Encoding

Expected performance with NVENC (actual speeds vary by GPU):

H.264 NVENC: 10-15x realtime @ 1080p
HEVC NVENC: 8-12x realtime @ 1080p
Power efficiency: ~90% less CPU usage vs software encoding
Quality: Comparable to software encoding at higher bitrates

Software Encoding

libx264: 1-4x realtime (depends on preset)
libx265: 0.5-2x realtime (slower but better compression)
libaom-av1: 0.1-0.5x realtime (best compression, very slow)

Usage Examples

Software Encoding (CPU)

# H.264 encoding with libx264
/usr/local/bin/ffmpeg -i input.mp4 -c:v libx264 -preset medium -crf 23 output.mp4

# H.265 encoding with libx265
/usr/local/bin/ffmpeg -i input.mp4 -c:v libx265 -preset medium -crf 28 output.mp4

Hardware Encoding (NVENC)

# H.264 NVENC encoding
/usr/local/bin/ffmpeg -i input.mp4 -c:v h264_nvenc -preset fast -b:v 5M output.mp4

# HEVC NVENC encoding
/usr/local/bin/ffmpeg -i input.mp4 -c:v hevc_nvenc -preset fast -b:v 5M output.mp4

# NVENC with hardware decoding
/usr/local/bin/ffmpeg -hwaccel cuda -hwaccel_output_format cuda -i input.mp4 -c:v h264_nvenc -preset fast output.mp4
/ffmpeg-static/bin/ffmpeg -i input.mp4 -c:v libx265 -preset medium -crf 28 output.mp4

# VP9 encoding
/ffmpeg-static/bin/ffmpeg -i input.mp4 -c:v libvpx-vp9 -b:v 2M output.webm

# AV1 encoding
/ffmpeg-static/bin/ffmpeg -i input.mp4 -c:v libaom-av1 -b:v 2M output.mp4

Hardware Encoding (NVIDIA NVENC)

# H.264 with NVENC
/ffmpeg-static/bin/ffmpeg -hwaccel cuda -i input.mp4 -c:v h264_nvenc -preset p4 -b:v 5M output.mp4

# H.265 with NVENC
/ffmpeg-static/bin/ffmpeg -hwaccel cuda -i input.mp4 -c:v hevc_nvenc -preset p4 -b:v 5M output.mp4

# With CUDA hardware acceleration
/ffmpeg-static/bin/ffmpeg -vsync 0 -hwaccel cuda -hwaccel_output_format cuda \
  -i input.mp4 -c:v h264_nvenc -b:v 5M output.mp4

Hardware Encoding (VAAPI - Intel/AMD)

# H.264 with VAAPI
/ffmpeg-static/bin/ffmpeg -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 \
  -i input.mp4 -c:v h264_vaapi -b:v 5M output.mp4

# H.265 with VAAPI
/ffmpeg-static/bin/ffmpeg -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 \
  -i input.mp4 -c:v hevc_vaapi -b:v 5M output.mp4

Troubleshooting

NVIDIA CUDA Not Found

If you see CUDA-related errors during build:

Check CUDA installation:
```
ls -la /usr/local/cuda
nvcc --version
```
Update CUDA_DIR in vars.sh if CUDA is in a different location
Ensure NVIDIA drivers are installed:
```
nvidia-smi
```

Package Manager Not Supported

If your distribution isn't automatically detected:

Check dependencies.sh for the list of required packages
Install them manually using your package manager
Continue with the build scripts

Build Failures

If a component fails to build:

Check the error message
Ensure all dependencies are installed
Check if you have enough disk space (build requires ~5GB)
Try building components individually to isolate the issue

Version Information

Current Versions (as of January 2026):

FFmpeg: 8.0.1
x264: latest (master branch)
x265: 4.1
libvpx: 1.15.2
AOM: 3.13.1
NASM: 3.01
NVIDIA Codec Headers: 13.0.19.0

License

This build script collection is provided as-is. Individual components (FFmpeg, x264, x265, etc.) have their own licenses:

FFmpeg: GPL/LGPL (depending on configuration)
x264: GPL
x265: GPL
libfdk-aac: Fraunhofer FDK AAC (check license restrictions)
Others: See individual component licenses

Contributing

To update versions, edit vars.sh and update the version variables.

Credits

Maintained by: Daniel Ramirez Based on FFmpeg static build practices for NVENC support.

Name		Name	Last commit message	Last commit date
scripts		scripts
.gitignore		.gitignore
README.md		README.md

Latest commit

History

Folders and files

Repository files navigation