How to Upscale and Enhance Video Quality with FFmpeg (2026 Guide)

Got a blurry 480p video you want to make HD? Old footage that looks noisy and soft? FFmpeg provides filters for upscaling, denoising, sharpening, and color correction—but there’s a hard ceiling on what traditional algorithms can achieve.

This guide shows you what’s possible with FFmpeg, explains the limitations, and introduces an AI-powered alternative that delivers dramatically better results.

Prerequisites: Install FFmpeg

FFmpeg

macOS:

brew install ffmpeg

Ubuntu/Debian:

sudo apt update && sudo apt install ffmpeg

Windows: Download from ffmpeg.org, extract, and add bin to PATH.

Upscale Video Resolution

Basic Upscale to 1080p

ffmpeg -i input_480p.mp4 -vf "scale=1920:1080" -c:v libx264 -crf 18 -c:a copy output_1080p.mp4

Upscale to 4K

ffmpeg -i input_1080p.mp4 -vf "scale=3840:2160" -c:v libx264 -crf 18 -c:a copy output_4k.mp4

Keep Aspect Ratio

Scale to 1080p height, auto-calculate width:

ffmpeg -i input.mp4 -vf "scale=-1:1080" -c:v libx264 -crf 18 -c:a copy output.mp4

Scaling Algorithms

FFmpeg offers several scaling algorithms. The choice significantly affects quality:

# Lanczos (best quality, slowest)
ffmpeg -i input.mp4 -vf "scale=1920:1080:flags=lanczos" -c:v libx264 -crf 18 output.mp4

# Bicubic (good balance)
ffmpeg -i input.mp4 -vf "scale=1920:1080:flags=bicubic" -c:v libx264 -crf 18 output.mp4

# Spline (sharp, good for upscaling)
ffmpeg -i input.mp4 -vf "scale=1920:1080:flags=spline" -c:v libx264 -crf 18 output.mp4

# Bilinear (fastest, lowest quality)
ffmpeg -i input.mp4 -vf "scale=1920:1080:flags=bilinear" -c:v libx264 -crf 18 output.mp4

Algorithm	Quality	Speed	Best For
lanczos	Best	Slowest	Final output
spline	Very good	Slow	Upscaling
bicubic	Good	Medium	General use
bilinear	Fair	Fastest	Previews

Reduce Video Noise

hqdn3d (Fast Denoiser)

ffmpeg -i input.mp4 -vf "hqdn3d=4:4:3:3" -c:v libx264 -crf 18 output.mp4

Parameters: luma_spatial:chroma_spatial:luma_temporal:chroma_temporal

Higher values = more denoising (but more blur)
Start with 4:4:3:3 and adjust

nlmeans (Better Quality, Slower)

ffmpeg -i input.mp4 -vf "nlmeans=s=3:p=7:r=15" -c:v libx264 -crf 18 output.mp4

s=3 — denoising strength (1–30)
p=7 — patch size
r=15 — research window size

Sharpen Video

unsharp Filter

ffmpeg -i input.mp4 -vf "unsharp=5:5:1.0:5:5:0.5" -c:v libx264 -crf 18 output.mp4

Parameters: luma_x:luma_y:luma_amount:chroma_x:chroma_y:chroma_amount

luma_amount — sharpening strength (negative values blur)
Start with 1.0 for mild sharpening, 1.5 for strong

cas Filter (Contrast Adaptive Sharpening)

ffmpeg -i input.mp4 -vf "cas=0.5" -c:v libx264 -crf 18 output.mp4

Range: 0 (no sharpening) to 1 (maximum)
0.4–0.6 is usually a good range

Color and Brightness Correction

Adjust Brightness, Contrast, Saturation

ffmpeg -i input.mp4 -vf "eq=brightness=0.06:contrast=1.2:saturation=1.3" -c:v libx264 -crf 18 output.mp4

brightness — -1.0 to 1.0 (default 0)
contrast — -1000 to 1000 (default 1.0)
saturation — 0 to 3.0 (default 1.0)

Auto-Level Colors

ffmpeg -i input.mp4 -vf "normalize" -c:v libx264 -crf 18 output.mp4

Stabilize Shaky Video

FFmpeg’s vidstab filter requires two passes:

Pass 1: Analyze motion

ffmpeg -i input.mp4 -vf vidstabdetect -f null -

This creates a transforms.trf file.

Pass 2: Apply stabilization

ffmpeg -i input.mp4 -vf vidstabtransform=smoothing=10:input=transforms.trf -c:v libx264 -crf 18 output.mp4

smoothing=10 — higher values = smoother (but more cropping)

Combine Multiple Filters

Chain filters together for a full enhancement pipeline:

ffmpeg -i input.mp4 -vf "scale=1920:1080:flags=lanczos,hqdn3d=4:4:3:3,unsharp=5:5:0.8:5:5:0.4,eq=contrast=1.1:saturation=1.2" -c:v libx264 -crf 18 -c:a copy output.mp4

This pipeline:

Upscales to 1080p with Lanczos
Applies denoising
Sharpens
Boosts contrast and saturation

The Hard Truth About FFmpeg Upscaling

Here’s what FFmpeg cannot do:

Recover lost detail — Upscaling 480p to 4K just stretches existing pixels. No new detail is created.
Reconstruct faces — Blurry faces stay blurry, just bigger.
Remove compression artifacts — JPEG/H.264 blocking artifacts get amplified.
Generate textures — Hair, fabric, skin texture cannot be hallucinated from a low-res source.

Traditional scaling algorithms (even Lanczos) are interpolation — they calculate pixel values between existing ones. The result is a larger image that’s smoother but not actually sharper or more detailed.

The AI Alternative: WaveSpeed Desktop Video Enhancer

This is where AI upscaling fundamentally changes the game.

WaveSpeed Desktop includes an AI Video Enhancer powered by deep learning models that:

Generate real detail — reconstruct textures, edges, and fine features that don’t exist in the original
Upscale up to 4x — turn 480p into true 1080p, or 1080p into 4K with actual new information
Enhance faces — restore facial details, sharpness, and clarity
Remove noise intelligently — distinguish noise from detail (unlike FFmpeg’s blur-everything approach)

The difference isn’t incremental—it’s a generational leap. What FFmpeg gives you in 10 lines of filter chains, AI enhancement does in one click with dramatically better results.

WaveSpeed Desktop Video Enhancer

Download WaveSpeed Desktop for free: https://github.com/WaveSpeedAI/wavespeed-desktop/releases

FAQ

Can FFmpeg truly upscale video to 4K? It can increase the resolution to 4K, but it won’t add real detail. The video will be 4K in pixel count but not in actual visual quality. AI upscaling (like WaveSpeed’s Video Enhancer) generates new detail that makes the result genuinely sharper.

What’s the best FFmpeg scaling algorithm? Lanczos for quality, bicubic for speed. But even the best traditional algorithm can’t match AI-based upscaling.

Does denoising always help? Not always. Denoising removes noise but also removes fine detail. If your video is clean, denoising will make it softer. Only use it on visibly noisy footage.

How long does FFmpeg enhancement take? A full pipeline (upscale + denoise + sharpen) on a 10-minute 480p video can take 30–60 minutes on modern hardware. AI upscaling with GPU acceleration is typically faster and produces better results.

Can I enhance old VHS or DVD footage? FFmpeg can clean it up somewhat, but the results are limited. AI enhancement tools are specifically trained on degraded footage and produce dramatically better restorations.