What Is CRF in Video Encoding? Quality and File Size Explained

---

When encoding a video, you may see a setting called CRF, or Constant Rate Factor. It can seem like an obscure number, but it controls an important trade-off: how much data the encoder uses to preserve visual detail. Understanding CRF helps you make more informed choices when creating a smaller copy, archiving a video, or preparing a file for sharing.

CRF is not a universal quality score, and there is no single value that suits every video. The result depends on the encoder, its settings, the source footage, and your own tolerance for visible compression. This guide explains the basic idea and offers a practical way to test settings.

What does CRF mean?

CRF stands for Constant Rate Factor. It is a quality-based rate-control method used by encoders such as x264 and x265. Instead of telling the encoder to use a fixed amount of data for every second, you give it a target level of quality. The encoder then uses more data for complex or detailed scenes and less for scenes that are easier to compress.

For example, a quiet shot of a plain wall may be encoded with fewer bits than a fast-moving scene with fine texture. This flexibility can make file sizes more efficient than applying one fixed bitrate to every part of a video. The final file size is an outcome of the content and chosen setting, not a precise target.

How CRF affects quality and file size

On common CRF scales, a lower number asks the encoder to preserve more detail. That usually means a larger file. A higher number allows more compression, which usually reduces file size but can make artifacts more noticeable. The exact direction and range depend on the encoder, so check the documentation for the software you use.

There is no fixed file-size change for moving from one CRF value to another. A short clip with simple backgrounds may stay small, while detailed foliage, film grain, rapid movement, or visual noise can require considerably more data at the same setting. Two videos encoded with the same CRF can therefore have very different sizes.

CRF also does not promise that a video will look identical to its source. It is a way to guide a lossy encoding process. Compression may remove information, and whether the difference is noticeable depends on the footage, display, viewing distance, and the encoder settings.

CRF versus a fixed bitrate

With a fixed bitrate, you specify a data rate, such as a certain number of kilobits per second. This gives you more control over the approximate amount of data used per second, which can be useful when a delivery platform or storage limit sets a size constraint. However, a fixed bitrate does not guarantee consistent visual quality across scenes: complex footage may look worse than simple footage at the same rate.

CRF works in the other direction. You specify a quality target, and the data rate varies as needed. This is convenient when you care more about a consistent viewing experience than a predictable file size. If you must meet a strict size or bandwidth limit, bitrate-based encoding or a constrained rate-control mode may be more suitable.

Choosing a starting CRF value

For x264, values around 18 to 23 are commonly used as a starting range for general-purpose encoding. That is a starting point, not a guarantee of a particular appearance or file size. Some projects need a lower value to retain subtle detail; others can use a higher value without visible problems in the intended viewing conditions.

Do not assume the same number will produce the same quality in x265 or another encoder. CRF scales and encoder behavior differ. Even when two encoders use settings with similar names, their output may not be visually equivalent. Test the specific encoder and preset you plan to use.

A practical test workflow

1. Choose a representative section of the source. Include difficult material if possible, such as motion, fine textures, shadows, or gradients.

2. Encode that section at two or three nearby CRF values using the same encoder, preset, resolution, and other settings.

3. Compare the versions on a display similar to the one your audience will use. Look for blockiness, smearing, lost texture, banding, or distracting changes around moving edges.

4. Check the resulting file sizes, then select the highest compression level that still meets your needs. If the choice is close, test a longer section before encoding the full video.

Use the same playback conditions for each comparison. Switching between different players or display settings can make it harder to judge whether the encoder caused a visible difference. For critical work, a frame-by-frame inspection can help reveal artifacts, but also watch the video at normal speed because some problems are more apparent in motion.

Why presets and source quality matter

An encoder preset usually adjusts how much time the encoder spends finding compression opportunities. A slower preset may take longer but can sometimes produce a smaller file at a similar visual quality. A faster preset may finish sooner but use more data for comparable results. Presets are not the same as CRF: CRF guides the quality target, while the preset influences the encoding process.

The source sets a limit on what an encode can preserve. If the original is already low quality or heavily compressed, re-encoding cannot restore missing detail. It can add further losses, and increasing the output resolution does not recreate information that is absent from the source. When possible, start with the highest-quality original you are permitted to use.

When CRF is not the right choice

CRF is useful when you want quality-oriented encoding and can accept a variable file size. It may be a poor fit when a job requires a specific maximum size, a steady delivery bitrate, or predictable bandwidth. In those cases, use a rate-control approach designed around the constraint, and inspect the result to make sure quality remains acceptable.

For online playback, the destination service may also re-encode uploaded video according to its own requirements. A carefully chosen CRF setting cannot control that later processing. Check the service’s current upload guidance when preparing a delivery file, and keep an original master if you may need to create other versions.

Key points to remember

CRF is a quality-oriented encoding control: lower values generally mean less compression and larger files, while higher values generally mean more compression and smaller files. The relationship is not a precise size calculator, and CRF numbers are not directly interchangeable between encoders. Test representative footage, keep the other encoding settings consistent, and choose based on both visual inspection and practical file-size needs.