FFmpeg CRF Values: Understanding Per-Title Encoding for Optimal Video Quality

When it comes to determining the complexity of an asset and setting individual encoding settings, a brute force approach is often employed. This is precisely what Netflix does when encoding their movie trailers. As a result, some trailers are easier to encode and less complex than others, requiring different bitrate values to achieve a good PSNR (Peak Signal-to-Noise Ratio) value.

The chart below illustrates the resulting PSNR and bitrate values at a resolution of 1080p for various movie trailers encoded with the same Constant Rate Factor (CRF) values. The findings show that different movie trailers have varying requirements, with some achieving a 'good' PSNR value of 40-45 at bitrates of 1Mbit - 2Mbit, while others require around 3Mbit - 6Mbit.

One key aspect of per-title encoding is its ability to adapt to the complexity of different video content. This approach recognizes that certain types of content, such as movies or sport streams with a lot of movement, require lower resolutions to maintain a good quality of experience for viewers. In contrast, easy-to-encode videos can be delivered at much lower bitrates while improving perceptual quality.

Compared to traditional approaches, which apply the same predefined encoding ladder to all types of content, per-title encoding has the potential to significantly decrease storage and delivery costs of video streams. This is because different types of video content require unique bitrate and encoding settings to achieve a certain quality.

To better understand per-title encoding and reproduce Netflix's results using ffmpeg, it's essential to comprehend the concept and its implications. In 2016, Netflix introduced the idea of per-title-encoding in a blog post and detailed paper, which provides valuable insights into their findings and how they can be interpreted.

By recognizing that different types of video content have distinct requirements for bitrate and encoding settings, per-title encoding offers a more effective way to achieve optimal video quality. By using ffmpeg to apply individual encoding settings based on the complexity of each asset, you can improve the overall viewing experience while reducing storage and delivery costs.

Ffmpeg crf values