Achieving Good HEVC/CRF Rate Balance for Various Media Sizes and Formats

To achieve a good HEVC/CRF rate balance for various media sizes and formats, you must adjust the Constant Rate Factor based on source complexity rather than using a single value. Higher CRF numbers produce smaller files with lower quality, while lower numbers yield larger files with higher fidelity. By understanding how DVD, Blu-Ray, and digital downloads differ in bitrate and resolution, you can select the optimal CRF to maintain visual integrity without wasting storage space.

Understanding the CRF Scale

The Constant Rate Factor (CRF) is the most important encoding option in x265 because it allows for the best quality-to-filesize ratio. It operates on a logarithmic scale where the maximum value is 51, representing the worst quality, and the minimum is 0, which is lossless. For most practical purposes, a CRF of 23 is considered visually lossless or nearly so for standard definition content. As you decrease the number, the file size increases exponentially while quality improvements become marginal. Conversely, increasing the number reduces file size rapidly but introduces visible artifacts like blocking or banding. For high-quality archival, many users aim for a CRF between 18 and 20. This range ensures that even when viewed on large screens or during fast-paced action scenes, the image remains clean. If your goal is efficient streaming or limited storage, a CRF between 24 and 28 is often acceptable for casual viewing. However, going below 18 rarely provides noticeable benefits for human eyesight and wastes significant disk space. Going above 28 usually results in noticeable degradation, especially in darker scenes or complex textures.

Adjusting for Source Material: DVD vs. Blu-Ray

Different sources require different baseline settings due to their inherent quality and resolution. DVD content is typically 480p or 576i with heavy compression artifacts already present from the original MPEG-2 encoding. Re-encoding DVDs with a very low CRF is pointless because the source quality is the bottleneck. A CRF of 22 to 24 is usually sufficient for upscaling or repackaging DVDs. Using a lower CRF here only inflates the file size without adding detail that was not captured in the original broadcast. Blu-Ray sources are significantly different. They offer 1080p or 4K resolution with much higher bitrates and cleaner video data. To preserve the sharpness and dynamic range of a Blu-Ray rip, you should use a lower CRF, typically between 18 and 20. This ensures that fine details in shadows and highlights are retained. If you are converting a 4K Blu-Ray down to 1080p, you might be able to increase the CRF slightly to 20 or 21, as the downsampling process itself adds some smoothing that masks minor compression artifacts.

Handling Digital Downloads and Web Content

Video downloads vary wildly depending on the platform and codec used. Many web videos are encoded with variable bitrates and aggressive compression to save bandwidth. When re-encoding these files, it is crucial to analyze the source first. If the download is already heavily compressed, using a low CRF will result in massive file sizes with no visible improvement. In such cases, a CRF of 24 to 26 is often the sweet spot. For high-bitrate web releases, such as those found on premium streaming services, treat them similarly to Blu-Rays. Use a CRF of 18 to 20 to maintain clarity. However, be aware that some modern codecs like AV1 or VP9 may handle certain types of motion differently than HEVC. While this article focuses on HEVC, if you encounter issues with specific playback devices, you might need to refer to resources on troubleshooting unsupported codecs. You can learn more about handling tricky codec scenarios by reading Troubleshooting Dolby Vision Video File issues, as similar principles apply to format compatibility and quality preservation.

Optimizing for Specific Media Sizes

File size constraints often dictate the final CRF choice. If you are preparing videos for mobile devices with limited storage, you may need to push the CRF higher. However, simply raising the CRF is not enough; you should also consider reducing the resolution. Encoding a 4K video at CRF 20 for a phone screen is wasteful. Instead, encode at 1080p with a CRF of 22 to 24. This approach yields a much smaller file while maintaining excellent perceived quality on small displays. For archiving purposes where space is less of a concern, prioritize quality over size. Use a CRF of 18 or lower. You can also add two-pass encoding if you have strict size limits, although this is less common with CRF. Two-pass encoding analyzes the video first and then encodes it with a target bitrate, offering more predictable file sizes. However, for general use, CRF alone is superior because it adapts to scene complexity automatically. Complex scenes get more bits, while simple scenes get fewer, ensuring consistent quality throughout the video.

Practical Encoding Settings

When using tools like FFmpeg or HandBrake, ensure you are using the correct preset. The preset controls the speed of encoding and has a minor impact on efficiency. Slower presets like medium or slow provide better compression efficiency, meaning you can achieve the same quality with a slightly higher CRF (smaller file). For example, using the slow preset might allow you to use CRF 22 instead of 21 for the same visual quality. Always verify your output. Play back the encoded video and look for artifacts in dark areas, which are prone to banding in HEVC. If you notice banding, try lowering the CRF by one or two points. If the video looks smooth and detailed, you have achieved a good balance. Remember that subjective quality varies by viewer, so find a setting that satisfies your personal standards. For further optimization techniques, you might explore methods for upscaling older content, as discussed in High-Quality 4K SWF to 4K Video Upscaling Without FFmpeg, which shares similar goals of enhancing visual fidelity.

Frequently Asked Questions

What is the difference between CRF and bitrate?

CRF is a quality-based metric that adjusts the bitrate dynamically per frame, while bitrate is a fixed average amount of data allowed per second. CRF generally produces better quality at smaller file sizes because it allocates more data to complex scenes and less to simple ones. Bitrate is useful when you have strict storage constraints, but CRF is preferred for balancing quality and size.

Should I use two-pass encoding with CRF?

No, two-pass encoding is designed for constant bitrate (CBR) or average bitrate (ABR) modes, not for CRF. When using CRF, the encoder automatically determines the optimal bitrate for each segment of the video. Using two-pass with CRF is redundant and will not improve quality or reduce file size further.

How do I fix banding in dark scenes?

Banding in dark scenes is often caused by insufficient bitrate allocation or poor dithering. Lowering the CRF by 1-2 points can help distribute more data to these areas. Additionally, enabling dithering filters in your encoding software can break up the visible bands by adding subtle noise, making the transition smoother to the human eye.

Is HEVC better than H.264 for file size?

Yes, HEVC (H.265) typically offers 50% better compression efficiency than H.264 (AVC) at the same quality level. This means you can halve the file size while maintaining identical visual fidelity. However, HEVC requires more processing power to encode and decode, which may affect performance on older hardware.

Can I use the same CRF for all resolutions?

While you can use the same CRF across resolutions, it is not always optimal. Higher resolutions contain more detail, so they may benefit from a slightly lower CRF to preserve sharpness. Conversely, lower resolutions may tolerate a higher CRF since fine details are lost anyway. Experimenting with slight adjustments based on resolution can yield better results.