Recommended encoding settings for video quality - MediaConvert
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Recommended encoding settings for video quality

When you create a job with AWS Elemental MediaConvert, the encoding settings that you choose affect video quality, file size, and player compatibility.

You can configure your job to allow MediaConvert to automatically select the best encoding settings for video quality, with a balanced output file size. Or, you can manually specify encoding settings to match your output or delivery requirements.

This section introduces basic concepts, describes typical settings, and provides guidance for choosing settings optimized for video quality.

Reference for GOP structure and frame types

When you create a job, the group of pictures (GOP) settings that you choose for your output affect video quality and player compatibility. This section introduces basic GOP concepts, describes typical GOP settings, and provides guidance for choosing settings optimized for video quality.

A GOP is a specific arrangement of compressed video frame types. These frame types include the following:

I-Frames

Intra-coded frames. Contain all of the information that a decoder uses decode the frame. Typically, I-frames use the most number of bits within a video stream.

IDR-Frames

Instantaneous Decoder Refresh frames. Similar to I-frames, they contain all of the information that a decoder uses to decode the frame. However, frames cannot reference any frame that comes before an IDR-frame.

P-Frames

Predicted frames. Contain the differences between the current frame and one or more frames before it. P-frames offer much better compression than I-frames and use fewer bits within a video stream.

B-Frames

Bidirectional predicted frames. Contain the differences between the current frame and one or more frames before or after it. B-frames offer the highest compression and take up the fewest bits within a video stream.

A typical GOP starts with an IDR-frame and follows with a repeating pattern of B- and P-frames. For example: IDRBBPBBPBBPBB

The following topics provide more information about individual GOP settings and recommend settings that are optimized for video quality.

GOP size recommended setting

GOP size is the number of frames in a GOP, and it defines the interval between IDR-frames. For example, if a GOP starts with an IDR-frame and has a combination of 29 B and P-frames, the GOP size is 30 frames.

A typical GOP size is 1–2 seconds long and corresponds to the video frame rate. For example, if your output frame rate is 30 frames per second, a typical GOP size is 30 or 60 frames.

When you set your output video codec to AVC (H.264) or HEVC (H.265), set GOP mode control to Auto. This allows MediaConvert to select an optimal GOP size.

Note

Streaming video formats, including HLS, DASH, CMAF, and MSS, require the fragment or segment length to be a multiple of the GOP size. For more information, see Setting the fragment length for streaming outputs. When you set GOP mode control to Auto for these video formats, MediaConvert automatically selects a compatible and optimized GOP size that's relative to the fragment or segment length.

B-frames between reference frames recommended setting

Defines the maximum number of B-frames that MediaConvert can use between reference frames.

A typical value is 1 or 2 if GOP reference B-Frames is set to Disabled, and 3–5 if GOP reference B-frames is set to Enabled.

When you set your output video codec to AVC (H.264) or HEVC (H.265), keep B-frames between reference frames blank. This allows MediaConvert to select an optimal number of B-frames between reference frames.

Closed GOP cadence recommended setting

Closed GOP cadence defines the number of GOPs a P- or B-frame is able to reference across. A GOP can either be open or closed. Open GOPs can have frames that reference a frame from a different GOP, while closed GOPs have frames that reference only within the GOP itself.

When you set your output video codec to AVC (H.264) or HEVC (H.265), keep Closed GOP cadence blank to allow MediaConvert to select an optimal closed GOP cadence.

Dynamic sub-GOP recommended setting

A dynamic sub-GOP can improve the subjective video quality of high-motion content. It does this by allowing the number of B-frames to vary.

When you set your output video codec to AVC (H.264) or HEVC (H.265), set Dynamic sub-GOP to Adaptive. This allows MediaConvert to determine an optimal sub-GOP.

GOP reference B-frames recommended setting

When you set your output video codec to AVC (H.264) or HEVC (H.265), set GOP reference B-frames to Enabled to allow B-frames to be referenced by other frame types. This improves the video quality of your output relative to its bitrate.

Min I-Interval recommended setting

Min I-Interval enforces a minimum number of frames between IDR-frames. This includes frames that are created at the beginning of a GOP or by scene change detection. Use Min I-Interval to improve video compression by varying GOP size when two IDR-frames would be created near each other.

When you set your output video codec to AVC (H.264) or HEVC (H.265), keep Min I-Interval blank. This allows MediaConvert to select an optimal minimum I-interval.

Adaptive quantization recommended setting

Adaptive quantization selects the strength applied to the different quantization modes that MediaConvert uses, including flicker, spatial, and temporal quantization. MediaConvert uses adaptive quantization to assign bits according to the complexity of your video.

When you set your output video codec to AVC (H.264), HEVC (H.265), or XAVC, set Adaptive quantization to Auto to allow MediaConvert to select an optimal adaptive quantization.