Compression, Codecs, and Wrappers
When editing video on any non-linear editor, you are looking at, and working with, digital files, which for most people have been created with a digital camera. But how do these original camera generated files differ from one and other? What is the difference between a file created on an Alexa camera and a file made on a Canon DSLR? Placing the size of the sensor, image processing and lens quality to one side, the answer mainly lies in the cameras compression technology and in turn the type of ‘wrapper’ it places the video in to.
DSLRs are big users of H.264 compression for HD and UHD images.
Compression is simply a way of reducing something in size. In the case of digital video cameras, in very broad terms, this affects the video file size that is created by the camera.
Uncompressed HD Video obviously has no compression on it and when shot at 25 frames per second would require 695 GB - Gigabytes - of storage for each hour of material! It would also need to write to the cameras storage medium at 198 MB (Mega Bytes) per second. UHD, 4K and 8K obviously needing even more!
This uncompressed data rate and storage demand, is far too high for most cameras, so some form of compression is applied, within the camera, to help reduce the size of the storage and in turn, reduce the data-rate needed to write to it.
The camera acts as the encoder, taking the image that falls on the cameras image sensor, or ‘pick up’ device, applying a mathematical formula to it, to compress the image to a given standard, and then write it down to the storage medium on the camera.
Editing systems use the same ‘codec’ formula ,to both Encode and Decode material when editing with that particular media. But this data just can’t be written down as a string of ones and zeros however we feel, there needs to be a framework for the ‘compressed’ information to be written into, so that other systems can find the information to ‘decode’ for playback. This is where a wrapper is used to contain the data stream.
The wrapper defines the file format and is usually reflected in the file extension. Two of the most common professional formats used are QuickTime (.mov) and MXF (.mxf).
In the context of video compression, a "codec" (which is the joining of the words "encoder" and "decoder"), is the type of compression that is used to reduce the size of the image. A mathematical map if you like, that allows the end user to decode the compressed image for use. If licensed it can also be used by an editing system to create new content with the same codec.
Some codecs are better than others. Some are great at making images very small, while maintaining the image quality, for example, the Panasonic AVC Intra and Ultra codecs. The mathematical effort that goes into the compression of an image varies from codec to codec. Codecs that offer a high image quality, while keeping a very low data-rate, 50Mbs and below, for example, require even more intensive and complex compression meaning that these type of codecs, generally, are 'harder' to work with for the editing suite.
For example, most Avid codecs such as 2:1 and DNxHD codecs are relatively 'easy' codecs to process as they are processor light, but disk space heavy. Packing and unpacking the images doesn’t take much effort on behalf of the NLE. H.264 compressed images on the other hand, a very popular codec that uses the MPEG4 standard, makes the image much smaller (disk space size wise) with minimum picture degrade but is more processor intensive. In short this means it takes more processing cycles to compress and decompress a H.264 image.
In Media Composer, the ‘codec’ used is referred to as a “resolution.” Avid Standard Definition (SD) codecs are listed as ratios, indicating the amount of compression used, such as two to one (2:1) or fifteen to one (15:1). The higher the ratio, the more compressed the image. Avid high-definition (HD) codecs are named DNxHD, with the number in the name representing the data rate. For example, DNxHD 120 has a data rate of 120 Mbps. The higher the number, the more data is being transferred with each frame, so the more disk capacity is needed. Avid's latest codec, DNxHR are named and no not refer to compression or data rates but rather image quality - LB Low Bandwidth or SQ Standard Quality as well as a number of others. For more details, see the Avid DNxHR formats and compression details.
Codecs developed by other companies are listed by name, such as XDCAM 50 or ProRes 422. The meaning of the number attached to each varies by manufacturer.