Lossless and Lossy Editing Workflows

From TreetWiki

In television and cinema, captured and edited footage is sacred and may go through many different processes before it reaches the intended audience. For example, captures may be assembled into scenes which are then later recombined by another editor. Or, portions of films or movies may be produced by outside contractors. Those parts are then combined to form the final edit. Even final edits can then be reused. For example, a motion picture edit may be recombined into a television edit, trailers may be created, or "Making Of" DVD extras may be created.

Since footage moves through the studio so often, it is essential that edited material never degrades in quality no matter how many editing sessions and re-editing sessions there may be. To understand how to achieve this, it's useful to explain both Lossy and Lossless editing workflow.

Note that these workflows are not specific to any one editing package. It is possible to achieve lossless workflows on PC's using Adobe Premier, and it is possible (by throwing caution to the wind) to achieve lossy and degraded editing workflows on expensive Mac-based Avid systems. The principles are the same for all packages and you'll need to learn how your package works to best understand what kind of workflow is appropriate.

Lossy Workflows

Lossy workflows result when input material is re-encoded by an editing package during editing, and possibly re-encoded again when the final export of video. Most prosumer editing packages such as iMovie or Windows Movie Maker do this by default to make life simple and easy. In addition, most consumers who do editing are not concerned with absolute quality standards and will never be submitting videos for broadcast or inclusion in HD studio productions. Most likely, consumer video will be presente on the web using sites like YouTube where lossy re-encoding is normal and quality standards are low.

Lossy workflows result because the editing package has no choice except to convert video from the timeline into some intermediate editing format, then to convert again when the final project is saved. While it may appear that frames are being transferred unchanged, and even close inspection may appear to reveal no differences, there are subtle differences in edges and "color fields" which gradually will degrade video edited using lossy workflows.

The reason why conversion must be performed is not always obvious. Here are some of the important reasons:

• Codecs which are convenient for internet transfer are not designed for editing purposes. Editors require good Intermediate Codecs in order to assure rapid editing. So, the format of codecs such as H.264 and Mpeg2 do not allow exact frame extraction and frames must be constructed and converted because of the very design of these codecs.
• Non-intermediate codecs use IBP groups to store picture data. So, a second of video may contain only one actual frame of video (called an I-Frame) followed by algorithmic frames (P and B) which instruct the codec how to assemble the full frame. When clips are dropped onto timelines, these frames do not align, and thus the editing package must decompress the IBP groups so that all the frames are stored as a full-frame internal graphic. Intermediate codecs are designed so that they do not have this limitation and each frame can be transferred in its exact form.

The figure below shows a lossy workflow. Note the following:

• Input clips have IBP groups which do not align with the I-only frames of the intermediate codec and don't align with the IBP frames of the final exported video.
• In the example, two conversions are done. One conversion to decompress the video on the timeline, and another conversion when the intermediate edit is exported to final video. This is typical.

Lossless Workflow

By contrast, lossless workflow preserves every frame of video exactly. In a good studio workflow, the video frames captured by cameras move through the system unmodified and are pixel-for-pixel identical through even hundreds of generations of editing and re-use. By contrast, a lossy workflow can only withstand two or three editing re-use sessions at best before visible artifacts and color shifts begin to appear in results.

In a lossless workflow, studios observe the following rules:

• A single format is chosen for production work and it is used throughout the editing, capture, and storage process. Codecs used for this purpose are almost always specially designed Intermediate Codecs.
• When video is provided by another source, it is always re-encoded immediately into the chosen format used by the studio. Therefore, the video will mix and match easily with existing footage, and archives and re-edits work normally.

The figure below shows a lossless workflow. Note the following:

• Because there are no IBP groups present in clips on the timeline, each frame is independent and can be moved into the intermediate edit unchanged.
• Professional editing systems are designed to recognise that clip format and intermediate format are identical. In such cases, they do not even store intermediate frames at all, but display information directly from the timeline. This results in almost instantaneous editing, scrubbing, and rendering.
• Because the final export format is the same as the clip and intermediate format, the final frames match pixel-for-pixel with the frames present on the original timelines.
• Avid and Final Cut systems have the ability to output what are called "reference movies" which actually refer to the exact clips on the timeline rather than copying the frames at all. Thus, an long and complex edit involving terabytes of content may result in finished files which are only a fraction of the original size.

Does it matter?

It should be obvious from the above that lossless workflows are superior. However, they require a great deal of discipline and require that archives and source imports be carefully converted and managed. While they produce better quality, they may not be worth the effort, especially if your editing package does not support the features necessary to control the internal workflow parameters it uses.

You can get away with lossy workflow if:

• You are only using your creations on the web and do not care about DVD or BluRay results.
• You will not be giving your footage to professional broadcasters (such as CNN, ABC, or HBO) for inclusion in their program material.
• You will not be using the output of one editing session in other sessions (such as when inserts or credits are created then used again).

But, if you are concerned about any of the following, it may be worth investigating lossless workflows:

• Broadcasters are picky about chroma and encoding quality. Many broadcasters will reject video which is not created with suitable workflow.
• If you are building an archive you feel may be reused by others.
• If you are storing video and may later re-combined or re-use it to create new programming.