Video compression workflow for shows
In small scale and informal production contexts where specific expertise cannot be assumed or required then the rules and workflow around compression and online publication need to be simple, understandable, and robust. This is the first iteration for some rules and workflow.
Compressing recorded video is necessary where the source material is at a higher resolution than is need for the publication/distribution medium. For example if I recorded at 1080p resolution (1920 x 1080 pixels) then each frame has around 2.1 megapixels and uncompressed is approximately 375MB per second, this means you would not quite contain 2 seconds of video on a single CD and one hour of content would need over 2000 CDs. In its original format this video would also take many days to be able to be uploaded online, and in turn take many days to be downloaded for anyone to be able to watch! When compressed using the recommendations here the same one hour of video will comfortably fit on a single CD and only specialists will notice any degradation to the image quality.
Compression is done using software that takes the original source video and makes it smaller. This is done by reducing the size of the image in pixels (for example my laptop has a maximum resolution of 1440 x 900 pixels so video at 1080p is literally too big for my screen) and by reducing the amount of data the video needs for each second. It can do this because it turns out an enormous amount of the information recorded when you film video or take a digital photograph can be removed without affecting the visible quality – we can’t see any difference.
Compression involves compromises between how large the files are going to be, image quality, speed and cost. The more compressed something is, the faster it will load and play for viewers, and the lower your bandwidth costs as there is less data to deliver, but there is a risk of poor image quality as it may contain what are known as ‘compression artefacts’ – pixellation, a lack of smooth gradation in colours, and visual stuttering. On the other hand the larger the file then generally the higher the image quality, but it will take longer to upload to your site, and for it to download to viewers, and it will cost more to download as it will require more bandwidth.
There is a ‘sweet spot’ that trades off each of these, where adding more data to the video (making the file size larger) leads to minor if any gains in visual quality and the art of compressing is to find this ‘sweet spot’. In the context of DIY production some experimentation is needed to find the quality that meets the technical requirements of your delivery platform and the aesthetic needs of your creators and the audience.
(It is very important to understand that there is usually a dramatic difference between what creators think is the minimum acceptable quality versus what the audience regards as a minimum acceptable quality.)
This is tech talk for a moment. Codec is short for ‘compression decompression’ and it is what all video file formats on computers use to compress and play back video and sound. A codec is an algorithm that is only about video and audio compression and playback. A codec is not the same thing as a file format. A file format is how the data is structured in a file for the computer, a codec is simply how the video has been compressed. Therefore you can have a QuickTime (Apple) video file that ends in .mov, .mp3, .mp4, .m4v, yet each is compressed with the same codec, and you can have a Windows Media File that ends in .asf which could contain video compressed using a variety of different formats.
Unfortunately this is complicated, and like many other things online could change dramatically in the future. However, at the moment things are quite straightforward because there is one specific codec that has been developed that is rapidly becoming a de facto standard for video online and so will work in most web browsers and most mobile devices, most of the time. This codec is known as H.264. It is part of what is known as MPEG4, is an international open standard for compression, and produces very high quality video even when compressed.
In online video the rule of thumb is that storage is cheap and bandwidth expensive. Bandwidth is a combination of how fast your internet connection is and how much data is delivered through that connection. When hosting video on a web server you can reasonably assume that it has a very fast connection to the rest of the world, so potentially it can deliver a lot of data (videos). This means that if you are paying for hosting and you have video that is very popular, then it will use an enormous amount of data, and this can be potentially expensive.
What counts as ‘good enough’ with the finished compressed video is subjective. If in doubt test with your main audience. (For example film scholars wanting to view film clips will generally appreciate access to high quality video, but they may not need such quality when browsing. Similarly a dance company may be happy that you can see enough detail to enjoy the performance and choreography and it doesn’t matter that you can’t identify what colour individual buttons are on a costume.)
1.1 If the video has been filmed on any device then it should be compressed using the following protocol. This includes domestic personal video recorders and smart phones as these generally now record video at some variety of HD which is higher quality than what is required for web access and use.
1.2 If the video is to be downloaded from a video hosting site then a copy that is compressed using H.264 (MPEG4) is required. Most sites provide this, and if in doubt selecting an option for iPhone or iPad will suffice. If it is not in this format then it should be compressed using this protocol.
Getting file info in QuickTime
1.3 It is possible to record video in a format that is already adequately compressed for use online. If unsure about the size (in pixels) and data rate of the source video then:
- open the file in QuickTime
- under the Window menu select Show Movie Inspector (keyboard shortcut = Command I)
- the window that appears shows you the dimensions of the movie (in pixels), its format (the codecs it uses), and its data rate
- if the dimensions are greater than 640 pixels wide (never mind the height) then compress
- if the data rate is greater than 1.1 Mbits/s (never mind the Data Size) then compress
- if the format is not H.264, then compress
The file information window in QuickTime, Format and Data Rate are what matters.
2.1 The recommended codec is H.264. It is an open standard, provides high quality compression, readily available, and a de facto standard for online video. It is highly recommended to use an ‘official’ (licensed) version of the codec to ensure compliance with H.264 standards. (For mobile use it is essential as mobile devices have hardware standards that define what can and cannot play.) H.264 is also commercial software and the risk of future IP issues is potentially lessened by using licensed software. QuickTime, QuickTime Pro, and Compressor all use legitimate versions of H.264. As the web moves towards what is known as ‘native’ support for video (which simply means no plugin needs to be installed to view video online) H.264 will play with no required plugins in most recent web browsers. For those web browsers that require a plug in then these are available.
3. Compression Software
3.1 The recommended software currently is to use Apple’s Compressor. It is affordable (AU$52.00), and allows for predefined settings that include sophisticated compression settings, watermarking, batch processing and notification of completion. It is possible to have predefined settings that can be easily shared between computers, and also provides ‘droplets’ that are in effect mini-programs that you can drag a video file upon and it will compress it automatically using predefined settings. This makes the process very simple for end users and removes any decision making around data rate, image size and so on.
Using Compressor a single predefined setting is applied and it is smart enough to preserve the aspect ratio of the source media. In addition the ‘droplet’ that can be created and distributed with a copy of Compressor will allow others to compress video to this format, for example while on location. The droplet prevents any of the parameters being changed by users, accidentally or deliberately, ensuring the process is simple and that technical problems and questions are minimised.
4.1 Compressor currently requires OS X 10.6.8 or better and a recent Macintosh. It will run on a laptop computer. Compression is computationally intense as it involves a lot of data and a lot of calculations. It is very difficult to estimate how long compression might take as it depends on the size of the source media (its data rate and total file size), its length, and the hardware being used. An iMac will be faster than a laptop, and a MacPro considerably faster again. Ideally the only software running while compressing a large amount of video (one or more hours of source media) is Compressor and for video that runs for over an hour I would allow up to four hours on a laptop dedicated to only running Compressor.
5. Joining of Sequences
5.1 If the source media has been split into multiple files by hardware or software the most effective way to deal with such files is to:
- compress them using this protocol individually (things work faster when dealing with smaller files)
- use QuickTime 10.x to do a simple ‘butt edit’ by
- open the first clip
- locate the next clip in the finder
- drag the second clip into the first (which is open in QuickTime Player)
- it will drop in where you place it
- click ‘done’ when finished and you will be prompted to save it
- name it, and when saved it will be ready for upload
Description: 640 px frame controls on 1000kbs max resizing
File Extension: mov
Email notification to:
source frames play at 24.000 fps
AAC, Mono, 24.000 kHz
Width and Height: Up to 640 x 360
Pixel aspect ratio: Default
Frame rate: 24
Frame Controls On:
Retiming: (Fast) Nearest Frame
Resize Filter: Linear Filter
Deinterlace Filter: Better (Motion Adaptive)
Adaptive Details: Off
Detail Level: 0
Field Output: Progressive
Codec Type: H.264
Multi-pass: On, frame reorder: On
Pixel depth: 24
Spatial quality: 75
Min. Spatial quality: 25
Temporal quality: 50
Min. temporal quality: 25
Average data rate: 1.024 (Mbps)
Fast Start: on
requires QuickTime 3 Minimum
Position: Lower Right–Title Safe
Scale By: 1.000
File Name: Beatrice:Users:amiles:Movies:00 oz tests:watermark.png
(These settings cannot simply be replicated outside of Compressor as a) some of the terminology is specific to Compressor, b) other software lacks this particular settings ability to scale to a maximum dimension using source aspect ratio.)