Beginner's guide to HD video

Lights, camera, action!

Over the last few years the line between the once totally separate worlds of video and stills photography has been getting gradually more blurred. This is hardly surprising: in the digital era the camcorder and stills camera share so much basic technology that they are, to all intents and purposes, the same thing. They just use components optimized for still or moving pictures.

The growth in video sharing (on places like YouTube) has ensured movie capture has long been a standard feature even on the cheapest compact camera. And now the increasing mainstream adoption of High Definition flatscreen televisions (and advances in sensor technology) has seen a new wave of HD-capable stills cameras appearing in the last year or so.

The HDSLRs are coming

Compact digital cameras have been able to capture short video clips for many years, usually at a maximum of around 640 x 480 pixels ('VGA') using a frame rate of 15 or 30 frames per second (fps). These movie modes have been more than sufficient for sending as email attachments and web use, but are not of high enough quality to do much else, and they certainly won't look too great on a large HDTV display. The good news is that a new generation of cameras - both compact and SLR - has arrived that's capable of capturing high definition video almost as well as it can capture high resolution stills.

Most modern digital stills cameras offer movie capture, and many now have some form of HD mode. Don't expect quality to match the best camcorders, and don't expect many movie specific controls either. Great for clips, maybe not for serious videography.	Video capable SLRs may be the new kids on the block, but they're already making a splash with both amateur videographers and serious movie makers. The main reason is the big sensor, which allows Hollywood-style focus effects and excellent image quality, even in low light. They may lack lots of video-specific features, but the versatility offered by the huge range of lenses (from fisheyes to extreme telephotos) is opening up a new world of creative possibilities.
Consumer grade camcorders are designed from the ground up as video capture devices, so they offer lots of movie-specific features (electronic viewfinders, power zooms, optimized autofocus, articulated screens, editing controls and so on) - but they have tiny sensors, meaning no depth of field control and noise in low light.	Most dedicated camcorders offer slightly better audio and video quality than compact digital cameras, and once you get to the top end of the camcorder market you're looking at professional tools with interchangeable lenses and broadcast quality 3-CCD capture. Unfortunately you're also looking at professional prices too.

The concept of a single camera that does it all is certainly appealing. Just imagine going out on a shoot and being able to switch between capturing stills and a High Definition movie at the flick of a switch. Well this is now possible. In 2008 Nikon introduced the D90 - the world’s first DSLR camera capable of capturing High Definition video (HD), since when it has been joined by a slew of models from other SLR manufacturers.

Although some professional DSLR cameras are able to shoot stills at up to 10 fps, this rate is nowhere near fast enough for movies and produces jerky video. Smooth action video needs to be shot at 25 or 30 frames per second - and current DSLR cameras can’t flip the mirror up and down that fast. To get round this problem, the latest generation of cameras either keep their mirrors up and capture video in a direct stream from the sensor or, in the case of interchangeable lens system cameras, do away with the mirror entirely (for the purposes of this article we're lumping DSLRs and hybrid cameras, such as Micro Four Thirds models, together). What we'll call HDSLR (High Definition Single Lens Reflex) cameras use the sensor's Live View functionality to capture video.

Many high end compacts now offer some form of HD movie capture (usually 720p), and many of the major manufacturers have introduced at least one DSLR (or other interchangeable lens system camera) with HD movie capabilities. Video in a DLSR and large-sensor cameras brings a wealth of creative possibilities including the ability to control depth of field and exposure, and to use exotic lenses for effects traditionally the preserve of the professional videographer.

Below is a table showing a selection of HDSLRs as of November 2009:

Camcorder replacement?

Although there's no denying the appeal of a combined stills / HD movie camcorder (especially one that offers all the photographic possibilities of an interchangeable lens SLR), it's not all plain sailing, and there are some important disadvantages to balance the obvious advantages over a dedicated camcorder - particularly once you start to consider the high end semi-professional and professional models, which offer better movie quality and more advanced video and audio controls. Without getting too bogged down in detail these can be summed up as follows:

Video capable DSLR vs Camcorder: Key points

Let's take a look in detail at a few of the more important considerations when looking at a video capable digital stills camera as an alternative to a camcorder.

Clip limits

One of the main disadvantages with using a stills camera to shoot movies is the short recording times available for HD video; Nikons limit a single take to 5 minutes while Canons and European Panasonics stop after 29 minutes, 59 seconds. This limitation is due to the different (European) import duty rates for still and video cameras. However, although this may seem like a handicap, in reality you would never need to shoot a sequence for longer than a couple of minutes or so (The celebrated opening take of Orson Welles' 'Touch of Evil' calls it a day after just three and a half). Look at any program or documentary on TV and notice that most shots are only held for a few seconds. Furthermore, a 4Gb card will store just 12 minutes of 1080p video from a Canon 5D Mark II, so you may never hit the 29 minute limit. The only time you would possibly need a longer recording time is in the case of shooting an entire wedding ceremony or event, in these situations a camcorder may be a better option.

Storage

Stills cameras use high speed memory cards to capture video, currently the popular choices are SDHC (Secure Digital High Capacity) and CompactFlash (which tends to be reserved for professional cameras these days). High speed cards are essential as video requires cards that are capable of keeping up with fast data transfer, this can be any rate between 15 to 38 M/bs (mega bits per second). Higher data rates produce better quality video, most TV broadcast channels have a minimum requirement for a capture quality of 50 to 100M/bs. SDHC cards are available in various capacity sizes from 4GB through to 32GB, larger capacities will become available. The cards also have class speed ratings; 2, 4 & 6, for video capture - Class 6 cards provide the most reliable option.

Shooting style

In order to get the best possible results with video, stills photographers will have to change their shooting techniques. One of the problems with using a HDSLR is that the camera itself is not ideally suited for capturing video. Trying to view an LCD screen on a bright sunny day can be very difficult, although some cameras are now offering flip out screens which can assist with awkward angles. Using an LCD screen as a monitor means you have to hold the camera away from your body to view the screen, this is not the ideal way to hold any camera and can introduce camera shake. A good tripod will cure the problem, but it also takes away some of the spontaneity that is enjoyed when shooting using a HDSLR.

Panning too fast across a static scene can make the image look blurry, this is due to the way images are compressed using MPEG – a key frame is recorded and then the next sequence of 11 frames use this as a reference, this is known as LGOP (Long Group of Pictures). A slow pan using a tripod will produce the best results, generally try to keep the pan duration to about seven seconds, this gives the viewer time to take in the scene.

The CMOS sensor, as used in most DSLRs, builds up the picture from top to bottom (a 'rolling shutter'). This can cause distortion when panning as the bottom section of the sensor will receive light after the top part, so if you are panning too fast then you may notice skew in upright lines (the so-called 'jello effect').

The rolling shutter causes a skewing of verticals if you pan too quickly (Nikon D90 - click thumbnail to see the video).

Focus and exposure

An out-of-focus HD sequence is really not worth looking at, especially when viewed on a 50inch plasma screen. The best way of ensuring sharp video is to set the focus on your subject before switching to Live View. Although a few SLRs allow you to autofocus in live view mode (and when shooting movies) it's always slow and you're better off leaving it in manual focus mode (this may be the only option).

Auto exposure continues to operate whilst recording, at first glance this may be seen as a good thing, but any changes in exposure during a clip will be very noticeable. The best method is to lock off the exposure before starting to record the video. Professional videographers rarely use auto focus or auto exposure, it’s easier to correct any exposure error on an entire clip than trying to correct small segments within a sequence.

Depth of field

One of the main advantages with using a HDSLR camera is in its ability to capture movies with a limited depth of field, especially when using full frame HDSLR cameras like the Canon 5D Mark II or Nikon D3S. The larger sensor size will give you amazing shallow depth of field, but this also means focus on the main elements in your clip has to be critical. The smaller sensor sizes, used by compact cameras and camcorders are more forgiving with focus due the extended depth of field, which can be from the front lens element to infinity – watch out for dust spots on the front of the lens.

Large sensor and (relatively) fast lenses allow digital SLRs to capture video with very limited depth of field - perfect for isolating a subject from its background or for creative effect. The results can look very professional but remember that focus is critical.
• Panasonic Lumix G Vario HD 14-140mm • 1280 x 720 pixels (M-JPEG, AVI) • 8:70 seconds, 32.8 MB	• Panasonic Lumix G Vario HD 14-140mm • 1280 x 720 pixels (M-JPEG, AVI) • 7:43 seconds, 31.7 MB

Audio

Most manufacturers seem to have stopped short on quality audio for the first generation of HDSLR cameras. Video usually uses audio sampling at 48 KHz, whereas Canon and Olympus have opted for 44.1kHz, Nikon 11kHz. Panasonic, however, has stretched all the way to 48 kHz for many of its cameras. Audio seems to be regarded as an added feature with very little importance, offering no proper audio controls or monitoring facilities. This is a little short-sighted as audio is an important element in any production. However, several models do at least have a stereo microphone (though many use a small mono microphone on the camera body itself), which annoyingly also picks up any handling noises. However, a few cameras are now including a stereo socket so you can use a microphone which is not attached to the camera.

Video resolutions

At one time there were just two (analog) video formats - VHS and Betamax - and no one gave a though to pixels and resolution figures, but those days have gone forever. Now we have a plethora of format choices led by High Definition (HD), and they all have their uses.

• Full HD. (aka 1080p) This is currently the largest frame size that most current HDSLR cameras offer. The frame size is 1920 x 1080 pixels and produces stunning quality when viewed on large plasma and LCD screens. The downside is that this size can make a lot of processing demands on your computer.
• HD. (aka 720p) Still high definition but at a smaller frame size of 1280 x 720. Nikon digital cameras use this format and although smaller than Full-HD it still offers outstanding picture quality. If you intend to produce DVDs or want to upload to the web then the 1280 x 720 HD format is a good choice.
• HDV. High Definition Video, this format is mainly used on tape based camcorders. The format is 1440 x 1080 anamorphic. Once downloaded the frame expands to Full-HD (1920 x 1080), but because the image is squeezed on the horizontal it can lose some definition.
• SD. Standard Definition. This format conforms to the 720 x 480 (NTSC – North America) or 720 x 576 (PAL - Europe) formats. Almost all commercial DVDs are produced in Standard Definition. The image format can be either 4:3 or 16:9, the larger size is an anamorphic expansion of the 4:3 size.
• VGA. Video Graphics Array, a format of up to 800 x 600 pixels (technically SVGA), but generally 640 x 480 pixels. VGA has its uses for web video, sending as email attachments, PowerPoint presentations etc., but for high quality video you should be looking at HD. Generally VGA allows longer shooting times, but at lower quality. Most digital cameras offer VGA as an option.

Although all televisions will scale up lower resolution video sources to fill the screen, you'll always get better results with movies that match the screen's native resolution. This is even more obvious when viewing movies on a computer screen. A VGA movie will look tiny on a 20" LCD PC screen, and will look terrible if you stretch it to fill the available area. This is partly because PC screens generally have a higher native resolution than a television, but mainly because you're usually viewing them from a lot closer.

The graphic above shows the relative size of each of the most common video recording resolutions,
from full HD (1920 x 1080, '1080p') down to 640 x 480 ('VGA')

Motion JPEG, MPEG or AVCHD?

The majority of digital cameras still use motion JPEG to capture video, as they have for many years. As the name implies, M-JPEG is unusual in video terms as it compresses each frame individually as a JPEG image. Compared to more advanced video file systems (which use complex inter-frame compression) M-JPEG produces relatively large files, but the picture quality is usually very good, the files can be viewed on almost any modern computer without the need for additional software and they can be uploaded directly to sites like YouTube or emailed as an attachment (so are great for sharing). M-JPEG movies are normally stored in either a QuickTime (.MOV) or an AVI file wrapper.

The main reason manufacturers choose M-JPEG is cost: it doesn't require any additional hardware in-camera or licence fees to be paid. As video becomes more important many are now moving to a more sophisticated compression system, namely MPEG-4 (usually using H.264 codec). This allows higher bit rates with smaller files, and is increasingly well supported by software and operating systems. Panasonic's Micro Four Thirds system uses the same AVCHD system as many digital camcorders. AVCHD is still based around MPEG-4/H.264 compression, but uses a complicated directory system that is derived from the Blu-ray Disc specification, though is not identical to it.

The biggest advantage of AVCHD is also its biggest problem: compatibility. Whilst it offers uniquely friendly interaction with consumer electronics devices (such as DVD recorders and Panasonic TVs), the files created are difficult for computers to deal with - movies cannot simply be dragged off the camera's card and double-clicked, and they put a strain on the processing capabilities of all but the most powerful desktop PCs. With the right editing software AVCHD is ideally suited to editing and creating home DVDs, for example, but for sharing simple clips PC to PC you're still better off with one of the more common formats.

Currently, all HDSLR capture video using Progressive mode. Progressive means that every frame is a complete picture (denoted by the 'p' at the end of the video size names). The alternative is Interlaced video which is actually twice the number of frames (60i or 50i) but each frame only contains half the image information, otherwise known as fields, Upper and Lower. The two frames are combined on alternate lines to make up a complete frame. This produces smoother action video but can also leave a tearing effect. Interlaced video is more commonly used on camcorders.

A still taken from a movie shot using interlaced mode, showing how the Upper and Lower fields are out of sync.	Interlaced stills can be fixed by using the De-Interlace filter in Photoshop, this can leave the image looking soft.
A frame grab from a 1280x720 sequence shot in Progressive mode. Click on the image to see the full size image.

File Sizes

When shooting stills we can either shoot RAW files which are generally large file sizes and occupy a large amount of memory or we can capture more pictures by using JPEG compression. The trade off with JPEGs is lower quality, but in reality most users are not aware of any significant drop in image quality. For video there isn’t a true RAW equivalent, instead there are several different CODECS for compression. HD movies can occupy several gigabytes for a short sequence, so in order to fit it all on a memory card the movie has to use some form of compression. MPEG-4 is a popular choice.

A four minute sequence captured in Full-HD (1920 x 1080) using MPEG-4 compression occupies 1048 megabytes of memory (just over 1 gigabyte), the same four minute sequence without compression occupies 32,457 gigabytes of memory. Many stills cameras offer Motion JPEG compressed AVI or MOV capture - these are also compressed file formats but use less sophisticated (and less effective) compression.

For example, the Canon 5D Mark II records to CompactFlash cards using QuickTime MOV files with the H264 codec, the Panasonic GH1 uses SDHC cards and uses 17 M/bs AVCHD format, which is an efficient codec and in many cases will outperform 25 M/bs HDV recordings. Nikon uses AVI (Motion JPEG) to record video.

Editing Video Files

There are many consumer applications for editing video which include iMovie (Mac), Adobe Premiere Elements 8, and Corel Video Studio X2. The popular applications offer many easy to follow wizards which should get you productive in no time. For more professional editing you should check out Final Cut Pro (Mac), Adobe Premiere Pro CS4, Avid, Sony Vegas, Adobe After Effects CS4 etc. For these high end applications there is a steep learning curve, but the rewards are tremendous. You can apply more or less the same color correction tools, such as color balance, Hue Saturation, curves etc., to a whole clip, as you would with a still image. Add in a music track and then export the entire production to a DVD or the internet.

Like stills images, JPEG, TIFF, PSD, GIF etc., video can also be delivered in several different formats, these are called “Wrappers”. A movie file is wrapped in a format that can be read by a player, this can be a MOV file for QuickTime, WMF for Windows Media player, FLV for Flash players etc. The output file can be decided after you have finished editing the file and are ready to export it. You will also be offered an endless list of output codecs, these will define how the movie is compressed etc. For the user who doesn’t want to become involved with video editing, then you can simply link up the camera to a domestic TV set via the HDMI output and watch your movie clips from an armchair or upload your video to YouTube or other sites.

TV standards

In North America broadcasters (and home appliances) use the NTSC standard (National Television System Committee) which plays at 29.97 frames per second (this may be indicated as 30fps). Most of Europe and the UK use the PAL TV standard (Phase Alternating Line), which plays at 25 frames per second. You can play NTSC DVDs on most PAL TV sets, but you can’t play PAL DVDs on NTSC TVs. The other option is to use 24fps which matches cine-film capture rates (actually it’s 23.976fps); almost all TV's can display this rate.

© 2009, www.dpreview.com |
Text by Vincent Oliver (www.photo-i.co.uk), additional text and editing by Simon Joinson