Choosing a Camera for Affordable Quality Videography (Updated 2011-Nov-05)

I do not usually blog about photography. However, I have just completed a 1.5 year long election campaign, and I now have time again to return to video production. That effort will begin by replacing my 5 year old Panasonic standard-definition video camera with a new High Definition (HD) camera. I have done a considerable amount of research, and it has not been easy – there are numerous things to consider, and I have not found a single place in which they have all been addressed. So, to save you time – and so I don’t forget what I have learned – I am writing this blog entry. May you find it as useful as I hope to find it while choosing a camera.

The Issues: Low Light Conditions & A Desire for Film-like Video

Most consumer-level video cameras have trouble with low amounts of light. I do not have formal training in lighting for a video shoot, and I have often shot video in low light conditions. It is possible to boost the brightness after the fact in a video editor, but the result is a grainy image (here is how bad it has gotten). Moreover, in low light, the auto-focus features of consumer video cameras often have a lot of trouble focusing properly. So I have a long-standing desire for a camera that handles low-light conditions better.

This was shot by candle light on an old consumer-grade
standard definition mini DV camera with a lousy sensor.

Accordingly, a few months back, I started researching consumer-grade video cameras, looking in particular for a camera that could better handle low light situations. The answer was clear: Panasonic’s high definition TM700. It had 3 MOS sensors in it which allowed that camera to provide quality images even in scenes with little light. The TM700 has been replaced by the TM900 (a very similar camera of equal or better quality). It sells for about $900 at present. It was priced at $1,100 a few months ago, when I decided I’d wait for the price to come down.

A few weeks ago, I was interviewed on camera by a documentarian. Knowing I was interested in doing more documentaries (my first effort was The Principle of Pot), he recommended I buy a camera like his: a Sony NEX VG-10. He said it had “a big sensor” and a removable/changeable lens, like professional video cameras, but that it sells for about $2,000.00 instead of many times that. I thought it might still be more than I was willing to pay, but I thought I’d look into it after the October 6, 2011 election. Last week, I did so.

I discovered that the Sony NEX VG-10 is still available, but it’s on sale (currently, for about $1,500) because it has been replaced with the VG-20 (which is selling for about $2,200). The salesman at my local Henry’s camera story told me that the VG-20 indeed has a big sensor – it is the same sensor used in a Sony digital SLR camera, which is used for taking still photographs. The big sensor, he explained, makes the VG-10 and VG-20 even better in low light than the Panasonic TM900.

This left me with the question: what are the differences between the Panasonic TM900 that make it less expensive than the Sony VG-10 or VG-20, and should I pay the extra to get a Sony? That question led to a host of discoveries, which I’ll now tell you about (trust me, it’s worth knowing).

1. Big Sensor: Not for Only for Low Light and High Definition, but Especially for Film-like Depth of Field

I had been led to believe that the only reason one wants a big sensor is to have a camera that handles low light conditions well. Not so. The main reason big sensors are desired is because they allow you to do what cannot be done with the relatively small sensors found in most consumer-level video cameras: big sensors allow you to have shallow depth of field.

In case you’re wondering “what is depth of field?” let me tell you in over-simplified layman terms. You will sometimes see a photo (or a scene in a film) in which something (e.g., a flower) is in-focus, but things in front of it and behind it are out-of-focus. The area (i.e., the range of distances from the camera) that is in focus is the “field”. The “depth” of that field is the range of distances from the camera that are in-focus. Now, let’s take two examples.

Example 1: If you set your camera so that you have an eight inch (i.e., 8″) depth of field, and you focus on a flower ten feet away from you, then things 4″ in front of the flower, and 4″ behind the flower will be more-or-less in-focus. Things closer than 9′ 8″ from the camera, and things further than 10′ 4″ from the camera, will be more-or-less out-of-focus.

Example 2: If you set your camera so that you have a four foot (4′) depth of field, and you focus on the same flower ten feet away, then things closer than 8′ from the camera will be out of focus, and things further than 12′ from the camera will be out of focus.

Philip Bloom, showing us the shallow depths of field made possible with a big sensor.

The point of depth of field: it allows the videographer to direct the audience’s attention to the thing that is in focus, by blurring out everything else. You will see this done in films, and in still photography all of the time. However – and this is key – you will almost never see shallow depth of field in consumer-level video recorders. The reason: small sensors do not facilitate depth of field, and most consumer-level video recorders use small sensors. That means that, with most consumer-level video cameras, everything is in focus, whether it is close or far away. This is one of the reasons that the videos made with most consumer-level camcorders don’t look like films or still photographs.

You can buy video cameras with bigger sensors that allow you to create shallow depth of field effects, but most of them cost over $5,000 and range into the tens of thousands. So, what is most innovative about Sony’s VG-10 and VG-20 is not that they handle low light conditions well, but that they give you the ability to have film-like, professional-looking depth-of-field effects for as little as $1,500 (i.e., the current price of the VG-10). So, the answer seems clear, right? I should fork out the extra dough for a Sony VG-10 or VG-20 and get professional-looking video, instead of buying the Panasonic TM900 (which has a small sensor and cannot give you shallow depth of field), right? Wrong. It turns out that the whole ‘big sensor’ thing has an interesting history and several draw-backs.

2. History

It turns out that the idea of giving consumers a big sensor to work with has its origin in news photography (stills, not video). As news media are faltering, they are requiring their photographers also to gather video for online and TV uses. So, perhaps among others, Canon a few years ago introduced their Canon 5D: a digital SLR (“DSLR”) camera for high-quality still photos (it has a full-sized sensor, about the size of a frame of 35 mm camera film), but it allows about 12 minutes of video to be taken with the camera.

Why 12 minutes? Not for any technical reason. It turns out that when event organizers sell exclusive rights to various events (e.g., a rock concert or hockey game), separate contracts are sold for still photography and for video photography. For whatever reason, a camera designed for shooting stills can be used to take up to 12 minutes of video by a person having only the contract for still photography. Go above that, and you’re expected to use a camera designed not for stills, but for videos.

Having a 12 minute limit isn’t necessarily the end of the world. All you have to do is hit record again to resume video recording. But, if you are videoing something that continues uninterrupted for more than 12 minutes (think something along the lines of a 60 minute speech, or a concert in which Rush is performing side 1 of their album “Hemispheres”), even a momentary interruption every 12 minutes could ruin your video project.

Now, understand this: even the high quality professional video cameras do not use the big sensors used in high quality cameras used primarily for stills. So, some videographers have purchased cameras designed for stills, having 12 minutes of video capability, because they produce very high-quality images, and offer stunningly shallow depth of field (for those video artists wanting very tiny depths to be in-focus, and everything else to be out-of-focus), as compared even to high-quality professional video cameras.

Seeing that the 5D was proving to be a popular camera, and that the idea of a still/video hybrid camera in the $3000 and under range was proving to have a market, Canon more recently introduced the Canon 7D. The 7D has a slightly smaller sensor than the 5D, but which provides awesome, film-quality video, 12 minutes at a time, and it has several features that make it much better-suited to video production…plus, it costs about $2,000 (about $,1000 less than the 5D). The 7D proved so popular, that Canon thereafter released the Rebel T2i. For shooting video, the T2i is pretty much the same camera as the 7D, but it sells for about $800 (yep, less than the Panasonic TM 900). But don’t get all excited just yet. If you also want the camera to shoot stunning stills, the 7D is a significantly better camera (and it is much more durable, in terms of its build).

The Rebel T2i has been slightly more refined by its recent successor, the T3i. So, should a person who doesn’t care about stills just buy a Rebel T3i instead of the Panasonic TM900?

3. Moire: a Visual Distortion Problem with the Big Sensor Still/Video Hybrid DSLRs

It turns out that there are some drawbacks to using a still photography big sensor as a video sensor. First of all, these hybrids tend all to have problems with various forms of distortions that are better handled by cameras designed primarily for video. The one I’ve seen the most complaints about is: Moire.

Watch the roof-top for moire. This is shot with a Sony NEX VG-10

If you’ve ever looked at a tight-nit, or checkered pattern while looking through the screen on a screen door, you’ll see a neat visual effect called Moire. Moire is like big curved or squiggly bands of light. Those bands look ugly, amateur, and un-film-like. Cameras designed first and foremost for video have been designed, in many cases, to minimize moire. The Panasonic TM900 reportedly does an excellent job of controlling Moire. The big-sensor hybrid cameras: nope. In fact, even the Sony VG-10 and VG-20 have horrible Moire problems (Moire is one of the things most commonly complained about when it comes to the VG-10 and VG-20, though Moire is not unique to the Sonys…it’s common to almost all videoing done with big sensor cameras design primarily for still photography).

4. Zebra and Peaking: Overexposure and Focus

If ever you have shot video near a bright point of light – for example, if you have shot video in the bright mid-day sun – you will be familiar with the problem that, sometimes, part of your image will be so over-saturated with light that that part looks like a plain, bright-white area, devoid of shape or detail. Unfortunately, such overexposure can be difficult to see by just looking at the image in your viewfinder, so some cameras use “zebra” technology to tell you if lighting is a problem. I have not yet used zebra but, from what I understand, the it overlays an overexposed part of the image (in your viewfinder) with zebra-like stripes. See an example here. By evening-out your light in the actual place you are videoing, and then adjusting the light intake on your camera, you can eliminate the overexposure. You’ll know that the overexposure has been dealt with, because excessive zebra striping will disappear in your viewfinder.

An example of zebra technology

If you would like manually to focus your image, it can be difficult – in the little viewfinder on most cameras – to be sure what is in focus and what is not. Peaking technology to the rescue. In your viewfinder, peaking basically highlights the edges of things that are in focus. Here is an example.

5. A Pox on All Their Houses: Shutter Roll Warping on Most MOS Sensors

Most of the newer sensors used in cameras use a sensor called a MOS, rather than another type called a CCD. From what I gather, the industry is moving away from CCDs and toward MOSs. One of the reasons may be that MOSs are more sensitive than the CCDs (e.g., better in low-light conditions). See the video below for one comparison.

A youtuber compares a CCD camera and an MOS camera under low-light conditions

Most (note: not all) MOS sensors record an image by scanning gradually (well, very quickly, but gradually nonetheless) across the scene, line by line. This method of obtaining the image is called “shutter roll”. In comparison, CCD sensors use a technique that grabs all of the image at the same time (a technique called “global shutter”).

Shutter roll does not cope well with things that move across the scene quickly. If your camera uses shutter roll, quickly moving objects within your scene (or fast pans from side to side) will warp your image. See two examples below.

Watch the bolt on the right appear to warp like it is made out of rubber.

Here is a comparison of CCD (left) to MOS (right) sensors.
The distortions of the MOS image are due to shutter roll.

So, we have a bit of a problem. First, it will be increasingly difficult to find any consumer-level CCD cameras as time goes on and things move toward using only MOS sensors. Second, although CCD sensors use a global shutter, which does not suffer from the warping found on cameras using shutter roll, the MOS sensors are more sensitive (e.g., better in low-light conditions). All of the cameras I describe above – the Canons, the Sonys, and the Panasonic – use MOS sensors. All of them, use shutter roll. All of them will demonstrate the warping found with quickly-moving objects or fast panning from side to side. So: shutter roll probably is not something that should enter into your decision-making. Just sigh, and hope they soon introduce a consumer-grade MOS sensor that uses a global shutter.

Conclusion:

As I see it – based upon my research so far – here is what you should ask yourself if buying a high-end consumer-grade camera capable of recording video:

Q1. Do I mind it if my video recording cannot continue for more than 12 minutes at a time? If you do mind, don’t buy a camera designed primarily for stills, such as the Canons mentioned above.

Q2. Do I want a camera that can give me film-like, shallow depth of field, or the possibility of using a variety of lenses. If so, do not buy the Panasonic TM900. If you don’t mind video recording that cannot go for more than 12 minutes at a time, the best options for stunning sharpness and shallow depth of field are the Canon 7D (if you want to do excellent stills with it as well), or the Canon Rebel T3i (which is less than half the price of the 7D). If you want pretty respectable depth of field effects, but you do not want a camera that prohibits video segments greater than 12 minutes in length, your best options (among those I’ve researched) are the Sony NEX VG-10 or VG-20. NOTE: The VG-20 has several improvements over the VG-10, not the least of which is that the VG-10 used interlaced images and had limited options for frame rates, whereas the VG-20 uses progressive scan and has a better compliment of frame rates.

Q3. Do a want a camera solely for video uses, that is good in low light (almost as good as the big-sensor Sonys), that controls Moire better than the big sensor cameras, but that does not allow me to have film-like, professional-looking depth-of-field controls, and does not allow me to use different lenses? If so, then the answer is the Panasonic TM900. Don’t bother forking out money for the big-sensor DSLRs.

Q4: Do I want my camera to have zebra and peaking technology? If yes, then don’t buy the Canons or the Sonys (unless you want to get into after-market hacks of the firmware on your camera). Get the Panasonic TM900, which comes with both of those technologies.

Well, that’s it for now. I hope that helps. And, please, if you have insights that would help me or those reading this blog entry, please post them in the comments section below. Happy video recording, everyone!