Digital camcorders have brought high-quality video capture to the masses, but not all models are equal. If you're looking to buy a video camera, you know there are a lot of specs listed on the box. Unfortunately, they won't tell you what you really need to know, which is how good the footage of your daughter's birthday party, your brother's wedding, or your first independent short is going to look. There's only one way to find out, and that's to shoot a lot of footage under a variety of conditions and take a careful look at the results. We test camcorders in our camera lab under controlled lighting conditions so that we have a solid basis for comparing different models. Under indoor-color-balanced tungsten lighting and low-light conditions, we shoot sample footage of our carefully created scene, which contains a diversity of colors, contrast, textures, and patterns. Under daylight-color-balanced HMI lighting, we shoot a series of test targets to determine how well a camcorder deals with important image-quality challenges such as noise, accurate color reproduction, and sharpness. Then we take the camcorders out in the field to see how they'll perform in typical shooting conditions.

Image-quality tests

Noise test

The laws of physics dictate that it is impossible to digitally capture an image that is completely free of noise. Although, the larger the physical size of a sensor is (not the number of pixels), the less susceptible a sensor usually is to noise. Numerous other variables impact how noisy digitally captured footage can be, including a camcorder's optics and integrated post-processing capabilities.

To evaluate the amount of noise generated by a camcorder, we start by capturing footage of an X-Rite Munsell ColorChecker Chart target (formally known as the GretagMacbeth ColorChecker Chart) using daylight-balanced HMI lighting. The camcorder's white-balance setting is set to either daylight or automatic, depending on which one is more color-accurate. Then we extract still frames, and analyze the images using Imatest image-analysis software to measurably evaluate the generated quantity of color and luminance noise. The footage is captured at the lowest-available compression setting, and at the camera's highest native resolution.

Color reproduction ability test

While color can be corrected via software after video is captured by a camcorder and copied over to a computer, it's usually best to start with captured footage that is as color-accurate as possible--especially for those instances when the footage will not be edited or might be burned directly to disc.

To evaluate how well a camcorder can reproduce accurate colors, we capture footage of the X-Rite Munsell ColorChecker Chart target using daylight-balanced HMI lighting. Footage is captured at the camcorder's daylight white-balance setting. We then extract still frames and analyze the images using the Imatest image-analysis software in order to measurably evaluate a camcorder's ability to reproduce colors correctly. The footage is captured at the lowest-available compression setting and at the camera's highest native resolution.

Sharpness test

The potential sharpness of an image is directly impacted by the resolution of the sensor that captured the image. However, several other variables impact the overall sharpness of a digitally captured image, especially the quality of a camcorder's optics and its integrated postprocessing capabilities.

To evaluate how well a camcorder can create sharp images, we capture footage of an ISO 12233 test target using daylight-balanced HMI lighting, with the camcorder set to automatic white balance. We then extract still frames and analyze the images using the Imatest image-analysis software to measurably evaluate a camera's capability when capturing sharp details. The footage is captured at the lowest-available compression setting, and at the camera's highest native resolution.

Additional lighting tests

In our camera lab, we shoot a fixed scene made up of objects with a wide range of colors, textures, and reflective properties. The video is shot under color-balanced (centered around 3,200K), tungsten studio lights. We use the camcorder's automatic exposure and set the focal length of the lens to a point in the middle of the zoom range for these tests. We set the camcorder to its highest-available quality setting, and test with three different white-balance settings:

• Automatic white balance
• Manual white balance
• White balance preset for tungsten or indoor lighting
  
  

We also test a camcorder's low-light capability by shooting the scene lit by only a shaded lamp with a three-way 150-watt incandescent bulb, at both 150 watts and 50 watts, and then again with all lights turned off. We also test all available special low-light modes using the camcorder's automatic white-balance setting with the lamp set to 150 watts, 50 watts, and turned off.

In addition to the video created in our camera lab, we shoot video in both indoor and outdoor settings. We shoot various scenes in a variety of lighting situations, capturing footage that contains a diversity of colors, contrast, patterns, and light levels. We also shoot video that includes people with a variety of skin tones in order to evaluate how well a camcorder captures them.

We evaluate the quality of the video for sharpness, detail, proper exposure, accurate color reproduction and saturation, white-balance accuracy, dynamic range, and tonal separation. We also look for typical problems that arise with camcorder footage, including image noise, artifacts, and lens distortion. All of the footage is reviewed on a professional video monitor.

If the camcorder offers still-image capture to a flash memory card, we follow our standard procedures for testing digital still cameras and capture photos for evaluation in addition to the video.

Image-quality characteristics

Here are some examples of the image-quality characteristics we consider when evaluating a camcorder.

White balance
	The sample at the top left has correct white balance; the whites and grays look neutral, and there's no discernible color cast. We consider the sample on the bottom left acceptable. Even though the image appears a bit too cool (bluish) and measurements indicate that the blue channel is a bit too strong, it still lacks an obvious color cast. The incorrect sample on the top right has an obvious, distracting color cast.

Exposure
	The top left and right samples are overexposed and underexposed, respectively. Using the correct exposure on the bottom left as a reference, you can see that in the overexposed sample, the sky looks blown out, and you can't tell that there's a glass wall on the right side because it loses the reflection. The underexposed version loses detail and definition in the white pipes.

Color saturation
	The sample to the immediate left most accurately represents our test scene. The sample on the top left has undersaturated colors; it looks as if there is a smoky filter in front of the scene. The sample on the top right is oversaturated, with colors that look too vivid to be real.

Dynamic range
We use various techniques to evaluate dynamic range, but these are two quick indicators: to judge quality of shadows, you should be able to clearly see the gorilla's face and the various shades of brown that comprise it. For good highlights, you should see well-defined reflections off the CD and the detail of the hat's weave.

Noise
	The top left and right examples respectively show low and high noise, and the sample to the immediate left displays moderate noise. The noise in these samples appears as white spots. We consider the quality of the first sample to be good, the second unusable, and the third acceptable.

Testing equipment

Our Labs-based, simulated, real-world scene is photographed using two Mole-Richardson Tweenie II Solarspot lights with 500-watt tungsten bulbs, and a Lowel Omni-light with a 300-watt tungsten bulb. Our fixed scene includes an X-Rite Munsell ColorChecker Chart target and an IRE 100 white card. We use a second IRE 100 white card to set a camcorder's white balance for our tests that require manually setting the white balance.

For our Labs-based low-light test, we use a shaded table lamp with a three-way, 150-watt incandescent bulb on its maximum setting as our sole light source.

Our daylight color-balanced test footage is shot using two 5-inch Altman 575SE-M HMI Fresnel lights. We use a Konica Minolta CL-200 Chroma Meter to regularly check the color temperature of our lights. When the color temperature starts to deviate by more than +/-200 degrees Kelvin from the norm, we swap out the bulbs for new ones.

We evaluate image quality by viewing our test footage on a calibrated 20-inch Sony PVM-20L5 professional multiformat video monitor.