DV Camera Tribunal

One of the key purchasing decisions when setting up a streaming production studio is the choice of camera. Since most streaming today is done at a resolution of 320x240 pixels, even on the high-end, some people are questioning whether or not it is worth the extra $1,000 or so to purchase a "prosumer" camera whose better video quality will ultimately be "dumbed down" for delivery.

To get to the bottom of the issue, we compared two prosumer DV cameras with a much lower-cost consumer camera. Sony’s DCR-VX2000, list priced at $2,999, and Canon’s GL1, listing at $2499, are both three-CCD units. Canon’s Elura 2mc is a single-CCD, very compact camera with a much lower list price of $1599.

Note: Because LCD size, battery life, and other "cool" features don’t generally affect the actual quality of audio and video produced by a DV camera, we decided to keep them outside the scope of this review. While these features can be important for overall ease of use, our intention with this comparison is to focus on the quality of the images and sound as they apply to streaming.

Exhibit A: Video

To test video quality, each camera was used to create a 20-second video with eight separate scenes. Each scene was set up with specific lighting intensities and distances, and different scenes were created to measure motion versus non-motion, color accuracy, and image resolution.

All three cameras record in NTSC DV format, which has a frame size of 720x480 pixels. The video was reduced to 640x480 and 320x240 resolutions for the purposes of our tests. Raw still frames taken from the DV file prior to streaming compression were used to arrive at the quantitative results in Figure 1.

	Lines of Resolution	Video Noise	Video Noise RR*	Image Contrast	Image Contrast RR*
Sony VX2000	430 (280)	4.3 (3.5)	6	64%	8
Canon GL1	425 (280)	2.7 (2.3)	8	66%	7
Canon Elura 2mc	400 (280)	6.2 (3.9)	4	74%	5

Resolution shown in Figure 1 is "limiting resolution," where adjacent vertical lines merge together. In our tests, the Sony VX2000 beat the others by a hair on high resolution (see Figure 3 for a test sample image), but the three cameras produced nearly identical resolutions at 320x240 (resolution values in parentheses).

Video noise is found from a histogram of the pixel values in a small area of the plain white card image (one of the eight scenes). Video noise level shown is the intensity variation as a percentage of the full range. Since pixels are averaged together when reducing resolution, the smaller frame size (represented in parentheses) shows less noise. Here, the Canon GL1 produced the least video noise — by a long shot.

	Bandwidth	Noise Floor (dBfs)	THD+N (dB)	Audio Quality RR*
Sony VX2000	30Hz - 21kHz	-67 (-62 manual)	-55 (-59 manual)	6
Canon GL1	44Hz - 18kHz	-66 (-73 att. on)	-42 (-42 att. on)	4
Canon Elura 2mc	26Hz - 16kHz	-66	-42	4
Aiwa AM-C80	10Hz - 20kHz	-74	-63
Turtle Beach Fiji (PC sound card, line-in)	10Hz - 22kHz	-93	-81

The image contrast test reflects the percentage of the full black-white range, separating the brightest from the darkest square on the color test chart available on the site. A lower contrast number means a more realistic image. A camera with higher contrast will saturate more quickly on bright areas of a scene, reducing the image quality. Here, the Sony VX2000 pulls ahead slightly, with the GL1 on its tail, but leaving the low-cost Elura in the dust.

•