Although both NTSC, HDTV, VGA, and SVGA are acceptable television and computer visual media, they have less than one megapixel resolution, the minimum deemed appropriate for even entry level digital cameras. Simple comparisons of pixel counts obscures the issue, however. First, television pictures are meant to be viewed from a relatively far distance, for example 5 ft for NSTC. Second, SVGA (and better) computer images are surprisingly clear even close up due to their relatively small, distinct, dot pitch size per pixel (e.g. 0.28 mm or smaller) compared to the large size of the screen.

Another important consideration is that television and computer screens display each pixel in potentially all of the colours of the rainbow. While the fundimental picture element in a television screen and computer monitor is a pixel, in a digital camera it is an image sensor. The pixels in a digital camera obtain their range of hues and brightnesses from an array of distinct red, green or blue image sensors. The distribution of digital camera image sensors is 50% for green, 25% red, and 25% blue There are almost as many pixels as there are image sensors. (If my math is correct, a camera with 1 million full-color pixels would have 1,004,004 image sensors: 251,001 red, 502,002 green, and 251,001 blue.)

The process of combining RGB sensor information into full-color pixels and the implication for digital camera picture quality are outlined as follows. The color of a 'pixel' is the weighted average of the respective luminance values of a block of 9 red, green, and blue image sensors. How this works is seen with the aid of the representation of a partial image sensor grid below; the numbers at the margin are for grid cell references.

	1	2	3	4	5	6	....
1	G	R	G	R	G	R	....
2	B	G	B	G	B	G	....
3	G	R	G	R	G	R	....
4	B	G	B	G	B	G	....
5	G	R	G	R	G	R	....
6	B	G	B	G	B	G	....
....	....	....	....	....	....	....	....

Starting with the upper left corner, the first image pixel is centered on cell 2/2 (row 2, column 2) and its colour value is calculated from the respective red, green, and blue luminance values of image sensors at cell 2/2 and those at the 8 cells surrounding it (i.e. 1/1, 1/2, 1/3, 2/1, 2/3, 3/1, 3/2, and 3/3. The next image pixel in the row is centered one image sensor to the right, and the next image pixel in the column is centered one image sensor below. (For a better pictural explanation refer to www.shortcourses.com/choosing/how/03.htm.)

The important effect of this scheme is that the calculation of pixel colors next to each other are based on the luminance information of many of the same image sensors. Six of the 9 image sensors are the same for immediately neighboring pixels. Three of the nine image sensors are common to pixels centered two cells away. Thus, the colors of pixels are a moving average of luminance data of blocks of 9 image sensors. This limits the ability of pixels next to each other to have large changes in color value. As a result distinct edge lines in digital pictures, upon close inspection, appear to be fuzzy or blurred. Thus the need for picture sharpening, discussed below. (Before leaving this particular topic, it is interesting to note that a new image sensor technology exists whereby an individual sensor obtains all RGB luninance values. The technology is described at www.dpreview.com/news/0202/02021101foveonx3.asp. Presumably, digital cameras incorportaing these new sensors would not be subject to bluring caused by interpulation averaging.)

I will not ponder on the fact that the colour value for each image sensor in a typical digital camera is limited to a scale of 8 bits, that is 256 discrete luminance levels of either red, gree, or blue. When the 8 bits of the red, green, and blue image sensors are combined in a pixel, however, they produce a 24 bit potential pallette, consisting of over 16 million colours. This is a lot of colours/shades and the human eye/brain may not be able to discern all of the subtleties between them. In any case, many computer video cards have the same limitation 24 bit limitation. Therefore, when viewing images on a computer screen the 24 bit colour resolution of a digital camera does not present any further restriction on potential image quality.

Another compromise in digital camera resolution compared to other electronic visual media is the use of image compression. Digital cameras invariably employ JPEG compression to allow an image to be stored in one-quarter to one-tenth of the computer/camera memory needed for an uncompressed image. The compression process averages hue information for clusters of pixels (perhaps 2x2 pixels) but retains their unique luminance (i.e. gray scale) values. The effect of compression may reduce effective pixel resolution by a factor of 25%, depending on the compression ratio. In extreme cases compression results in 'blotchy' artifacts in images.

Most digital cameras process images to 'sharpen' them. Images can also be sharpened outside of the camera by computer image editing software. The effect is to de-blur high-contrast boundaries within the image to make them appear distinctive. The result can be visually pleasing, but sharpening does represent a further adulteration of image information. Too much sharpening results in a brittle, crystalline looking picture.

When viewing images on a computer monitor there is a further compromise in image quality when the pixel dimensions of the screen are not an exact multiple of the pixel dimensions of the image. Viewer software processes image information to force fit a 2048x1536 picture, for example, onto an 800x600 pixel screen, which tends to blur the picture on the screen. One lesson here is that if a picture needs additional sharpening by software, the sharpening should be applied to a picture in its final, showable size. My first digital camera did not have internal sharpening so I had to apply sharpening via image viewer software. To process the images for the slide show I first converted them to 800x600 pixels, then sharpened them in that format.

So what is the conclusion of all this discussion about pixels? The conclusion is that what we know about pixel resolution from dealing with television and computer monitor (and printer) images does not strictly relate to digital cameras. If it did, then a digital camera that takes the 800x600 (one-half megapixel) image would be all of the resolution that is ever needed to produce perfect pictures for viewing on a typical computer screen. Quite the contrary, a digital camera with only one-half megabyte resolution does not yield very detailed and clear computer screen images. As noted above, digital camera images are subject to limitations that no amount of resolution will completely overcome. As their resolution increases there is a definite improvement in picture quality, all else being equal. But it is not accurate to say that images from a two megapixel camera are twice as clear as those from a one megapixel camera, etc.

WHAT IS THE CONCLUSION ABOUT DIGITAL CAMERA RESOLUTION BASED ON EXPERIENCE

As I mentioned before, my first digital camera was a very basic one megapixel model. It did not have optical zoom nor extensive controls to manually override the camera's automatic settings for picture lighting and focus. Never the less, the pictures from this camera produced a very satisfying computer slide show of 500 pictures documenting my wife's and my RV trip. I played the slide show, and individual computer images, to over a dozen people. The responses were always about the beautiful and interesting content of the pictures. Not once did anyone complain about their quality. I have also printed a number of the images in 8"x10" pictures, and their quality is quite acceptable. Again, when people look at these pictures the potential lack of quality does not deter from the enjoyment of the pictures' content.

The one megapixel camera does a particularly good job of capturing detail in close-up pictures, within 2 meters or so. If the primary interest in using a digital camera is to take intimate pictures of the kids and pets, then a one megapixel camera may be all that is needed.

My second digital camera is a three megapixel model, there is a noticeable improvement in picture quality. Where this makes the most difference is in landscape pictures with the higher resolution providing more detail in far off foliage, mountain crags, and animal shapes. I prefer the quality of the pictures of this camera to that of my one megapixel unit, but in this regard I may be more critical than most. Certainly, not all that many people are willing to spend the considerable extra money to buy a three or more megapixel camera. If my first digital camera had been a reasonably capable two megapixel model I might not have been tempted to replace it.

OTHER IMPORTANT CONSIDERATIONS

Pixel resolution is only one aspect bearing on the capability of a digital camera to produce satisfying results. The camera's underlying optics must be of high quality. Even with film cameras, poor optics produce poor pictures. My first digital camera did not have optical zoom; the next one had 3X optical zoom. Having used both I like zoom, because it is easier to frame a picture to cover and highlight the most interesting parts of a scene.

There is a lot of similarity between digital camera models, but each has features and characteristics that make it different and worthy of religious loyalty. I will not delineate the pros and cons of the various models here. I have one experience-driven opinion, however, that is important for RVers. I strongly recommend a digital camera that uses easily replaceable AA batteries. RVers may be faced with the need for fresh batteries when shore power is not available to charge a camera's internal battery. Rechargeable AA batteries are economical, and chargers are available that run off an RV's 12 volt power supply and that are solar powered. In a pinch, an RVer can usually buy a fresh set of AA batteries at almost any destination. Having a dead camera in a location of beauty or interest is a very frustrating experience.