My old Pentax Spotmatic II still gets a workout now and then. For critical work, it's much more versatile than digital. Furthermore, when I scan the 35mm negative, it's the equivilent of about a 35MPixel digital camera -- provided you use film that even supports that resolution. More importantly, the film scanner has 16-bit/channel color depth, most digital cameras have only eight and better digitals have ten.
There are good, technical reasons for both of these differences and they all boil down to time. One of the specs for a digital camera is how many pictures per minute it can take with most in the ten to twenty range. This means that the camera has three to six seconds per image to process everything. But, even more critical is this: the instant you close the shutter, the data in any kind of imaging device (CCD) begins to deteriorate, so you have to get the data out of data device quickly. The rate of that deterioration can be predicted and so the deterioration can be compensated for. But, that rate of deterioration can't be predicted perfectly; it varies from device-to-device, even from pixel-to-pixel within a device, it depends strongly on temperature, some on battery voltage, and it changes over the age of the device. The rate of decay is not linear either. So, the longer you dilly-dally getting those data out of the device, the less accurate the prediction of its decay will be and, therefore, the less accurate the correction will be. The more pixels you have to shuffle around, i.e. the higher the resolution of the camera, the more challenging -- and expensive -- that becomes.
It's a little bit like fresh salmon: the sooner you can get it off the hook, onto the fire, and into your mouth, the better it will taste.
A camera takes the whole image in a split-second and then has to get it all out of the imaging device and processed and stored it as quickly as possible. But a scanner can take just a small fraction of the image, maybe just 1/100th of it, read out, process, and store just that fraction, then do the next 1/100th and so forth. This is why taking a picture takes maybe three second while scanning a frame of film might take 120 seconds.
A camera is a bit like going out fishing on the first day of the month and catching 30 salmon, one to have on each subsequent day of the month, and trying to keep them on ice through the month. By the last week of the month, the remaining few salmon probably will be a litte bit less desirable. A scanner is alot more like going fishing for a fresh salmon just before each meal.
High-end digital cameras have imaging devices with multiple reading ports, but multiple reading ports means multiple copies of the circuitry to read those ports which drives up cost and physical size and drives down battery life.
Increasing the color depth, more bits per channel, also slows down the conversion process which is one reason why digital cameras seldom have more than about ten bits per channel of color depth.
More bits per channel, more color depth, means more detail in an image. One way we perceive detail in an image is as focus. So, no matter how good your lens is and no matter how optically-perfectly you've focused it, your image may look a bit blurry if you don't have color depth.
The number of bits-per-channel is not necessarily a true measure of color depth. Just because an A/D converter has ten bits doesn't mean that all of those ten bits are meaningful. If you don't believe me, then take a picture of a piece of solid-colored paper and examine that picture in your editting software. Watch the color numbers for each pixel and you'll see that even two adjacent pixels are often not exactly the same color. There's noise in them there A/D converters and that noise affects the least-significant bit... or two bits... or even three. So, even if your camera has a ten bit converter, you are probably only getting nine bits and maybe only eight bits of usable color depth.
Furthermore, when you "correct" or adjust color or a brightness or contrast using Photoshop-type software, you necessarily loose color depth. In fact, with the exception of cropping or moving the image, most of the things you can do in editting software have the side-effect of reducing color depth. This is one reason why I preach minimum editting. Instead of trying to "fix" the image in Photoshop, take the image correctly in the first place. But, if you do want to do some editting, the more depth you have going in, the more depth (and remember, depth is also known as detail and focus) you'll have left when you're done.
So, if I take an image with a digital camera that has ten-bit converters and realizes eight bits of usable color depth per channel and then take that image into Photoshop and loose two bits per channel of depth, my final image has six bits per channel of real color depth. But, if I take an image out of a film scanner that has 16-bit A/D converters and delivers 15 bits of meaningful depth (the film scanner losses fewer bits than the camera for a variety of reasons) and then loose the same two bits in editting, my final image has 13 meaningful bits. That's over twice the bits of the image coming out of the camera, but it's actually 128-times the color depth (and remember, depth is also known as detail and focus) of the finished image that started with the camera. Pictures on the web are generally eight bits per channel. Most monitors and printers can't display or print more than eight bits per channel. So, I can take the finished image that started with the film scanner and convert it to eight bits depth and have all eight bits meaningful. That's four times as much color depth -- four times as much detail and focus -- as the image that started from the camera.
By the way, I just swagged that editting would cost two bits of color depth per channel. This depends to some extent on the algorithms used. Not all editting software is created equal. That free stuff that came with your camera or something that you downloaded from a freeware site may cost you more color depth for the same operation as something like real Photoshop.
The alternative is to buy a digital camera that can deliver 14-bits/channel meaningful data. But, those extremely expensive, tens of thousands of dollars.
My Spotmatic II, which -- if bundled with a lot of lenses and accessories -- might fetch $100 on eBay, and my film scanner which cost just shy of $500 several years ago, can deliver images comparable to high-end professional digital cameras costing $20,000 or more.
All of this, of couse, assumes film capable of supporting the resolution and color depth of the scanner.
