huugh said:
Gollnick - for the three ccd cameras - I think you confuse video camera and photo camera, AFAIK there is only the Foveon (e.g. in Sigma SD9 and SD10) chip employing exposure for seperate colors but even that is still one chip.
You haven't looked at some of the really high-end professional, studio stuff.
I don't know where did you take your story about A/D converters, but the image is processed by CPU (that's why it is advertised sometimes e.g. "DIGIC II"(Canon) on the box, it advertises better(faster) CPU), slow storing is usually because of low transfer speed to memory.
WHAT? Trust me, transistors and signal amplifiers are the the devices consuming minimum power. They don't slow down image processing.
Let me take you to school.
The image sensor in most digital cameras is one of several forms of a Charge Coupled Device, CCD. It's a little integrated circuit chip not unlike the CPU in your PC. It's fabricated in much the same way.
The transistors on most digital ICs are field-effect transistors, FETs. Without getting overly technical, a FET has two terminals, one called the "source" and the other called the "drain." Electrons enter through the source and exit through the drain, hence the names. Between the source and the drain is a region made out of a material whose conductivity varies with applied electric field, hence the name "field effect" transistor. The easiest way to create an electric field is between the plates of a capacitor. And that's how a FET works. The sensitive region is physically between the plates of a capacitor. Charge can be put into or taken out of that capacitor through a third terminal called the "gate." When the gate is charged, the resulting electric field makes the sensitive region conductive. When the gate is discharged, there is no electric field and so the sensitive region is not conductive. This is a simplification, of course, but it's basically correct and will help us understand how a digital camera takes a picture.
If you were to go to a semiconductor fabrication facility, you'd see one area of the factory where the light is very subdued. This is where the finished IC dies are tested before the wafer is cut apart. Economically, this is a very critical function. The next steps are cutting, bonding, and packaging. These are expensive steps. You don't want to cut, bond, and package bad dies. On the other hand, the steps preceeding this are also very expensive, so you don't want to throw away good dies. This test to determine which dies will be cut out, bonded, and packaged is very important. It's done in the dark for a good reason. Photons of light can actually knock charge off of the gates of FETs causing a good die to appear bad.
It's possible to construct a FET in such a way as to optimize this light-sensitivity. That's how image sensors are made. And this is how a digital camera works: you charge all the gates, expose the die to the image, and then measure the source-to-drain conductivity -- which is a measure of the amount of charge knocked off the gate by photons, i.e. a measure of the brightness of the light in that pixel -- of each transistor. That measurement is made by a sort of Analog-to-Digital converter. For a five megapixel camera, there are classically 15 million transistors to be measured, five million for red, five million for green, and five million for blue. Making that many measurements takes time. But, capacitors, including the gates of FETs, loose their charge naturally over time for several reasons. This sets up the race I referred to.
Cheap digital cameras have one, slow A/D converter. Better cameras have three (or more) fast converters so they can read the pixels faster before that decay process degrades the data. But, three A/Ds consume three tiems as much power and faster ones use more power still.
Cheap digital cameras sometimes try to compensate for this decay in software. They just inflate the levels on the later pixels. But, it's not possible to know the exact rate of decay. It will vary from pixel-to-pixel, with temperature, even over time. Furthermore, one of my favorite laws of physics is: there's no such thing as a free lunch. In this case, when software inflates the levels, it does so at the expense of detail in the picture. That detail can express itself in the perceived sharpness of the image. So, this is yet another reason why cheap digital cameras take fuzzy pictures.
By the way, just as it was possible to design a FET to optimize its light sensitivity, it's also possible to design one to minimize the rate of decay of gate charge. Unfortunately, the two optimizations are mutually exclusive. But, it is possible to design one to retain gate charge for years. This is the basis of the memory cards used in digital cameras.
So, this is why I said that cameras with fewer and slower A/D converters, while cheaper and using less power, take poorer pictures. It gets back to that free lunch.
There is a Chinese proverb that I think defines photography in general. A strict observance of this proverb has helped me improve both my film and my digital photography. The proverb: If pure water flows from upstream, there is no need to filter it downstream. Couple this with the law of free lunches, and you now know all you need to know about digital photography. Start with clean light, keep it clean the whole way, do as little to it as possible, and you will have a good picture. Every time you try and cut a corner, every time you try for a free lunch, it will degrade the quality of your image.
) where some of the extra money you pay is buying high end lens materials with an extra high refractive index allowing the lens to be smaller and lighter.
) and take spare set of batteries with you, the Li-Ion battery packs usually don't last so long and are more expensive. Usually optical zoom function of lens consumes most power. Therefore avoid so called ultrazooms (wide range of focal length, e.g. 28mm - 200mm (equivalent to 35mm)) like Olympus C-750 UZ because the lens is heavier. Ultrazoom lenses are poorer quality too.
