Typically one frame will be at a proper exposure, then the rest are overexposed and underexposed at various increments to let you capture the details in the brightest highlights and the darkest shadows. Then these images are merged together to create a 16-32 bit file.
As I've explained in a few other blog posts and videos on our YouTube channel, most every camera available on the market currently has a sensor in it that can record information with a very specific dynamic range, (somewhere between 5-14 stops of light), which is FAR LESS than what the human eye can perceive. Therefore we combine these bracketed exposures together to create an HDR image.
If you try to take a high contrast shot with a single exposure you typically end up with either completely blown highlights, or all details lost in the shadows.
When you've taken a proper mix of exposures (under, balanced, and over), and merged them with an HDR software app, you're left with a relatively flat and low contrast image. This is where tone mapping comes into play.
What is tone mapping?
It's the process of converting the tonal values of an image from a high range to a lower one. For instance, an HDR file merged from multiple images with a dynamic range of 100,000:1 will be converted into an image with tonal values ranging from around 1 to 255.
Why do we want to reduce that tonal range so much? Well the reason is simple. Most standard display devices (and printers) can only reproduce a low range of dynamic values (between 100 or 200:1 or lower). The goal of tone mapping is to reproduce the appearance of images having a higher dynamic range to fit/display properly on standard display devices, thus keeping the image looking realistic.
The algorithms that tone mapping use to scale the dynamic range down attempt to preserve the appearance of the original image captured by breaking the information up into two categories: global and local.
Global operators map each pixel based on it's intensity and global image characteristics. The process ignores its spacial location or if it's in a dark or light area. Using global only tends to leave you with a flat non-contrasty image after the conversion process.
Local Operators uses the pixels location in the image when analyzing the appropriate scaling for it. This allows each pixel of a given intensity will be mapped to a different value depending on whether it's found in a dark or light area. Local tone mapping requires the system to look up surrounding values for every pixel mapped. This makes it slower (and more memory/system intensive), but leaves you with a much richer and eye pleasing image when correctly done.