About color depth p.2 - 32 bit

I've already talked about 8 bit and 16 bit color space. Now i'll tell you about 32 bit.

So what's the difference? There are two of them. Let's look at a range of shades of gray.


At first glance it looks weird: from zero to one. But if in 16 or 8 bits, the minimum scale is 1, then there is an integer, then the 32 bits scale is one hundredth, thousandth and ten thousandth. I looked into different programs, the maximum partition ratio was 1 / 100000.
100,000 shades per channel against 32,729 in 16 bit and 256 in 8.
First difference is: in 32 bits, all operations are with floating point, ie not with integer numbers.

Now let's have a look at this: It was taken the same image in 16 and 32 bits, and exposure in each image has been reduced by one.


At 16 bits, white became gray that is predictable, and in 32 bits in white area appeared details and disk of the sun became visible. This is because the 8 bit and 16 bit images all the information that above white just clip. If we talk about the 32 bit images, then they have such concepts as whiter than white and blacker than black.


Note the difference of color values in different color spaces. In 32 bits area of the sun disk has a value greater than 100.

Remembering the example of a ruler: 8 bits is a 30 centimeter ruler, in which color is measured in centimeters, 16 bits is the same ruler, but measured color in millimeters, and 32 bits is much more longer ruler:


The second difference can be stated as follows: 32 bit images contain information beyond the black and white. 32 bit do not clip anything.

The most common formats for storing 32-bit images are an exr and hdr.

Practical application:
Photographers can shoot without worrying about exposure settings, since it can be corrected later in a photo editor.

For 3d artists: you can render without worrying about the brightness of light sources, and set it later in any image editor or the composing program.

Advantages:
-Storing information whiter than white and blacker than black.

Disadvantages:
-Heavy files
-Slow processing

Resume
I always work in either 16-bit or 8, because they are processed faster. And edit carefully and take care of the absence of overburned areas is not hard.
I'm using 32 bits only if i am going to work with it in compositing program, because changes that i do can be devastating for non-float point images. It is also reasonable to use 32 bits for a very bright image with tiny shades (ie high-key). In such images it is sometimes difficult to not clip some color of the image.