- Posted by Jon Tarrant
- Comment on this post
Picture lines and image detail
Before explaining why we need lines-per-picture, let's look at what it means... The obvious analogy is an old-fashioned (CRT) TV screen: if you look closely at an image on one of those screens you will see that it is built-up using a series of closely-spaced picture lines. Ancient b&w TVs used 405 lines whereas colour TVs use 625 lines (of which only 576 are used to display the image). Each picture line is separated from its neighbour, so effectively each picture line is really a pair comprising an information-bearing picture line and a black-spacer line.
Any picture that is defined using lines is limited in terms of how much information it can contain by the number of picture lines used. In the digital-TV era and LCD screens we talk about pixels rather than lines but the same principles apply and the common specifications are 720 pixels or 1080 pixels (the "p" that sometimes follows these numbers means "progressive", not "pixels", and refers to the way in which the image is drawn on the screen). Again, the pixels are separated by black spaces: if you grab a magnifying glass (or turn a medium focal-length lens back to front and look at your computer screen through the lens) you will be able to see the individual dots, or tiny coloured rectangles, that define the picture.
So what we are saying is that digital images are defined by their lines or dots - and the more picture lines or dots there are, the more detail the image can convey. And that takes us back to image sharpness testing.
So if we have a lens that can capture 50 line-pairs-per-millimetre and we use a full-frame camera, which has a sensor height of 24mm, then the image can contain a maximum of 1200 line-pairs. This is the figure for the vertical direction and it is used partly for historical reasons (compatibility with TV specifications) and partly because it is smaller than the horizontal figure and is therefore less misleading because it is true, as a minimum, in both directions.
When it comes to sharpness testing using Imatest, the equivalent figure is line-widths-per-picture-height. Note that this is INDIVIDUAL lines, not line-pairs, and it is probably more valid to halve the Imatest value to get line-pairs-per-picture.
This is the point at which we can start to compare the performance of lenses that are used on different-size sensors; specifically, we can now ask whether a certain lens used on an APS-C camera body performs better or worse than another lens (or even the same lens, if it is compatible) used on
a full-frame body.
We have waded through quite a lot of theory here so the practical analysis will have to wait for next time but readers who want to learn more about sharpness and resolution are invited to visit the following URLs for more detail (excuse the pun!).