"The process of focus is the adjustment of the relative positions of the optics and the imaging surface so as to yeild the smallest possible star size on the imaging surface when all other conditions are unchanged".
What is unchanged? Optical size and quality can affect the size of the star even in a well focused image, but those can be considered constant (since you probably won't be switching scopes in mid-exposure). Seeing can also affect star size but for the sake of comparing one level of focus accuracy to another, even this must be considered unchanged (even if it probably won't be).
Certainly the image on the right below is unfocussed. Although in this case it was blurred after the fact, it illustrates the point well. It is the detail, the high frequency information, that is missing in a blurred or unfocussed image. Since this detail is the "holy grail" of astroimagers, it is clearly not something we can sacrifice lightly.
One needs to have good focus not just to improve the detail alone. If science is your goal, good focus will almost always (with a few exceptions) result in better science. Also be aware that with today's modern guided systems, poor focus will make for poorer guiding which will only make a bad situation worse.
OK, this is a good news/bad news situation. The good news is that it is not necessary to achieve a precise POINT of focus. It is only necessary to stay within a RANGE of focus. Wallis and Provin put it this way in their 1988 book "A Manual of Advanced Celestial Photography":
"With all cameras and imaging systems there exists a range of distance that the focal plane can be moved without making a visible change in the image that is formed..."
The bad news is that this range is not very large as the table below illustrates.
| Focal Ratio | Depth of Focus (+ or -) | Total (maximum) Range |
|---|---|---|
| F4 | .002 | .004 |
| F5 | .0025 | .005 |
| F6 | .003 | .006 |
| F7 | .0035 | .007 |
| F10 | .005 | .01 |
It was suggested by Wallis and Provin that the photographer use the smaller, first set of numbers as a practical maximum. I have found this to be correct. The larger second set of numbers simply allows one to be too "close to the edge". Since it is very easy to drift out of focus, the extra margin is probably a necessity under most conditions.
The fact that depth of focus is focal ratio dependent brings another point into focus (pun intended). The diagram below shows why with the same focus spot size, a faster system with its steeper light cones will have a narrower band of focus (distance "B" vs. distance "A"). This means it is harder to be within the focus band on an faster system. I have sometimes heard imagers say that their faster systems "snap" into focus more easily. What I think this shows is that they are moving toward focus more quickly as they turn the focus control, but in reality will have a harder time achieving a truly accurate "within the band" focus.
