Goodness, I would have to re-read all 30 pages of the seminal paper by Mike Kriss. If you can get a copy, it will save a lot of time in discussion here, but he did write a paper on new methods of studying image quality. I'll have to get it out I guess. I did incorporate some graphs from his paper in my book (with permission of the author).
Anyhow, we study grain and sharpness both, but you are interested in how sharp a photo is. Perrrin shows in the first figure, the cone shaped nature of an exposed image, say a dot. It should be a cylinder, but it is cone shaped and the cone can be reversed from what is shown by Perrin. In his example, attenuation makes the image form with the point down, but if you have scatter, then the image is formed with the point up. This is more the case with modern materials in which are added absorbing or acutance dyes to make the image more cylindrical.
Now, how can we observe the ideal? We use X-Ray "dots" instead of or along with light dots. The difference between them is the contribution of light vs X-Ray in scatter. X-Rays do not scatter. Thus, no cone shape.
Now Ross showed that even with ideal exposures, development posed its own problems. For example, HQ is a hardening agent (Quinone is the hardener). This causes pseudo "edges" due to hardening, and formation of Silver metal causes the swell of the gelatin into "bumps" around such a "dot" exposure.
We have to analyze all of this as we go into an R&D cycle. And, we have to compare micro and macro effects. You see, micro and macro contrast are not the same and thus images vary from 4x5 to 35mm due to these effects.
'nuff for now.