Sharpness can be well defined for focus at infinity. Typically, one can measure the so-called point-spread function
, how well focused a point of light becomes in the image plane. I would call a lens 'tack sharp' on my camera body if stars in an image of the night sky remained unresolved by the sensor (e.g. FWHM
<~ 2 pixels).
There are also more stringent conditions. For telescopes, 'tack sharp' would correspond to the diffraction limit
, which is the smallest angle you can possibly resolve given your aperture and "perfect" optics. Typically, a 10 cm aperture corresponds roughly to 1 arcsecond
diffraction-limited resolution. You can compute the diffraction limit for your lens by multiplying the f-number with 100 mm and then divide with the focal length to get the resolution in arcseconds, if you e.g. have a 400/2.8 lens the diffraction-limited resolution would be
2.8 * 100 / 400 = 0.7 arcsec
With a FF sensor the field of view would be ~5 degrees = 18000 arcsec, meaning we would need a 2 Gpix sensor to optimally sample a diffraction-limited 400/2.8 lens. Clearly the diffraction limit is way beyond most regular lenses and mostly relevant for the much longer focal lengths used in telescopes, which regularly do achieve diffraction-limited resolution.