Aperture is the first technical control most photographers learn. Open it for less depth of field, close it for more. Open it for more light, close it for less. The exposure triangle slides into place, and the picture brightens or darkens accordingly. This is correct as far as it goes. It is also a small fraction of what aperture is doing inside the lens.
Stop a lens down from f/1.4 to f/8 and several things change at once. Depth of field deepens. Light transmission drops by roughly five stops. Sharpness usually improves at first then degrades again past a point. Vignetting in the corners eases. Chromatic aberration reduces. The shape of out-of-focus highlights shifts from circular to polygonal. The diffraction spikes around point light sources grow longer and more defined. Each of these is a consequence of the same geometric change, but they have different practical implications for the image, and a photographer who treats aperture as only a depth of field and exposure control is leaving choices on the table.
Why most lenses are not sharpest wide open
A lens is a stack of glass elements designed to bring light from a scene to a point on the sensor. The designer has chosen, in compromise after compromise, how the optical aberrations distribute across the image. At the lens’s maximum aperture, all of the available glass is in use, and most of the aberrations are at their largest. Spherical aberration, coma, and longitudinal chromatic aberration are all visible at the edges of the frame and, often, in the centre as well. The image is sharp enough to be usable, but the corners are softer than the centre and the edges of high-contrast areas have a slight haze.
Stop the aperture down by one stop, sometimes two, and the marginal rays of light that hit the edges of the lens glass are blocked off. Only the central, better-corrected portion of the lens is doing work. The image sharpens, the corners catch up to the centre, and the aberrations that depended on the marginal rays diminish. This is why almost every lens has a sweet spot two to three stops down from wide open. A f/1.4 lens is usually sharpest around f/2.8 to f/4. A f/2.8 lens is usually sharpest around f/5.6 to f/8.
Past the sweet spot, diffraction starts to take over. The image gets softer again, not because the lens has worse optical correction at smaller apertures, but because the physics of light passing through a small opening means the focused point is no longer a point. It becomes a small disc of light called an Airy disc, which spreads outward as the aperture closes. For more on the underlying optics, the diffraction tutorial at Cambridge in Colour walks through the math without burying the practical takeaway.
Where the diffraction limit lands
The diffraction-limited aperture depends on the sensor’s pixel pitch. A 24-megapixel full-frame sensor has pixels around 6 microns wide, and diffraction softening becomes visible around f/11. A 60-megapixel full-frame sensor has pixels closer to 4 microns, and diffraction softening becomes visible around f/8. The exact threshold is not crisp, and many photographers shoot well past it when they need the depth of field. But it is real, and the diffraction softness at f/22 on a high-resolution sensor is enough to make a tripod-stabilised image at f/22 look softer than the same scene shot handheld at f/8.
This is one of the reasons macro photographers often stop at f/11 or f/13 and focus-stack to extend depth of field, rather than shooting at f/22 and accepting the softening. The math comes out in favour of the stack.
Vignetting and the recovery of the corners
Vignetting, the darkening of the corners relative to the centre of the frame, has two main causes. Mechanical vignetting is the lens barrel itself blocking off-axis light, and it is worst at wide focal lengths and large apertures. Optical vignetting is the natural fall-off of light intensity at angles away from the lens’s centre, and it follows the cosine-fourth law where light intensity at an angle theta from the optical axis is reduced by cos to the fourth of theta.
Stopping down reduces both. Mechanical vignetting often disappears entirely by f/4 to f/5.6 on a lens that vignettes badly wide open. Optical vignetting reduces but never disappears completely, since the cosine-fourth fall-off is a geometric property of any lens. The practical implication is that a lens with two stops of corner vignetting at f/1.4 may have almost none at f/4, and a portrait shot at f/4 on the same lens will have an even brightness across the frame that the same shot at f/1.4 will not.
Why this is sometimes desirable
Corner darkening is not always a fault. Many portrait photographers add vignetting in post-processing because the natural fall-off of light draws the viewer’s eye toward the centre of the frame. A lens that produces some vignetting wide open is doing this in-camera, for free. Photographers who like the look will sometimes choose to shoot wide open partly for the vignette, accepting the slight loss in corner sharpness as part of the rendering.
Out-of-focus rendering changes as you stop down
Wide open, the aperture diaphragm is essentially a circle, and out-of-focus highlights render as discs. As the lens is stopped down, the diaphragm contracts and the shape becomes a polygon whose number of sides equals the number of blades. A lens with five straight blades produces pentagonal highlights at moderate apertures. A lens with seven rounded blades produces near-circular highlights at the same aperture. By f/11 or so the diaphragm has contracted enough that the polygon is small and the highlights are also small, so the shape difference matters less.
This is one of the practical reasons portrait photographers shoot wide open. Not only for the shallow depth of field, but because the circular bokeh is more pleasing at maximum aperture than the polygonal bokeh that emerges as you stop down.
Sunstars and when they appear
Diffraction has a useful artistic side effect at small apertures. Point light sources in the frame, like the sun glimpsed through trees or a street light at night, produce diffraction spikes that radiate outward from the source. The number of spikes equals the number of aperture blades, doubled if the blade count is odd. A nine-bladed lens produces 18 spikes. A seven-bladed lens produces 14. An eight-bladed lens produces 8, because the spikes from opposite blades line up.
The spikes only appear at small apertures. Stopping down from f/8 to f/16 makes them dramatically more visible. A landscape shot of the sun rising over a ridge will have prominent sunstars at f/16 and barely visible ones at f/5.6, even though the rest of the exposure has shifted to compensate for the aperture difference.
The reading
Aperture is a single number on the camera’s display, but it is several decisions inside the lens. Most of the time the depth of field consideration dominates, and the other effects ride along as consequences. But there are situations where the other effects are the point. A landscape photographer choosing between f/8 and f/11 may be making a sharpness-versus-sunstar tradeoff, not a depth of field one. A portrait photographer choosing between f/1.4 and f/2.8 may be making a bokeh-shape choice as much as a depth of field one.
The body of knowledge a photographer builds about a particular lens is largely a body of knowledge about how that lens responds to aperture. The way the rendering changes across the f-stop range, the sweet spot for sharpness, the aperture at which sunstars start to appear, the aperture at which vignetting disappears. None of this is in the spec sheet, and none of it is obvious from looking at one or two sample images. It comes from shooting the lens through its range and paying attention to what each stop does. The mechanics are described in our notes on the front of the lens, the use of neutral density filters, and our notes on lens mount adapters for moving glass across systems. Once you have done that work for a particular lens, the next image is a series of choices instead of a guess.