The opening controls ray cones as well as light collection. Smaller is not always sharper.
By the end: Explain the benefits and costs of stopping down without assuming one universal best aperture.
Focal length is the distance from a thin lens to the focus of incoming parallel rays. Real compound lenses use effective reference planes, so the number is not their physical barrel length. F-number is focal length divided by entrance-pupil diameter. The entrance pupil is the apparent opening seen through the front of the lens, not necessarily the physical iris diameter. At 50 mm focal length, f/4 corresponds to a 12.5 mm entrance pupil; f/8 corresponds to 6.25 mm.
Area depends on diameter squared. Halving diameter quarters area. This is why an f-number doubling loses two stops. For distant subjects, the f-number largely normalizes light per sensor area across focal lengths, but actual transmission, vignetting, and close focusing still matter.
An f-number compares two lengths, so the units cancel. For a 50 mm lens at f/4, divide 50 by 4 to find a 12.5 mm entrance pupil. For a 100 mm lens at f/4, divide 100 by 4 to find 25 mm. The second pupil has twice the diameter and four times the area. Equal f-numbers therefore do not identify equal holes. They describe opening size relative to focal length.
Why can both lenses give similar exposure per sensor area? Consider a distant, evenly lit wall, equal transmission, and the same exposure time. Doubling focal length doubles the linear size of a small wall feature in the ideal image, spreading its image over four times the sensor area. The pupil's fourfold area admits correspondingly more light from that feature. These factors balance approximately. This concerns exposure per area, not field of view, depth of field, or total light across different sensors.
Keep the focal length at 50 mm and instead change f/4 to f/8. The pupil shrinks from 12.5 to 6.25 mm. Circular area equals a constant multiplied by diameter squared. Because the same constant appears in both areas, their ratio is simply one half squared, or one quarter. At f/16 the diameter halves again to 3.125 mm, giving one sixteenth of the original area. Counting the diameter changes as one-stop changes would underestimate the light loss by a factor of two in stop count.
Suppose f/4 and 1/200 second collect 800 arbitrary light units from a stationary patch below saturation. At f/8 and the same time, predict 200 units. Changing time to 1/50 second multiplies that by four and restores 800. At f/16, 1/200 second gives 50 units, so restoring 800 needs sixteen times the time, or 1/12.5 second. That is 0.08 second. These are arithmetic examples; an actual camera may offer nearby rounded settings rather than precisely these values.
Raising ISO instead does not restore those units. If f/8 and 1/200 second remain fixed, the patch still supplies the modeled 200 units even when a higher ISO produces similar displayed brightness. Conversely, an automatic camera might lengthen time when you select a smaller opening. A similarly bright result then hides a real change in capture duration. To compare aperture effects, read the actual time and ISO if the device reports them. A mode label alone does not establish what remained fixed.
The equal-transmission assumption matters when comparing lenses. An entrance pupil describes geometry, while glass, coatings, internal obstructions, and off-axis shading affect delivered light. Close focusing also changes the simple distant-scene relationship. Do not infer a defective meter from a small disagreement with the ideal ratio. Use the calculation to predict the direction and approximate size of a change, then distinguish transmission or automatic-setting differences from the geometric mechanism. The comparison below keeps those complications absent so that the area and time reasoning remain testable.
| 50 mm lens | Pupil diameter | Time | Light units |
|---|---|---|---|
| f/4 baseline | 12.5 mm | 1/200 s | 800 |
| f/8, time fixed | 6.25 mm | 1/200 s | 200 |
| f/8, time extended | 6.25 mm | 1/50 s | 800 |
| f/16, time fixed | 3.125 mm | 1/200 s | 50 |
| f/16, time extended | 3.125 mm | 0.08 s | 800 |
A point away from the focused distance makes a blur spot rather than a point at the sensor. Restricting the ray cone with a smaller opening reduces that spot. More subject distances then meet a chosen acceptable-sharpness criterion. This interval is depth of field, not a slab in which all objects become perfectly focused.
The criterion depends on final display size, viewing distance, eyesight, and enlargement. A small print can look sharp where a large crop reveals blur. Depth of field also changes with focus distance and framing. Aperture alone does not explain every portrait with a blurred background.
A ray-cone drawing separates focusing from deciding what looks sharp. Imagine one object point whose rays converge 100 mm behind an ideal lens. Let the entrance pupil be represented by a 20 mm opening at the lens plane. Place the sensor 90 mm behind that plane. It intercepts the cone 10 mm before convergence. The remaining distance is one tenth of the full 100 mm cone length, so the cone's width there is one tenth of 20 mm, or 2 mm.
Now halve the opening to 10 mm without changing the object, convergence position, or sensor. The same one-tenth fraction gives a 1 mm spot. Rays still aim at the same convergence point; the sensor has not become correctly focused. Instead, it samples a narrower cone. A sensor 110 mm behind the lens gives the same geometric diameters in this symmetrical construction because the rays have spread for 10 mm after crossing. These large dimensions aid visibility, not camera calibration.
The general arithmetic is opening diameter multiplied by distance from convergence, divided by lens-to-convergence distance. This is similar-triangle geometry for the stated ideal cone. It ignores diffraction and lens aberrations, and it represents the pupil at the simplified lens plane. It is not a way to measure an unfamiliar camera's internal distances with a ruler. What transfers is the proportional reasoning. Halve the opening at a fixed focus error and geometric blur halves; halve that error at a fixed opening and blur also halves.
To connect geometry with output, use a separate, more photographic-sized example. Suppose a recorded defocus spot is 0.020 mm across on a sensor region 20 mm wide. Display that whole region at 200 mm wide. The enlargement is 200 divided by 20, or ten, so the displayed spot is 0.20 mm across. Display the same region at 600 mm wide and the enlargement is thirty, making the same spot 0.60 mm across. Nothing about the optical focus changed between these two outputs.
For a constructed viewing task, choose 0.30 mm as the largest acceptable displayed spot at one fixed viewing distance. This is a declared decision threshold, not a universal property of eyesight. The 200 mm output passes because 0.20 is smaller than 0.30. The 600 mm output fails because 0.60 is larger. If a smaller aperture halves the recorded geometric spot to 0.010 mm, the large output reaches 0.30 mm and just meets this criterion. Diffraction, motion, and sampling could still make the actual image unsuitable.
Viewing the large output from farther away might hide its blur, so dimensions alone do not settle acceptability. Fine text, eyelashes, and broad shapes also demand different detail. When someone reports a depth-of-field distance, ask what output and blur criterion it assumes. The distances within that reported range still produce different spot sizes; they have merely passed the chosen test. This prevents the mistaken inference that every face inside a group photograph's quoted depth of field receives identical optical focus.
Both focal length and pupil diameter use millimeters. Halving diameter quarters circular area.
| At 50 mm | Pupil diameter | Relative opening area |
|---|---|---|
| f/4 | 12.5 mm | 1 |
| f/8 | 6.25 mm | 1/4 |
| f/16 | 3.125 mm | 1/16 |
Use the aperture circles in the original model to predict a depth trade rather than treating them as photographs. Imagine a group with one row near the camera and another farther away. At a chosen focus setting, suppose the near row's geometric spot is 0.024 mm and the far row's is 0.040 mm. Declare a 0.025 mm sensor-plane tolerance for the planned output. The near row passes and the far row fails. These are stipulated spots, not a prediction for a particular lens.
If the pupil diameter halves while the focus setting remains fixed, the cone model halves both spots. They become 0.012 and 0.020 mm. Both now pass the same tolerance. At fixed focal length, this is a doubling of the f-number, for example f/4 to f/8. It costs a factor of four in collection at unchanged time. More depths pass, but best focus stays fixed. Whether this trade helps depends on what else the longer exposure would change.
For example, start at f/4 and 1/200 second with people who sway during capture. Changing to f/8 and 1/50 second restores ideal collection but records four times as much movement at constant projected speed. At 100 output pixels per second, that means 0.5 pixel of motion at the first time and 2 pixels at the second. An optical improvement for the back row can therefore coincide with worse motion detail everywhere. Keeping 1/200 second avoids that time penalty but leaves only one quarter of the original collection.
| Setting | Near / far spot | Pass 0.025 mm? | Motion at 100 px/s |
|---|---|---|---|
| f/4, 1/200 s | 0.024 / 0.040 mm | Near only | 0.5 output px |
| f/8, 1/200 s | 0.012 / 0.020 mm | Both | 0.5 output px |
| f/8, 1/50 s | 0.012 / 0.020 mm | Both | 2 output px |
Changing the arrangement can remove the need for the smaller aperture. If both rows move into nearly the same subject plane, their convergence planes become closer together. A suitable focus setting can then make both geometric spots small without restricting the pupil as much. Moving the camera also changes framing and perspective. The useful question is which depths contain necessary detail. A distant background may not need to pass the face-detail criterion, even if autofocus selects it.
Cropping can make a previously sufficient choice fail. Return to a 20 mm-wide captured region with a 0.020 mm spot, displayed 200 mm wide. Its spot appears 0.20 mm across. Crop to the central 10 mm and enlarge that crop to the same 200 mm output width. Enlargement doubles to twenty, and the displayed spot becomes 0.40 mm. Under the earlier 0.30 mm task threshold, the full view passes and the crop fails. The crop did not change capture focus; it changed how strongly the recorded blur is enlarged.
A paper exercise can test the decision without manual controls. Keep the stipulated group spots and choose either a 0.025 mm or stricter 0.015 mm tolerance. At the stricter threshold, even the halved far-row spot of 0.020 mm fails. Explain whether you would change aperture, focus or subject arrangement, and name the competing cost. A real-device version can compare full views and crops of a stationary textured object. Use this to inspect output dependence, not to certify spot diameters.
Light behaves as a wave. Even perfect focus through a finite aperture spreads a point into a diffraction pattern. At a fixed wavelength, its central-disc diameter on the sensor increases with f-number. Stopping down can reduce geometric defocus and lens aberrations while increasing diffraction blur.
Choose the aperture for the required subject depth and output size, then inspect the result. For a deep still scene, a tripod and careful focus may help; focus stacking is another option but moving leaves or subjects can create merge artifacts. Do not assume that the smallest opening or a software sharpness slider can recover arbitrarily fine detail.
Geometric rays alone predict that an ever-smaller opening always improves an out-of-focus spot. Wave behavior prevents that prediction from continuing indefinitely. For an ideal circular pupil, light from an in-focus point forms a bright central region surrounded by rings. The central region ends at the first dark ring. A common diffraction calculation gives its diameter as 2.44 multiplied by wavelength and f-number. This describes an ideal pattern, not a uniformly bright disc or measured camera blur.
Wavelength is the distance between repeating peaks of a light wave. Choose green light with wavelength 0.00055 mm and distant focus. Multiply 2.44 by 0.00055 to get 0.001342 mm per unit f-number. At f/4, multiply by four to obtain 0.005368 mm, about 0.0054 mm. At f/8 the diameter is 0.010736 mm, about 0.0107 mm. At f/16 it is 0.021472 mm, about 0.0215 mm. Doubling f-number doubles this diameter, whereas it quarters pupil area and collection at fixed time. The two relationships describe different consequences of the same aperture change.
For a 20 mm-wide captured region enlarged to 400 mm, multiply sensor distances by twenty. Those diffraction diameters become about 0.107, 0.215, and 0.429 mm on the output. This does not mean all detail smaller than those numbers suddenly disappears. Real images contain patterns of many sizes and contrasts, and diffraction reduces their contrast gradually. Wavelength, aberrations, sensor sampling, and processing also matter. It explains why stopping down can weaken detail despite correct focus and a motionless camera.
Contrast two scenes. In a flat document perpendicular to the camera, all important detail can lie close to one focus plane. There may be little geometric-defocus benefit from changing f/8 to f/16, while diffraction spread grows. In a small object with important front and back surfaces, the same change may reduce distracting defocus enough to outweigh the diffraction cost. You cannot decide between those cases by ranking f-numbers alone. Nor should you simply add the geometric and diffraction diameters as if they were independent hard-edged circles.
For a controlled optional comparison, use a stationary textured object with detail at two depths. If aperture control exists, keep viewpoint and focus target fixed, change aperture, and note how time and ISO compensate. Support the camera safely and check for subject movement. Compare the near and far details together at the intended output size, then inspect a larger crop separately. A longer exposure, missed focus, or automatic sharpening change can explain softness too. One pair of images rarely isolates every optical cause without additional controls.
Without aperture control, use the calculated cases to make a decision. Suppose the required output tolerates the f/8 diffraction scale but the near or far surface remains visibly defocused there. A smaller opening is a candidate, not a guaranteed improvement. Rearranging the subject, changing the focus choice, or reducing enlargement may satisfy the purpose with less cost. Focus stacking requires suitable captures and alignment, and motion can spoil it. State the required detail and the evidence you would inspect before claiming that any particular setting is best.
A 50 mm lens and a 100 mm lens both use f/4. Are their entrance pupils the same size?
Divide each focal length by four.
| Lens | f-number | Pupil |
|---|---|---|
| 50 mm | 4 | 12.5 mm |
| 100 mm | 4 | 25 mm |
No. Their entrance pupils are 12.5 and 25 mm. Equal f-number does not mean equal physical opening. Under the distant-scene approximation it gives similar light per sensor area at equal time and transmission, not equal framing or total sensor light.
A still scene has all important objects near one focus plane. A tripod image at f/22 looks softer than one at f/8. Is this impossible?
Separate focus error from diffraction.
| Effect of stopping down | Typical direction |
|---|---|
| Geometric defocus blur | Smaller |
| Diffraction spread | Larger |
| Net useful detail | Depends on scene and output |
It is possible. The scene may not need f/22 for depth of field, while diffraction spreads point detail more at f/22. Check focus, vibration, lens behavior, and output size before diagnosing the cause, but a smaller aperture is not a universal sharpness improvement.
The lessons and answers stand alone. These optional secondary references explain the optics and imaging concepts. Numerical examples are constructed, not measured camera specifications. Visual judgments depend on the intended image; the course does not claim a universal aesthetic score.
Definitions, light per unit area, transmission limits, and diffraction. Selected passages checked; the illustrations here are original geometric constructions.
Acceptable blur criteria, viewing conditions, and focus geometry. No universal circle-of-confusion value or camera-specific depth scale is assumed.
Selected textbook passages on diffraction patterns and the wavelength/aperture dependence of angular resolution. This does not establish a universal best camera aperture or an output-independent sharpness threshold.