A lens, a hole, a curtain and a sensor. Change focus, aperture, shutter and ISO, then follow the rays and see what the photo does.
START WITH A QUESTIONPick one · predict · then look
EXPOSURE±0 stopsMeter reads EV 10
SUBJECT FOCUSSharpBlur below 0.03 mm
DEPTH OF FIELD2.73–3.33 m0.60 m deep
HYPERFOCAL29.8 mFocus here: sharp to ∞
FIELD OF VIEW39.6°2.12 m wide at subject
MOTION SMEAR0.14 mm≈ 23 px on 24 MP
01 / EXPLORE THE CAMERA
Take it apart.
Light to viewfinder or sensorLight to autofocus
Subject lightBackground lightNear lightDepth of field
Drag to orbit · scroll or pinch to zoom · tap a part
Not to scale: the distance ruler is logarithmic and centred on the runner, so the far side (out to 40 m) and the near side are squeezed by different amounts. Read distances off the ruler. The millimetre gaps behind the lens are exaggerated so you can see them. The numbers are not exaggerated.
IRISf/2.8⌀ 17.9 mm opening
ON THE SENSORSharp pointRing = 0.03 mm limit
02 / THE VIEWFINDER
See the photo.
Open shade · EV 10
1/125f/2.8ISO 100
−3−2−10+1+2+3
● AF
HISTOGRAMTones fit
ShadowsMid-tonesHighlights
YOUR SHOTS · select one to restore its settings
WHAT’S HAPPENING
Read the light.
TRY THIS
Three classic problems.
How the model works
The camera is an ideal thin lens in front of a full-frame 36 × 24 mm sensor. Light from a point at distance u comes to a point at v = f·u / (u − f). The sensor sits where the focus distance would come to a point. If the subject is somewhere else, its cone of light is cut by the sensor as a disc: blur = (f/N) · |v − vsensor| / v.
Depth of field is the range of distances whose blur disc stays below 0.03 mm, the usual full-frame circle of confusion. The near and far limits use the hyperfocal distance H = f² / (N·c) + f. They agree with the blur formula above to rounding.
Exposure counts stops. Doubling the f-number quarters the light (the hole’s area). Halving the shutter time halves it. Doubling ISO adds one stop of gain. The meter compares these settings with the scene light, written as EV at ISO 100. Correct means log₂(N²/t) − log₂(ISO/100) = EV.
Motion smear on the sensor is speed · t · v / u. On a 6000-pixel-wide sensor, 0.03 mm is about 5 pixels.
Honest limits
An ideal thin lens. There is no diffraction, so f/16 never gets softer here, but it would on a real camera. There are also no aberrations, vignetting or focus breathing correction.
The viewfinder is an illustration driven by the model. Each depth layer gets a Gaussian blur sized to its blur disc, and point lights show the disc itself. Real out-of-focus blur is a disc, not a Gaussian.
Background bokeh switches from round to a seven-sided shape when the aperture is stopped down. Real lenses vary.
Grain is an illustration of how gain amplifies noise. It is not a measured sensor noise model.
Field of view is quoted for focus at infinity. Scene light levels are typical values, not measurements.
The f-numbers and shutter speeds are the marked third-stop values. Exposure uses the marked numbers.
The 3D camera is a teaching model, not a particular camera. Its parts sit in the right order and roughly the right places. The rays inside it are drawn to show the path, and the gaps behind the lens are exaggerated. The iris opening follows the f-number exactly. The rings turn with focal length and focus.
“Take a photo” is slowed down 12× to 200× so you can see it, and the status shows the factor. A typical curtain takes about 3.5 ms to cross the sensor. Exposures shorter than that are made by a moving slit, and the animation shows this.