Moving the camera changes spatial relationships. Zooming from one position mainly changes framing.
By the end: Explain apparent background compression without assigning viewpoint changes to focal length alone.
Viewpoint sets relative scale
In an ideal rectilinear projection, image height is proportional to object height divided by distance, multiplied by focal length. Two equal objects at different distances have different apparent sizes. Their size ratio depends on their distances from the camera, not on a common focal-length multiplier.
Moving changes those distances unequally. A camera close to a face makes the nose relatively larger than features farther away. Stepping back reduces the distance ratio. A longer lens often accompanies that move to preserve framing, but the viewpoint change causes the altered perspective.
Calculate which size relationships change
Start with two upright markers, each one meter tall. Put the front marker two meters from the camera and the back marker six meters away. Assume an ideal rectilinear camera, with straight lines remaining straight, and measure distances along the viewing direction. For this distant-subject teaching calculation, projected height equals focal length times object height divided by distance. Using millimeters throughout, a 24 mm focal length gives the front marker a height of 24 times 1,000 divided by 2,000, or 12 mm.
The back marker projects to 24 times 1,000 divided by 6,000, or 4 mm. It therefore appears one third as tall. Replace 24 mm with 48 mm without moving the camera. The front projection becomes 24 mm and the back becomes 8 mm. Both have doubled. Dividing 8 by 24 still gives one third. These calculations describe projected sizes before a sensor boundary crops anything. A projection can exceed the available sensor height, so doubling focal length may exclude an object rather than display all of it.
Now move the camera backward by six meters. The marker distances become eight and twelve meters, because both markers stayed in place. To restore the front projection to 12 mm, use a 96 mm focal length in the ideal calculation. The front becomes 96 times 1,000 divided by 8,000, or 12 mm. The back becomes 96 times 1,000 divided by 12,000, or 8 mm. The front matches the original size, but the back is now twice its original projected height.
Hold one projected height constant
Equal one-meter markers. Vertical scale is ten drawing units per projected millimeter; horizontal positions only separate the markers for comparison. Zooming doubles both heights. Moving back and restoring front height changes their ratio. These are unconstrained projections, not sensor frames or photographs.
This change does not mean that the markers became closer together. Their separation remains four meters. The camera now sees a smaller difference between their distances relative to the distance to either marker. Originally, the back was three times as far away as the front. After moving, it is only one and a half times as far away. Equal objects therefore differ less in apparent size. Calling this compression describes the appearance; it does not identify an optical force exerted by the longer lens.
Equal object height was an assumption, not a general property of backgrounds. Suppose the back marker is two meters tall while the front remains one meter tall. At the near viewpoint, its image height is twice 4 mm, or 8 mm, compared with the front's 12 mm. Its ratio is two thirds. At the far viewpoint, it becomes 16 mm compared with 12 mm, a ratio of four thirds. A background object can therefore appear larger than a foreground object because its actual dimensions differ.
Use a photograph with two recognizable repeated objects to practice this reasoning. First ask whether their real heights are actually similar. Then compare their image heights using any consistent unit, such as millimeters on paper or pixels in an editor. A ratio of 40 to 120 is one third regardless of the unit. The ratio alone cannot establish exact distances unless object heights and projection assumptions are known. It can still reject the claim that a common magnification change, with the viewpoint fixed, changed their relative heights.
Object height and distance both enter the ratio
View and back height
Front image
Back image
Back/front
Near, back 1 m
12 mm
4 mm
1/3
Near, back 2 m
12 mm
8 mm
2/3
Far, back 1 m
12 mm
8 mm
2/3
Far, back 2 m
12 mm
16 mm
4/3
The front marker is always one meter tall. Near distances are 2 and 6 m with 24 mm framing; far distances are 8 and 12 m with 96 mm framing. Relative image size cannot by itself identify distance when real object sizes are unknown.
Framing is not perspective
At a fixed position, changing focal length narrows or widens the field of view. Cropping a wider image can reproduce the central framing of a longer lens in the ideal model, though resolution, optical blur, distortion, and noise need not match. A crop cannot reveal a previously hidden side of an object.
Sensor size also affects field of view at a given focal length. A “35 mm equivalent” focal length compares field of view to a reference format; it does not change the physical focal length. Phone lens buttons may switch modules or crop computationally. Their labels need not imply a continuous optical zoom.
Compare a crop with a narrower optical view
Imagine a hypothetical sensor recording 6,000 pixels across and 4,000 pixels high. From one fixed viewpoint, a longer focal length doubles every central object's linear size on that sensor. In the ideal projection, the same framing can be obtained from the wider picture by retaining its central 3,000 by 2,000 pixels. This crop keeps half the width and half the height. It keeps one quarter of the area, because 3,000 times 2,000 is six million pixels, compared with the original twenty-four million.
Display both results at 3,000 by 2,000 output pixels. The crop already has those dimensions. The full longer-lens capture starts with 6,000 by 4,000 samples and must be reduced. It has more samples available over the shared field of view, but that is not a guarantee of twice the useful detail. Focus, motion, diffraction, lens quality, sampling, and processing may limit recorded structure before reduction. The model predicts framing; image quality requires additional evidence.
To isolate cropping, now compare the full wider image with its central crop, both delivered at 3,000 by 2,000. A feature spanning four source pixels becomes two output pixels in the full-frame delivery and four in the crop. A two-source-pixel blur becomes one and two output pixels, respectively. Cropping doubles their displayed sizes relative to the whole-frame delivery, not relative to source pixels. The feature-to-blur ratio stays two to one. Interpolating a larger export would add values, not new optical measurements. Acceptable sharpness still depends on viewing size and distance.
A half-width, half-height crop keeps a quarter of the samples
The amber source rectangle becomes the entire right frame. Its feature doubles in displayed size, along with existing blur. The upright shape illustrates framing, not an optical resolution test. Interpolated output pixels are not extra source measurements.
Sensor dimensions change the boundary around a projection. Consider a hypothetical 36 mm wide sensor and an 18 mm wide sensor behind a 24 mm lens at the same viewpoint. The smaller sensor records the central half-width. For distant subjects, a 48 mm lens on the wider sensor gives the same horizontal field of view as 24 mm on the narrower sensor. This is the arithmetic behind a two-times focal-length equivalence. The physical 24 mm lens remains 24 mm; the sensor does not change its focal length.
Equal field of view does not settle exposure or total light. With scene light, time, transmission, and f-number matched, exposure per unit sensor area is approximately equal in this distant-scene comparison. If the narrower sensor also has half the height, its area is one quarter as large. It can then collect roughly one quarter as much total light over that frame. ISO or a brighter rendering does not replace that collection. Pixel count, aperture diameter, depth of field, and processing need separate comparisons before making a quality judgment.
On a phone, treat a zoom label as a requested view rather than proof of a particular optical operation. A device may switch cameras, crop, merge measurements, or change processing, depending on its design and conditions. Without a known device, do not infer which occurred. You can still compare central framing from a stationary position and inspect whether nearby and distant objects retain their relative sizes. Small differences may reflect a shifted module position or correction software rather than a general failure of projection geometry.
The sensor selects a boundary within a projection
Both boundaries occupy one image plane. The smaller boundary excludes more of the same schematic projection; it does not enlarge that projection physically. A smaller sensor is not a focal-length multiplier inside the lens. Scene shapes are illustrative, not calibrated optical output.
Work through the mechanism
Move back, then restore foreground size
Blue and gold facade shapes have equal real dimensions and are four meters apart in depth. Both panels preserve foreground height at 120 drawing units by changing the framing multiplier. The gold object's relative height changes with viewpoint. Ideal projection only: no lens distortion, defocus, occlusion reconstruction, or image-quality comparison. Presets immediately replace viewpoint.
Camera to front
Camera to back
Back/front size ratio
2 m
6 m
1/3
4 m
8 m
1/2
8 m
12 m
2/3
The equal-height objects stay four meters apart in depth; restoring foreground framing does not restore perspective.
Use sideways movement to test depth and overlap
The original model changes distance while restoring foreground height. A sideways move exposes another consequence of viewpoint. Imagine a narrow front post at two meters and a back post at six meters, initially aligned with the camera's central direction. Move the camera 0.3 meters to the right while keeping it pointed in the original direction. Both posts now lie to its left. For an ideal projection with a 24 mm framing multiplier, the front shifts by 24 times 0.3 divided by 2, or 3.6 mm.
The back shifts by 24 times 0.3 divided by 6, or 1.2 mm. Their projected separation is therefore 3.6 minus 1.2, or 2.4 mm. The closer post shifts farther because the same sideways displacement is a larger fraction of its distance. This relative shift is parallax. The calculation uses lateral distance and depth in the same units. It describes a fixed camera orientation; turning the camera to recenter the front post changes the exact projection, but does not generally remove the relative-depth effect.
Doubling focal length at the original aligned viewpoint gives no such separation. Both post centers still project to the same location. Their widths increase, but one remains behind the other. A crop around the alignment also cannot reveal the hidden center. This distinction matters when a background line seems to touch a face. A safe sideways or vertical viewpoint change can alter alignment. Merely requesting a tighter frame may magnify the merger. Moving the subject is another geometric solution when the scene permits it.
A sideways move separates initially aligned directions
The posts stay still. The solid sight lines from the new camera position differ from the original common line. Front and back centers were aligned, not necessarily fully hidden. Projection values assume fixed orientation; this plan is not a view through the lens.
Parallax depends on depth difference, not simply on whether you moved. Put both posts at two meters, side by side in a flat display. A sideways translation produces the same projected displacement for each under the fixed-orientation model. Their separation stays unchanged. With objects at two and six meters, their displacements differ. Consequently, a flat poster and a layered arrangement can respond differently even when their first photographs look similar. This is one reason a single image does not uniquely specify the three-dimensional scene that produced it.
Occlusion means that one object blocks rays from another. A newly visible strip after moving is genuinely new scene information, not an enlargement of previously recorded pixels. Its color and texture could not be read from the blocked view alone. Perspective-correction tools can stretch visible surfaces to change their shape in an image. Such a transformation does not measure the wall hidden behind a post. Generative filling may propose a plausible wall, but that proposal is different from documenting the newly revealed surface with another photograph.
For a paper exploration, draw a camera point and two aligned posts on a top-down plan. Add a second camera point to one side and draw straight sight lines to each post. Notice where those lines no longer coincide. Then repeat with the posts on a single depth line across the page. The drawing need not reproduce an actual lens. Its purpose is to predict whether an overlap can change and to identify which surfaces require a new viewpoint before they can be observed.
A new viewpoint can measure a previously blocked surface
This pair is a qualitative projected arrangement, not a scaled camera-movement calculation. The wall pattern behind the original post is absent from that first measurement. Cropping can remove the post from the frame but cannot record the wall it hid.
Control the comparison
The model keeps a foreground object the same displayed height while moving the camera back. This requires a tighter framing multiplier. The background then grows relative to the foreground. It shows geometric rectangles, not a simulation of lens blur or a promise that long lenses always make backgrounds soft.
To test the mechanism, place two stationary objects at different distances. First photograph from one position and compare a wider crop with a tighter framing. Then move back and restore foreground size. Compare the background ratio and occlusion. Avoid standing in roads or unsafe locations to achieve a viewpoint.
Design a comparison that isolates viewpoint
Use two stable objects of similar height on a table or across a room. They need not be cameras or photographic equipment. Keep them separated in depth and slightly separated sideways so that both remain visible. Mark a near viewing position and a farther one with safe floor references. Avoid placing yourself or equipment where people might trip. If photographing is inconvenient, perform the comparison by sight through a rectangular hole in paper, changing the hole's distance from your eye to adjust the apparent framing.
At the near position, record or sketch both objects and note the front object's displayed height. Without moving your viewpoint, make a tighter view if the device permits, or crop the first image afterward. Compare the back-to-front height ratio, not the number of pixels occupied by either object separately. For example, heights of 180 and 60 give one third. A fixed-viewpoint enlargement might produce 360 and 120. The ratio is still one third, even though the tighter view may exclude useful surroundings.
Now move farther away and restore the front object's displayed height by cropping or by any available framing control. Suppose the front again spans 180 pixels, while the back spans 120. Its ratio is now two thirds. That difference is the measurement to explain. Check that you did not move the objects or compare different parts of them. If your device automatically changes cameras, the physical viewing position can shift slightly. Use a generous distance change rather than expecting tiny measurements to isolate unknown device processing.
Compare ratios after matching the foreground
Comparison
Front height
Back height
Back/front
Near view
180 px
60 px
1/3
Same viewpoint, enlarged
360 px
120 px
1/3
Farther view, front restored
180 px
120 px
2/3
These are constructed measurements for equal-height objects. The fixed-viewpoint enlargement preserves the ratio. Restoring only the foreground after moving does not restore the background. Pixel counts must refer to comparable object boundaries.
Restoring one height does not restore the whole scene. Look at side overlaps, how much of the farther object's side is visible, and the spacing between recognizable points. A person photographed nearby may have a nose much closer to the camera than the ears. As a simplified depth calculation, distances of 0.4 and 0.5 meters give an equal-size front-to-back scale ratio of 0.5 divided by 0.4, or 1.25. At distances of 2.0 and 2.1 meters, that ratio is only 1.05.
This face example compares equal-size markers at two depths, not actual nose and ear dimensions or a calibrated portrait prediction. It isolates the depth contribution to relative magnification. Facial shape, pose, lighting, and lens distortion also affect a real portrait. Moving back may reduce near-feature exaggeration but leaves less room for a wide environmental view unless framing changes. The useful choice depends on whether the photograph should emphasize facial proportions, surrounding space, or deliberate proximity. There is no numerical ratio that selects an aesthetically correct portrait.
Finish by explaining a new case without using a lens label as the cause. A mountain looks larger behind a person after the photographer steps far back and tightens the frame. The person is relatively nearby, so the move changes their distance substantially as a fraction of its original value. The already distant mountain changes distance by a much smaller fraction. Tightening the frame then restores the person while enlarging the mountain's projection. This explanation remains valid whether the tighter frame comes from a lens, a crop, or both.
Depth differences matter more at close range
Equal-size markers represent only the relative-depth contribution to magnification. Back height is normalized to 100 drawing units in each comparison. These are not nose and ear dimensions or a simulated face. The same 0.1 m depth difference has a smaller proportional effect farther away.
Exercises
The zoom-only claim
From a fixed tripod position, you double focal length. In an ideal projection, does the back/front size ratio double?
Hint
Both object images receive the same focal-length multiplier.
Worked answer
Quantity
Effect of doubling focal length
Front image height
×2
Back image height
×2
Back/front ratio
Unchanged
A common multiplier cancels in a ratio.
No. Both apparent heights double before any reframing, so their ratio is unchanged. The field of view narrows. This ideal statement excludes lens distortion and other optical differences, but the camera has not moved to a new viewpoint.
Move and restore framing
Two equal objects are 4 m apart in depth. You move from 2 m to 8 m away from the front object, then restore its displayed size. What happens to the background ratio?
Hint
Use distances 2 and 6, then 8 and 12.
Worked answer
Position
Back/front
Near
1/3
Far
2/3
This is the geometry often called background compression.
It rises from 2/6 = 1/3 to 8/12 = 2/3. The background appears twice as tall relative to the front object. The distance change produces this relationship; tighter framing restores only the front size.
Sources and further reading
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.
Relationship between viewpoint, relative distance, framing, and apparent compression. Composition exercises are reasoned alternatives, not experimentally validated aesthetic rules.