Perspective distortion in product photos is why a 600 mm-deep sideboard can look 42% deeper at its front edge than at its back edge — in a photo where every measurement you took was correct. The buyer is not being misled by a careless photographer. They are being misled by arithmetic, and there is no lens on the shelf that fixes it.
Five beliefs keep circulating in seller groups about this. Four are wrong, and the fifth is right for the wrong reason. Here they are, with the numbers attached.
Myth 1: "Wide-angle lenses distort your product"
The truth: distance distorts your product. The lens only decides where you stand.
Perspective distortion is defined as a warping of an object relative to its surroundings caused by the relative scale of nearby and distant features — and the cause is stated flatly in the optics literature: linear perspective changes are caused by distance, not by the lens itself. Two frames shot from the identical spot with a 24 mm and a 105 mm lens have identical geometry. The 105 mm frame is a crop of the 24 mm frame, nothing more.
Wide lenses got the blame for behavioural reasons. A wide lens lets you fill the frame from close range, so you do, and closeness is what stretches the near edge.
The mechanism is one line of arithmetic. Apparent size falls off with distance, so for a product of depth d, with the camera D away from its front face, the front face sits at distance D and the back sits at D + d:
front-to-back exaggeration = (D + d) ÷ D
Take a sideboard 2,000 mm wide and 600 mm deep — round numbers, easy to follow:
| Camera distance to front face | (D + 0.6 m) ÷ D | What the buyer actually sees |
|---|---|---|
| 1.0 m | 1.60 | Front edge renders 60% larger than the back edge |
| 1.5 m | 1.40 | 40% larger |
| 2.0 m | 1.30 | 30% larger |
| 3.0 m | 1.20 | 20% larger |
| 6.0 m | 1.10 | 10% larger |
The rule falls straight out of that column: stand back about ten times the product's depth and front-to-back exaggeration drops below 10%. At five times the depth you are at 20%. At two times the depth you are at 50%, and the photo is now telling a different story than your spec sheet.
Myth 2: "Switch to a 50 mm and the problem goes away"
The truth: 50 mm has no special powers. It just makes you walk backwards.
The "normal" focal length for a format is defined as roughly the diagonal of that format. For the 24 × 36 mm frame that diagonal is 43.27 mm, and anything from about 40 mm to 58 mm counts as normal on full frame. Fifty became the default because Oskar Barnack fitted one to the first Leica — not because 50 is optically neutral. The often-repeated claim that 50 mm matches human vision is contested; the eye's own nominal focal length is closer to 17 mm.
So a 50 mm does not correct anything. It narrows your angle of view, which forces you to step back, and stepping back is what does the work. You can get the identical result by staying on a wide lens and cropping — a crop and a longer lens are the same operation.
That matters because of frame-fill rules. Amazon requires the product to fill at least 85% of the main image frame, so you cannot solve distance by standing far away and leaving air around the product. You stand far away and then crop in. Same geometry, compliant frame.
Myth 3: "It's my phone's fault"
The truth: the focal length is a symptom. The 1.4 metres it forces you to stand at is the disease.
Most smartphone main cameras land between 24 mm and 27 mm in 35 mm-equivalent terms — the iPhone's main camera is commonly quoted at 26 mm equivalent, Samsung's Galaxy main camera nearer 23 mm. That number is not a defect. It is a working distance sentence.
For horizontal framing on a full-frame-equivalent sensor, the distance needed to just fill the frame width W is:
D = (W ÷ 2) × f ÷ 18
(18 mm being half the 36 mm frame width.) Run it for that same 2,000 mm-wide, 600 mm-deep sideboard:
| Lens (35 mm equivalent) | Distance to fill frame width | Front-to-back exaggeration |
|---|---|---|
| 26 mm (typical phone main camera) | 1.44 m | 42% |
| 35 mm | 1.94 m | 31% |
| 50 mm | 2.78 m | 22% |
| 85 mm | 4.72 m | 13% |
| 105 mm | 5.83 m | 10% |
Same product, same frame fill, same "correct" photo — and the near edge swings from 42% oversized to 10% oversized purely on where your feet are. This is also why furniture shot in a small studio reads wrong no matter who holds the camera: the room will not let you get to 6 metres. Our walk-through of how to photograph a product for measurement covers the setup side of that constraint.
Myth 4: "Lens distortion and perspective distortion are the same thing"
The truth: one is a lens defect your software already fixes. The other is geometry your software cannot.
Barrel and pincushion distortion are optical distortion — straight lines bow outward or inward because of the lens design. Phones and raw converters correct them automatically from a stored lens profile, which is why you rarely see bowed edges any more.
Perspective distortion is not a defect at all. It is the mathematically correct projection of the scene from the point where you stood. There is no lens profile for "stood too close," because nothing went wrong optically.
The related-but-separate case is the keystone effect: tilt the camera up or down and vertical edges converge. Levelling the camera on a tripod removes convergence. It leaves the depth exaggeration exactly where it was, because that comes from the ratio of distances, not from the angle of the camera back. Square-on and wrong is still wrong.
Myth 5: "Put a hand or a coin in the shot and scale is solved"
The truth: your scale reference obeys the same rule your product does.
Place a hand at the front of the product and it sits closer to the lens than the product's back edge, so it is inflated by the same (D + d) ÷ D factor — except now the inflation is applied to the one object the buyer is using as their ruler. A coin taped to the front face of a 600 mm-deep cabinet, shot from 1.5 m, is rendered 40% larger relative to the back of that cabinet. The buyer reasons from the coin and lands on the wrong number.
There is a second, quieter failure. A scale reference only works if the viewer already knows the reference's size, and knows it in their units. A model in the frame has an unstated height. A coin is a different diameter in every market. Comparative props narrow the guess; they do not answer the question. We went through the trade-offs of props versus labels in show product size in a photo.
What actually fixes perspective distortion in product photos
In order of payoff, not effort:
- Move your feet before you touch the lens. Get to roughly ten times the product's depth, then crop or zoom to hit frame fill. This is the whole fix for exaggeration.
- Shoot the depth-critical view square-on and level. A tripod at the product's mid-height, camera back parallel to the front face. This removes convergence so the remaining error is only the distance effect.
- When the room won't let you back up, stop trying. Shoot the front elevation flat — accept that it shows no depth — and carry the depth on a second image as a labelled number.
- Put the measured number on the image. A photo, however well shot, contains no scale. A labelled dimension does.
That fourth point is the one sellers skip, and it is the only one that closes the loop. Even a geometrically perfect photo answers "what shape is this?" and never answers "how big is this?" A dimension that is measured against the product's real edges and printed on the image is the only version of the answer the buyer cannot misread. It is also the difference worth understanding between measuring and generating: an image model asked to add dimensions will produce a plausible-looking number, because plausible is what it optimises for; snapping a measurement to the actual edge of the actual object produces a correct one. Plausible costs you a return. If you want that in currency, run your own numbers through the return cost calculator.
Quick reference: how far back for under 10% error
| Product depth | Stand back at least | Realistic? |
|---|---|---|
| 100 mm (small appliance, boxed goods) | 1.0 m | Yes, trivially |
| 300 mm (wall cabinet, shelf unit) | 3.0 m | Yes in most studios |
| 450 mm (dining chair, side table) | 4.5 m | Tight but doable |
| 600 mm (sideboard, desk) | 6.0 m | Needs a real studio |
| 900 mm (sofa, armchair) | 9.0 m | Rarely — use elevation plus labels |
Read down that last column and you see why furniture and large industrial goods are the categories where photography alone stops being a viable answer to size.
FAQ
Why does my product look distorted in photos?
Almost always because the camera was too close. Perspective distortion in product photos has one dominant cause: the near face of the product is a shorter distance from the lens than the far face, and apparent size is inversely proportional to distance, so the near face renders larger. At 1.5 m from a 600 mm-deep product that error is 40%. Lens choice matters only through the distance it pushes you to.
Does a longer lens reduce perspective distortion?
Only indirectly, and only if you actually move. A longer lens shot from the same position gives an identical perspective to a wide lens shot from that position — it is a crop. The reduction happens because a longer lens forces you to step back to keep the product framed.
What focal length should I use for product photography?
Work backwards from distance, not forwards from focal length. Decide how far back you can physically get, then pick the focal length that fills the frame from there using D = (W ÷ 2) × f ÷ 18. In a normal studio that usually lands between 85 mm and 135 mm full-frame equivalent for medium goods. Anything wider than about 35 mm equivalent should be treated as a compromise you are accepting, not a choice you are making.
Can I fix perspective distortion in Photoshop or Lightroom?
You can fix lens distortion and converging verticals — both are standard corrections. You cannot recover the true depth ratio, because that information is not in the file. A perspective warp forces the edges straight by stretching pixels, which fixes the look of the box and leaves the proportions wrong.
Will a better photo remove the need for a size chart?
No, and this is the useful thing to internalise: a photograph carries shape, finish and context, and carries no scale at all. The reliable pattern is one clean, square-on, far-enough photo plus the measured dimensions locked onto it where the buyer is already looking — snapped to the product's real edges, exported at each platform's spec-diagram size. A worked example of that pairing is in this furniture spec diagram breakdown. Tools built for dimension annotation do this deterministically; general photo editors and image generators do not, because they have no measurement to snap to.
Sources & References
- Perspective distortion (photography) — definition, extension and compression distortion
- Normal lens — diagonal-of-format rule and the 43.27 mm figure for 24 × 36 mm
- Distortion (optics) — barrel and pincushion distortion
- 35 mm equivalent focal length — how phone camera focal lengths are quoted
- Amazon Seller Central — product image requirements (85% frame fill)