What Is Image Stabilization? IBIS vs Lens IS

Hold a camera up to your eye and try to keep it perfectly still. You cannot. Your heartbeat moves it, your breath moves it, the small corrections your muscles make to hold your arms up move it. In bright light none of that matters, because the shutter closes before the tremor has time to write anything. Let the light drop, let the shutter stretch to a fifteenth of a second, and the same tremor becomes visible as a soft, doubled, slightly smeared frame that looked fine on the back screen and falls apart on a monitor.

Image stabilization is the engineering answer to that problem, and it is one of the most misunderstood features in photography. This field note covers what stabilization physically does, why camera shake grows with focal length, how the stop ratings work and how much of them you should believe, the honest differences between in-body and in-lens systems, when the two combine, when to switch stabilization off, what electronic stabilization costs in video, and the one hard limit that trips up nearly everyone. It sits alongside our shutter speed field note and our field note on sharp photos, because stabilization only ever solves one of the several reasons a frame comes out soft.

What image stabilization actually is

Inside a stabilized camera or lens sits a pair of tiny motion sensors, gyroscopes and accelerometers of the same family found in a phone. They read how the camera is rotating and shifting, hundreds or thousands of times per second, and hand that stream to a small processor. The processor works out where the projected image is drifting on the sensor, and commands an actuator to move something in the opposite direction by exactly the amount that cancels the drift. The thing it moves is either a floating group of lens elements or the sensor itself.

The effect, described plainly: the image the sensor records is held still relative to the sensor even though the camera is not still relative to the world. During a quarter second of exposure your hands wander a fraction of a degree, and the correction system spends that quarter second continuously nudging the optical path back onto the same spot. What lands on the sensor is a frame that behaves as though the camera had been clamped down.

That is the whole idea, and everything else in this field note follows from it. Stabilization is not sharpening, not noise reduction, and not a shutter speed multiplier in any general sense. It is a servo loop that removes one specific source of image movement, the movement of the camera, and leaves every other source completely untouched. Its power and its limits both come from that one sentence.

Camera shake: the problem stabilization was built to solve

Camera shake is what happens when the whole camera moves during the exposure. Every part of the frame smears in the same direction by the same amount, so the background is exactly as soft as the subject. There is no crisp edge anywhere in the image, and zooming in reveals doubled highlights and edges that look like they were drawn with a slightly wet pen. Once you learn to recognize it, shake becomes unmistakable, and the recognition matters because it is the one blur stabilization can actually fix.

Shake is not one motion but several. Your hands rotate the camera up and down (pitch) and side to side (yaw), which are the two largest contributors in ordinary shooting. They also twist it around the lens axis (roll), and slide it bodily up, down, left, and right without rotating at all (the two shift axes). At normal shooting distances the rotations dominate by a wide margin, because a fraction of a degree of rotation moves the projected image far more than a fraction of a millimeter of sideways slide does. Up close, in macro work, the shift axes become significant.

The trigger press is its own contributor, and a large one. The jab of a finger onto a shutter button applies a downward force at exactly the moment the exposure begins, which is why the first frame of a burst is often the softest and the middle frames the sharpest, and why a two second timer transforms tripod work. Stabilization helps with all of this, but technique still removes the largest single spike for free.

Why shake gets worse as focal length grows

A long lens magnifies the scene, and it magnifies your tremor by precisely the same factor. Rotate the camera by one tenth of a degree and a 24mm lens moves the image by a small distance across the sensor; the same rotation with a 240mm lens moves it ten times as far, because focal length is the lever arm that converts angle into displacement. Nothing about your hands changed. The optics simply enlarged the error along with the picture.

This is why a wide angle frame at a thirtieth of a second can look perfectly sharp while a telephoto frame at the same speed looks like it was shot from a moving car. It is also why the viewfinder of a long lens feels so unsteady even before you press anything: you are watching your own tremor at ten times life size. That jitter makes framing and tracking harder, which is a separate cost from the blur itself, and one that stabilization also addresses when it happens inside the lens.

Crop sensors magnify the same way. A 100mm lens on a body with a 1.5x crop factor frames like a 150mm lens on full frame, and the shake behaves like 150mm too, because the sensor is recording a smaller slice of a projected image that got no more stable. Whenever you reason about shake, convert to full-frame-equivalent focal length first. Our telephoto field note works through the reach arithmetic in detail; here the only thing that matters is that equivalent focal length, not the number engraved on the barrel, sets the difficulty.

The reciprocal rule: your unstabilized starting point

The traditional guard against shake is the reciprocal rule: handheld and unstabilized, keep the shutter at least as fast as one over the full-frame-equivalent focal length. A 50mm lens wants roughly 1/50 second, a 200mm lens wants roughly 1/200, a 24mm can survive at about 1/25. The rule is old, it predates digital sensors and the pixel-level scrutiny they invite, and it is best understood as a rough floor rather than a promise.

Two forces push the honest floor faster than the rule suggests. High-resolution sensors resolve smaller amounts of movement, so blur that a lower-resolution file never showed becomes visible at full magnification. And people are not equally steady: cold, fatigue, caffeine, adrenaline, and how much you have walked that day all move the number, sometimes by more than a stop. Many photographers hold themselves to one stop faster than the rule, 1/100 on a 50mm lens, and treat that as the real starting line.

One force pushes the other way, and it costs nothing. Bracing works. Elbows tucked against the ribs, a forearm on a railing, a shoulder against a doorframe, an exhale before the press: for most people that combination buys about a stop over free-standing shooting, and setting the camera on something solid removes the ceiling entirely. This field note treats bracing as worth one stop, and stabilization as a separate benefit stacked on top of it.

How stabilization is measured: stops, and what a stop buys

Stabilization performance is quoted in stops, the same unit the rest of exposure uses, and it means exactly what it means everywhere else: one stop is a doubling. If your unstabilized floor at 50mm is 1/50 second, then one stop of stabilization moves it to 1/25, two stops to about 1/13, three stops to about 1/6, four stops to about 1/3, and five stops to roughly 0.6 seconds. The ladder is clean because time doubles cleanly, which makes the arithmetic easy to do in your head at the moment you need it.

The practical way to use the number is as a subtraction from the focal length denominator. Take your equivalent focal length, that is your unstabilized floor. Then halve it once per stop of stabilization you trust. At 200mm with three trusted stops, 1/200 becomes 1/100, then 1/50, then 1/25. At 24mm with three stops, 1/25 becomes about 1/3 of a second, which is genuinely remarkable and also the point at which most people discover that their own body, not the system, has become the limit.

Notice what the ladder does not say. It says nothing about how long a subject can move, nothing about depth of field, and nothing about noise. It is a statement about one axis of one problem. The chart below shows the ladder at 50mm so the doubling is visible, and every number in it is illustrative rather than a specification for any product.

How honest stop ratings really are

Published stop figures come from a standardized bench procedure, which is a good thing: it makes numbers comparable across products rather than pure invention. It also means the number describes a controlled test rather than a person standing in a dim room at the end of a day of walking. Commonly claimed figures sit somewhere in the range of five to eight stops for current systems, and real-world results for most people fall meaningfully short of that.

The gap is not deceit, it is definition. A bench figure typically reflects a particular success threshold at a particular focal length under a particular shake profile. Your version of the test includes fatigue, cold fingers, a heavier lens than the one on the bench, an off-center grip because you are also holding a coffee, and a much harsher standard for what counts as sharp because you are inspecting at full magnification rather than assessing a print. Any of those alone costs a stop.

A realistic planning approach: assume roughly half the claimed figure, or two to three stops fewer, and treat anything better as a bonus. This field note uses three stops as a working assumption for a single modern system and four when a body and a lens correct together, because those are conservative numbers that hold up in ordinary shooting rather than optimistic ones that hold up on a test rig. Then run the twenty minute test described near the end and replace the assumption with your own measured floor, which is the only number that governs your photographs.

In-body stabilization: moving the sensor

In-body image stabilization, almost always abbreviated to IBIS, floats the sensor on a moveable mount driven by electromagnets. When the motion sensors report that the camera pitched slightly upward, the mount slides the sensor to follow the image rather than letting the image slide across the sensor. Because the correction happens at the last stage of the light path, it works regardless of what is attached to the front of the camera.

That universality is the feature. An in-body system stabilizes your fast prime, your kit zoom, the fifty year old manual-focus lens you bought for the character of its rendering, and anything on an adapter, none of which contain a single electronic component. With manual lenses you usually have to tell the camera the focal length in a menu, since the body cannot ask a lens that has no contacts, and getting that number right matters because the system scales its correction by focal length.

Sensor movement also unlocks corrections that in-lens systems generally cannot perform. Roll, the twist around the lens axis, is naturally handled by rotating the sensor, and the two shift axes are handled by sliding it, which is why in-body systems tend to be the stronger choice for close-up and macro work where shift dominates. The trade is physical: the sensor can only travel so far inside the body before it runs out of room, and that travel limit is what caps how much angular movement a body-based system can absorb at long focal lengths.

In-lens stabilization: moving the glass

In-lens stabilization, variously badged as optical stabilization by different makers, puts a small floating group of elements inside the lens and shifts it perpendicular to the optical axis to steer the projected image back onto the sensor. Because the group sits in the middle of the light path, a very small physical movement produces a large correction at the image plane, and the amount of correction available scales with the design of that particular lens.

Two structural advantages follow. First, the correction is tuned for one optical formula rather than for everything that might ever be mounted, so a long telephoto’s stabilizer can be designed around the very large angular corrections that long focal lengths demand. Second, the stabilized image is what everything downstream sees: the viewfinder, the rear screen, and the autofocus system all receive an already-steadied image. On a long lens that steadying is not a luxury; a jumping viewfinder makes it genuinely hard to keep a distant subject framed, and an unsteady image is harder for autofocus to lock onto.

The costs are also structural. Stabilization adds elements, motors, and weight to a lens, which shows up in price and in size, and it only helps the lens that contains it. Every unstabilized lens in your bag stays unstabilized. On an unstabilized body, that means the benefit you get varies lens by lens, which is a real planning problem when you are deciding what to buy next against your upgrade budget.

IBIS vs lens IS: the honest trade-offs

The comparison is often framed as a winner and a loser, and that framing is wrong because the two systems have different strengths that map onto different kinds of shooting. In-body stabilization is broad: it applies to every lens you own, it handles roll and shift, and it makes old and cheap glass behave better than it has any right to. In-lens stabilization is deep: it is tuned to one focal length range, it can be built with enough correction range for a long telephoto, and it stabilizes what you see and what the autofocus sees.

The honest summary is that in-body is a property of your system and in-lens is a property of one lens. If you own five lenses and one of them is stabilized, a stabilized body improves four more lenses at once. If you own one very long lens and mostly shoot with it, a stabilizer designed for that lens will likely serve you better than a general purpose sensor mount straining at the edge of its travel.

Cost enters too. In-body stabilization is a body feature, so it arrives once and applies forever, and it tends to appear on mid-range and higher bodies rather than the least expensive ones. In-lens stabilization is priced into each lens that has it, which can mean paying for the same capability several times across a kit. Neither pattern is universal across makers or generations, so the practical move is to look at the specific body and lenses you actually intend to own rather than at the general argument.

Why in-lens works better on long telephotos

At 400mm, a tiny angular error becomes a large displacement at the sensor. Correcting it requires moving something far enough to compensate for that displacement, and here the geometry favors the lens. A stabilizing group sitting inside the optical path has leverage: shifting it a small amount swings the projected image a comparatively large amount, and the designer can size that leverage for the focal length the lens actually has.

A sensor mount has no such leverage. It corrects by moving the sensor the full distance the image moved, which means the correction range is capped by how far the sensor can physically slide inside the body before it hits the edge of its travel or the edge of the image circle. At wide and normal focal lengths that travel is plenty. At extreme telephoto lengths the required displacement grows until the sensor simply runs out of room, and the benefit tapers off.

There is a second, less discussed reason. Long lens work is usually about a distant subject you are trying to hold in a small frame, and the stabilized viewfinder that an in-lens system provides makes that dramatically easier. Even if the file would have been equally sharp either way, keeping a bird or a player inside the frame is a task the steadied finder helps with directly. Our sports photography field note leans on this in practice, where shutter speeds are short enough that stabilization contributes more to framing than to sharpness.

Why in-body helps every lens you own

Flip the situation and the in-body case becomes obvious. Most photographers do not own a wall of exotic telephotos; they own a normal zoom, a fast prime or two, maybe a modest telephoto, and often something old and interesting. In-lens stabilization on that kit is patchy: the zoom probably has it, the fast prime probably does not, the vintage lens definitely does not. An in-body system stabilizes the whole shelf at once with no further purchases.

The benefit is largest exactly where fast primes live. A stabilized normal or short telephoto prime is uncommon, and those lenses spend most of their working lives in low light, which is where slow shutter speeds appear. Adding three stops to a 35mm or 50mm prime moves a 1/50 second floor down toward 1/6, and at that point interiors, evening streets, and dim venues become shootable at a reasonable ISO instead of an ugly one. Our low light field note treats this as a system property rather than a lens property for exactly this reason.

Adapted and manual lenses gain the most in relative terms, because they had nothing before. The one caution is the focal length entry: a body cannot query a lens with no electrical contacts, so it applies whatever focal length is set in the menu. Set it wrong and the correction is scaled wrong, which produces either under-correction or, more confusingly, over-correction that looks like a new kind of blur. It takes five seconds to set and is easy to forget when you swap lenses.

When in-body and in-lens work together

Some systems can run both at once and divide the labor rather than duplicating it. In the common arrangement, the lens unit handles the two rotational axes it is best at, pitch and yaw, while the sensor mount handles roll and the two shift axes that the lens cannot address. Because they are correcting different components of the same movement, the combined result is better than either alone rather than a case of two systems fighting.

This only works when the body and the lens are designed to talk to each other, which generally means matched brands and compatible generations. Attach a stabilized lens from one maker to a stabilized body from another through an adapter, and the two systems have no shared language. Depending on the combination the camera may disable one of them, or run both independently and produce a worse result than either alone, because each is trying to correct movement the other has already corrected.

The practical rule: on a matched pair, leave both on and let the system decide. On a mixed pair, check what the camera actually does, and if there is no clear indication, run the twenty minute test with one on, the other on, and both on, and believe the frames. This field note assumes about four stops of practical benefit from a well-matched combined system, one stop more than a single system, which is a deliberately conservative estimate.

What actually gets corrected, axis by axis

It helps to know which movements a system is correcting, because that tells you which failures it cannot touch. Pitch and yaw, the up-down and side-to-side rotations, are the dominant contributors to shake at ordinary distances and are what nearly every stabilization system addresses first. Roll, the twist around the lens axis, matters for horizons and for handheld video, and is generally a sensor-shift capability. The two shift axes, sliding the camera bodily sideways or vertically, matter most in close-up work where the subject is inches away.

Distance changes which axes dominate. Photographing a building across the street, rotation is everything and shift is negligible, because sliding the camera a millimeter barely changes the geometry of something thirty meters away. Photographing an insect at life size, sliding a millimeter moves the framing by a visible fraction of the subject, so shift becomes the loudest failure and a system without shift correction will feel useless. This is the clearest situation where in-body and in-lens systems genuinely differ in kind rather than in degree.

There is also a category no stabilization addresses: sustained movement that is not tremor. Walking, riding, or leaning is a real displacement over a real distance, and correction systems have a limited range of travel and a design assumption of small oscillation around a stationary point. Ask any of them to absorb a stride and they will run out of range instantly. That is the boundary where hardware stabilization ends and gimbals, rigs, and electronic correction begin.

The limitation that matters most: stabilization does nothing for subject motion

This is the section to remember if you remember nothing else. Stabilization measures the camera’s own movement, using sensors physically attached to the camera. It has no information whatsoever about what is happening in front of the lens. A person walking through your frame, a dog shaking itself dry, a hand gesturing during a toast, a car passing behind your subject: none of that registers anywhere in the system, so none of it is corrected.

Play out the consequence. You are in a dim restaurant at 1/8 second, hand held, with a very good stabilization system, photographing someone across the table. The tablecloth is razor sharp. The glassware is sharp. The person’s face, which moved slightly while they were talking, is a smear. The system did exactly what it was designed to do and the photograph still failed, because the failure was subject motion and the tool addresses camera motion. Every stop of stabilization in the world does not shorten the exposure by a single millisecond.

The rule that falls out is simple and worth making a habit. Stabilization lowers your floor for still subjects. It does not lower your freeze speed for moving ones. When anything in the frame is moving, choose the shutter speed the subject demands, get there with aperture and ISO, and treat the stabilization rating as irrelevant to that decision. The freeze speeds themselves are worked through in our shutter speed field note, and the ISO half of the bill is priced in our ISO field note.

How to tell which blur you have

Diagnosing the blur takes ten seconds and saves entire shoots. Magnify the frame to full size and look at a static, detailed part of the background: a doorframe, a brick edge, printed text on a wall. If that edge is smeared, the camera moved, and everything in the frame will be smeared in the same direction by the same amount. That is camera shake, and it is the case where stabilization, bracing, or a faster shutter all help.

Now look at the moving element. If the background edges are crisp and only the subject is streaked, the camera was steady and the subject was not. That is subject motion, and only a shorter shutter fixes it. A third pattern, background smeared and subject relatively crisp, is a pan, deliberate or accidental, where the camera happened to follow the subject.

The fourth possibility is not blur at all. If something in the frame is genuinely, critically sharp, just not the thing you wanted, that is a focus error, and no shutter speed or stabilization setting will address it. Frames often show two of these at once, and the direction of the smear usually tells you which is which. Our sharp photos field note treats all four causes as a single diagnostic system.

When to turn image stabilization off

The classic case is a tripod. A stabilization system is looking for movement, and on a rigidly mounted camera there is essentially none, so with some systems, particularly older ones, the loop can end up chasing its own residual signal and introducing a small oscillation that was not there before. The symptom is maddening: tripod frames that are slightly softer than they should be, at exposure times where nothing should be soft. Many current systems detect a locked-down camera and idle or adapt, so the behavior depends entirely on your specific gear.

The second case is deliberate panning. Swing the camera horizontally to follow a subject and a stabilization system that does not know you meant it will try to cancel the swing, which produces a jerky correction and a smeared frame. Many stabilized lenses and bodies offer a panning mode that ignores horizontal movement and corrects only the vertical, which is exactly what a pan needs. If yours does not, switching stabilization off for panning is usually better than leaving it on.

The third case is not a hazard but a non-benefit. At very short shutter speeds, well above your reciprocal floor, there is not enough time for tremor to matter and the system contributes nothing to sharpness. It still steadies the viewfinder, which is worth having on a long lens, and it draws a little battery, which matters on a long day. Leave it on if you like the finder, switch it off if you are counting power, and stop expecting it to change the files.

Panning mode and what it changes

Panning mode deserves its own explanation because it is widely misread as a general purpose setting. In a panning mode the system stops correcting along one axis, usually the horizontal one, and keeps correcting the other. The logic is that during a pan the horizontal movement is the photograph, and the vertical movement is the mistake, so you want the vertical wobble in your swing removed while the sweep itself is left completely alone.

That makes it a specialized setting, not a better setting. Leave a body or lens in panning mode for ordinary shooting and you have thrown away correction on the axis where much of your shake lives, so your static frames get worse. Some current systems detect a sustained sweep and switch behavior automatically, which is convenient and also means the mode switch may do nothing on that gear. Read what your equipment actually offers rather than assuming.

Vertical panning exists too, following a subject moving up or down, and where a system supports it the axes simply swap. The underlying principle is the one worth carrying: stabilization cannot distinguish movement you meant from movement you did not, so any mode selection is you telling it which is which. The panning technique itself, the swing, the follow-through, the shutter speed, is covered in the panning section of our shutter speed field note.

Electronic and digital stabilization in video

Video adds a second family of stabilization that works in software rather than hardware. Electronic stabilization reads the camera’s motion, then shifts and sometimes warps each recorded frame so the sequence appears steadier than the camera was. To have somewhere to shift the frame to, it needs sensor area outside the recorded picture, which it takes by cropping in. That crop is the price, and it is unavoidable: room to move has to come from somewhere.

The crop is commonly on the order of ten to thirty percent of the frame depending on the mode and how aggressive the correction is, with stronger modes taking more. That means a wide lens stops being as wide, and in a small room or a tight interior that loss is genuinely felt. Aggressive modes may also warp the edges of the frame to compensate for rotation, which can bend straight lines near the border, and on wide angle footage that bending is visible if you are looking for it.

Electronic stabilization also inherits a video specific problem. Rolling shutter, where the sensor is read line by line rather than all at once, already skews fast movements, and heavy software correction applied on top of a skewed frame can produce the wobbling effect that video people call jello. The most reliable handheld video comes from hardware stabilization doing the heavy lifting with modest electronic correction on top, plus the boring truth that a smooth operator with good technique beats any amount of correction applied to a bouncy walk.

Stabilization or higher ISO: which stop to buy

When a scene is too dark for your shutter speed, you have a small set of moves and they all cost something. Open the aperture and you pay in depth of field and possibly in lens sharpness at the extreme. Raise ISO and you pay in noise and some editing latitude. Add light and you pay in setup and in the look. Slow the shutter and you pay in blur risk. Stabilization is unusual because it lets you take the fourth option at a much smaller risk than normal, and it charges nothing in image quality to do so.

That makes the ordering clear for a still subject. Spend the stabilization stops first, because they are free. Only when you have slowed the shutter as far as your tested floor allows should you start paying in ISO. A photographer with three trusted stops shooting a still interior at 50mm can work at 1/6 instead of 1/50, which is three stops of light gathered, which is three stops of ISO not spent: ISO 800 instead of ISO 6400 for the same brightness, with all the file quality that implies.

The ordering inverts the instant anything moves. A moving subject sets a hard shutter floor that no amount of stabilization lowers, so the stabilization stops are simply unavailable and ISO has to pay the whole bill. This is the practical reason a fast lens and a stabilized lens are not interchangeable purchases: the fast lens buys stops that work on moving subjects, the stabilized lens buys stops that work only on still ones. The exposure triangle field note sets out the whole trade, and the upgrade budget planner is where you price which stops are worth buying.

What stabilization does not replace

Stabilization does not replace a tripod. Tripod work is not about a stop or two, it is about exposures measured in seconds and minutes, about identical framing across multiple frames for blending, and about a camera that stays exactly where you put it while you adjust something. No correction system holds a frame still for thirty seconds, because thirty seconds of drift is not tremor, it is travel. Our phone tripod field note covers the small end of support, which is where most people start.

It does not replace a fast lens either, for the reason described above: aperture stops work on everything, stabilization stops work only on still subjects. It does not replace good technique, because trigger jab and a poor stance still contribute movement, and stabilization has to spend part of its correction range absorbing errors you could have removed for free. It does not replace focus accuracy, and it certainly does not replace light.

What it does replace, honestly, is a category of missed shots. The frame in the museum you would not have gotten. The evening street at 1/8 second. The handheld telephoto at dusk. The video clip that would have been unwatchable. Those are real photographs that exist because of stabilization, which is a substantial thing for a feature to do. The mistake is only ever in expecting it to do the other jobs too.

A worked example: one dim room, four ways

Take a specific scene. An interior at the end of the day, 50mm equivalent lens, aperture wide open at f/2, and a meter reading that says a correct exposure needs 1/50 second at ISO 3200. The subject is a still room, and there is also a person in it who is talking. Four approaches, all illustrative.

Unstabilized and free standing, the reciprocal rule puts the floor at 1/50, which is exactly what the meter asked for, so there is no room to move and ISO 3200 is the bill. Brace against a doorframe and buy a stop: 1/25 second, ISO 1600, with the same brightness and a visibly cleaner file. Add three stops of stabilization on top of the bracing and the arithmetic reaches 1/6 second at ISO 200, four stops down from where you started, which is the difference between a noisy file and a clean one.

Now include the person who is talking. Their face needs roughly 1/125 to be reliably sharp, which is faster than the meter’s 1/50, so the shutter goes up rather than down and the exposure has to be bought back somewhere else. Wide open already, so ISO climbs from 3200 to about 6400. The stabilization stops sat there unused, because they were never available for this problem. Two photographs, one room, one setting, and the only variable that changed was whether something in the frame was moving.

Common image stabilization mistakes

The largest by far is expecting stabilization to freeze a subject, and it is worth restating because the failure is so convincing: the frame is well composed, the technique was good, the camera was steady, and the picture is still soft where it matters. The guard is a habit rather than a purchase. Before you slow the shutter to take advantage of stabilization, ask whether anything in the frame is going to move during the exposure. If the answer is yes, that shutter speed is not available to you.

The second cluster is configuration. Leaving a body permanently in panning mode and losing correction on the axis that matters. Forgetting to enter the focal length for a manual lens on an in-body system, so the correction is scaled for the wrong lens. Running two mismatched systems that were never designed to cooperate. Each of these produces frames that are worse than no stabilization at all, which is why they are so confusing to diagnose.

The third is trusting the badge over your own frames. A published stop figure describes a bench test, not you, and the difference is often two or three stops. Photographers who plan their evening shooting around a claimed eight stops end up with a folder of soft frames and a theory that their copy is defective. Run the test, write down the honest number, and use that. Our camera settings reference collects the stop ladders you will need while doing the arithmetic.

How to test your own stabilization in twenty minutes

The test is simple, and the result is the only stabilization number that governs your photographs. Find a detailed static subject in reasonable light: a bookshelf, a brick wall, a page of text taped up at eye level. Set the camera to shutter priority, put the lens at a focal length you actually use, and stand at a distance that fills the frame with detail. Turn stabilization off to start.

Shoot five frames at each speed down a ladder: 1/125, 1/60, 1/30, 1/15, 1/8, 1/4, and slower if the light allows. Then repeat the entire ladder with stabilization on, and if you have a matched body and lens pair, run a third pass with both engaged. Do not change your stance between passes, and do not brace, because you are measuring the system rather than the furniture.

Now inspect every frame at full magnification and, for each speed, count how many of the five are acceptably sharp. Your honest floor is the slowest speed where at least three or four of five pass. The gap between the stabilization-off floor and the stabilization-on floor, counted in stops, is your real benefit: from 1/60 to 1/8 is three stops. Repeat the test at a long focal length, since the answer changes, and repeat it braced, since that answer changes too. Twenty minutes buys you two or three numbers you will use for years.

How much stabilization should weigh in a buying decision

Weight it by what you actually photograph, not by what sounds impressive. If a large share of your shooting happens indoors, at dusk, in venues, at long focal lengths, or handheld on video, stabilization changes which photographs exist rather than merely making them nicer, and it deserves real priority in the budget. If you shoot mostly outdoors in daylight, sports and action at short shutter speeds, or from a tripod, it is close to irrelevant and the money does more good elsewhere.

The comparison that trips people up is a stabilized slow zoom against an unstabilized fast prime at a similar price. Both solve dim light and they solve it differently. The stabilized zoom gives you flexible framing and slow-shutter capability on still subjects. The fast prime gives you genuine light, which works on moving subjects and also produces the shallow depth of field that separates a subject from its background. Neither is the right answer in general; the right answer depends on whether the things you photograph hold still.

One more consideration: in-body stabilization is bought once and applies to everything you will ever mount, which makes it a compounding purchase in a way a single stabilized lens is not. If you expect to build a kit over years, that argues for weighting it at the body decision. If you expect to own two lenses forever, it matters less. Run your own version of this against the upgrade budget planner before deciding which line item wins.

The bottom line

Image stabilization is a servo loop that senses camera movement and cancels it, either by shifting glass inside the lens or by floating the sensor inside the body, and everything true about it follows from that one description. It buys you stops of handheld exposure time, roughly two to four of them in honest practice rather than the larger figures quoted from bench tests, and each stop doubles the shutter time you can hold. In-lens systems hold long telephotos steadier and stabilize the viewfinder and autofocus; in-body systems improve every lens you own, including manual and adapted glass, and handle roll and shift that lens units generally cannot. On matched pairs the two divide the work and do better together. Switch it off on a tripod if your system needs that, use a panning mode when you swing deliberately, and remember that electronic stabilization in video pays for its smoothness with a crop. Above all, hold on to the limit: stabilization corrects the camera and nothing else, so a still scene gets slower shutters for free while a moving subject still demands every millisecond of genuine shutter speed you can pay for.

This field note describes how stabilization systems behave in general and names no camera, lens, or manufacturer on purpose, because the engineering principles hold across every system while specific performance does not. Every stop count, shutter speed, percentage, and crop figure in it is an illustrative reference point chosen to make the relationships legible, not a measured specification or a claim about any product, and published stop ratings come from standardized bench procedures whose conditions differ from yours. Run the twenty minute test on your own body and lens, write down the floor you actually achieve, and trust your own magnified frames over any number printed on a box or repeated in an article, this one included.