The Ultimate Guide to Thermal Scopes in Australia →
Thermal Academy · Core Curriculum
It’s a still night in an open paddock and your thermal picks up a warm shape out near the tree line. The rangefinder says 780 metres. The box the scope came in promised a detection range of 2,600 metres, so this should be easy — except it isn’t. You can see something is there, glowing softly against the cool ground. What you can’t tell is what it is. Pig? Dog? A neighbour’s calf? A roo sitting in the grass? The heat blob refuses to become an animal, no matter how long you stare.
Nothing is wrong with your thermal. It’s doing exactly what the number on the box said it would: at 780 metres it has comfortably detected a warm object. The trouble is that detection was never the question you actually needed answered. The question you needed answered was can I tell what that is, and is it safe and legal to shoot? — and that’s a completely different, and much shorter, range.
So: how far can a thermal really see? The honest answer is that “see” isn’t one distance — it’s three. A thermal can detect much further than it can recognise, and recognise much further than it can identify. The framework that explains why, and by how much, is nearly seventy years old and still sits underneath every detection-range figure ever printed on a thermal spec sheet. It’s called the Johnson Criteria, and once you understand it, you’ll never read a detection-range number the same way again.
The hero statement
Every thermal can detect further than it can recognise. Every thermal can recognise further than it can identify. These aren’t three names for the same range — they’re three separate, shrinking distances, and the biggest one is the one printed on the box.
Here’s the line to keep in your head for the rest of this guide, because everything else hangs off it.
Detection tells you something exists.
Identification tells you whether you should take the shot.
Thermal Academy
Detection is the number that sells the optic. Identification is the number that keeps you legal, ethical and safe behind the trigger. The whole point of this article is to show you why they’re so far apart, exactly how far apart, and what actually moves the gap — so you can read a spec sheet like someone who understands what they’re looking at.
What Is the Johnson Criteria?
The Johnson Criteria is a model — a rule of thumb backed by real experiments — that predicts how much detail a sensor has to put on a target before a human can perform a given visual task: notice it’s there, tell what kind of thing it is, or identify exactly what it is. It was worked out in 1958 by John Johnson, an engineer at the US Army’s Night Vision Laboratory, who ran observers through trials with early night-vision equipment and measured the point at which they could reliably detect, recognise and identify targets.
What Johnson found was that these tasks don’t depend on some vague notion of “sharpness.” They depend on a countable amount of detail spread across the target — and each task needs a predictable amount more than the last. That single insight is why, decades later, a manufacturer can print three different range numbers for the same scope and have them all be true at once.
Quick Definition
Johnson Criteria — a 1958 model that links a visual task (detection, recognition, identification) to the amount of sensor detail, measured in “cycles across the target,” needed to achieve it with a 50% probability. It’s the reason detection range, recognition range and identification range are three different distances rather than one.
Before we get into the numbers, it helps to see where this all sits. A detection-range figure isn’t a property of the sensor alone — it’s the end result of a whole chain of components, each of which has its own Thermal Academy guide. The target gives off heat; the lens gathers it; the sensor turns it into pixels; the processing builds a picture; the display shows it to you; and your eye reads whatever detail survived the trip. The Johnson Criteria is simply the yardstick we hold up at the very end to ask: was there enough detail left to identify what you were looking at?
Every spec on the sheet — lens, sensor, pixel pitch, NETD, processing — exists to get detail down this chain and onto your eye. The Johnson Criteria is the final test: of all the detail that started at the target, was there enough left at the end to tell what it was? Detection, recognition and identification are just three different pass marks on that same test.
Detection, Recognition, Identification: Three Questions, Not One
The three Johnson tasks are really three escalating questions you ask about the same warm shape, each demanding more detail than the last.
Detection answers “is something there?” — nothing more. A warm blob stands out against a cooler background. You can’t tell what it is, which way it’s facing, or whether it’s alive; you only know that heat is present where the background isn’t. This is the easiest task, so it happens at the longest range, and it’s the number that makes the box look impressive.
Recognition answers “what class of thing is it?” — a pig versus a deer, a person versus an animal, a vehicle versus a tree. Now the shape has to carry enough detail that its general form is unmistakable: four legs and a low body, or two legs upright, or the blocky mass of a ute. Recognition needs several times more detail than detection, so it happens much closer.
Identification answers “exactly what is it, specifically?” — this particular boar and not a sow with piglets nearby; a fox and not a small dog; a legal target and not a protected or non-target animal. Identification is the demanding one: you need enough detail to be certain, because this is the range at which a responsible shooter is willing to act. It happens closest of all.
Left to right, the same warm animal carries more and more detail. A blob is enough to detect. A clear four-legged outline is enough to recognise a pig. Only a detailed image lets you identify a specific, legal target — and each step only becomes possible as the animal gets closer.
What Is a “Cycle”? The Currency of the Whole Model
To make any of this measurable, Johnson needed a way to count “detail on target.” The unit he used was the cycle — sometimes called a line pair. Picture a set of black-and-white stripes laid across the width of the target: one black stripe plus one white stripe is one cycle. The question the model asks is simple: how many of those stripes can your sensor still tell apart across the critical dimension of the target? That count of resolvable cycles is the currency the whole Johnson Criteria is priced in.
Why stripes? Because resolving a pair of light-and-dark lines is the fundamental test of whether a system can render detail at a given fineness. And there’s a hard floor set by physics: to capture one cycle you need at least two pixels — one for the light line, one for the dark. That’s the Nyquist limit, and it’s why the whole game comes down to how many pixels the sensor lands across the target. More pixels on target means more cycles resolved, which means you can climb from detection to recognition to identification.
One cycle across the target is barely a smudge — enough to know something’s there. Four cycles gives you a shape you can classify. Around six and a half gives you enough to be sure. It’s the same target every time; all that changes is how much fine detail the sensor can still resolve at that range.
The Johnson Numbers: 1, 4 and 6.4 Cycles
Here’s the part that turns the whole idea into something you can actually use. Johnson’s trials produced a set of numbers — how many cycles across the target each task needs, for a 50% chance of a trained observer succeeding. They’ve been refined over the decades, but the classic figures still quoted today are these:
| Task | Cycles across target | What you can tell | Real example |
|---|---|---|---|
| Detection | 1.0 ± 0.25 | Something warm is present | “There’s a heat source by the fence” |
| Orientation | 1.4 ± 0.35 | Which way it’s facing / rough aspect | “It’s side-on, moving left” |
| Recognition | 4.0 ± 0.8 | What class of object it is | “That’s a pig, not a deer” |
| Identification | 6.4 ± 1.5 | The specific object, with certainty | “A lone boar, no piglets, legal to take” |
The classic Johnson cycle criteria. Detection needs about one cycle across the target; identification needs roughly six and a half. Orientation is the often-forgotten fourth task that sits between detection and recognition.
Look at the jump from detection to identification: from about one cycle to about six and a half. Identification demands roughly six times as much detail on the target as detection does. And detail on target is set by how many pixels the sensor lands on the animal, which depends directly on how close the animal is. That six-fold gap in required detail is exactly what turns into a very large gap in range — which is the whole reason the number on the box and the number that matters to you as a shooter are so far apart.
Why Detection Range Isn’t Identification Range
This is the heart of the article, and it follows directly from the cycle numbers. The amount of detail a sensor puts on a target shrinks in proportion to distance: push the animal twice as far away and it subtends half the angle, so you get half as many cycles across it. Turn that around and you get the rule that governs everything: the range at which you can perform a task falls in proportion to the cycles that task demands.
Detection needs about 1 cycle and happens at the longest range. Recognition needs about 4 cycles — four times the detail — so it happens at roughly a quarter of the detection range. That figure carries a published tolerance of ±0.8 cycles, so the honest version is a band rather than a point: about a fifth to just under a third of the detection distance, for the same target under the same conditions. Identification needs about 6.4 cycles, so it happens at roughly one-sixth of the detection range — again a band once the ±1.5-cycle tolerance is applied, roughly an eighth to a fifth. That’s not a manufacturing flaw or a marketing trick; it’s simply what the geometry forces.
The single most important picture in this guide. The detection zone reaches all the way out; recognition reaches about a quarter as far; identification only about a sixth. The animal that’s a clear, shootable boar up close is nothing but a warm smudge out at the detection limit — same optic, same animal, three very different distances.
Put real numbers on it and the gap becomes vivid. Take a scope advertising a 2,600 metre detection range on a man-sized target. By the cycle ratios, recognition lands somewhere around 650 metres and identification down around 400 metres — or, applying the published tolerances, roughly 550–800 m and 330–530 m. Every one of those numbers inherits the assumptions baked into the 2,600 m headline: a man-sized target, clean air, and a 50% chance of success. They are ceilings calculated under ideal conditions, not distances to plan around. Thinner thermal contrast, humidity, dust, clutter, an animal that is nothing like man-sized, or simply wanting to be more than half sure all pull the usable figures shorter than the model predicts — never longer. Either way the lesson is identical: the range at which you can safely take the shot is a small fraction of the range printed on the box.
What this model is, and what it is not. The Johnson criteria are a probabilistic imaging model — not a rule, not a guarantee, and not a measurement of your optic. Every cycle figure above is an N50 value: the point at which an ensemble of trained observers, viewing a defined target under defined conditions, succeeds about half the time. It is also a 1958 model built on scale models viewed through early image intensifiers; modern sensor engineering has largely moved to the Targeting Task Performance (TTP) metric, which handles today’s focal-plane arrays more faithfully. The cycle numbers endure because they remain the clearest way to explain why the three ranges differ so much — not because they are the last word. A model prediction is never confirmation of a target’s identity. Never use a predicted detection, recognition or identification range to decide whether a target is safe or legal to engage. Positive identification is made with your own eyes, at a distance where you are certain, and the responsibility for it is yours alone.
If you want to see where the numbers come from rather than take them on faith, the arithmetic is refreshingly plain — it’s all just pixels landing on an animal. Work an example through and the whole model stops feeling abstract.
The maths in action
Take a 640-wide sensor on a lens giving a field of view about 12 metres wide at 400 metres. Spread 640 pixels across 12 metres and each pixel covers roughly 19 mm of the scene at that range. A pig’s critical dimension — call it its shoulder height — is around 750 mm, so about 40 pixels land across it. Two pixels make one cycle, so that’s roughly 20 cycles on target: comfortably past the ~6.4 needed to identify. Now push the same pig out to 1,600 metres, four times further: each pixel now covers ~75 mm, only about 10 pixels land across the pig, and you’re down near 5 cycles — enough to recognise it’s a pig, not enough to be sure it’s a legal target. Same optic, same animal; distance alone dropped you from “identify” to “recognise.” That’s the Johnson Criteria doing its work.
The one takeaway
If a thermal advertises a big detection range, your usable range — the distance at which you can identify a legal target and take a responsible shot — is roughly a quarter to a sixth of it, before you even factor in your rifle, your hold and the conditions. Never treat the detection figure as your shooting range.
Are Manufacturers Lying About Detection Range?
It’s tempting to feel cheated when you learn all this — as if the big number was a con. It isn’t, and it’s worth being fair about why. When a spec sheet says “detection range 2,600 m,” that is very often technically true: under the right conditions, on the target size they’ve defined, the sensor really can put roughly one cycle on that target at that distance. The number isn’t fabricated. It’s just answering the least useful of the three questions.
What manufacturers are doing is choosing, quite reasonably from a marketing point of view, to quote the largest of the three legitimate ranges. Detection is always the biggest number, so detection is what gets printed in bold. The problem isn’t dishonesty; it’s that buyers assume “range” means “the range I can shoot at,” when the figure quoted is the range at which the scope can first tell that a warm object exists.
Two more things quietly inflate these numbers, and knowing them makes you a sharper reader of any spec sheet. First, the target size: detection range is always quoted against an assumed target, and a bigger assumed target gives a bigger range. Many manufacturers use a NATO-standard 2.3 × 2.3 m target or a 1.8 m “man-sized” one — larger than a fox or a rabbit, so your real detection range on small game is shorter than the sheet suggests. Second, ideal conditions: the figures assume high thermal contrast and clear, dry air. A humid night, rain, dust or a target barely warmer than its background all pull every one of the three ranges in.
The 50% Catch: It’s a Prediction, Not a Promise
There’s one more piece of fine print in the Johnson Criteria that almost nobody mentions, and it matters. Those cycle figures — 1 for detection, 4 for recognition, 6.4 for identification — are pegged to a 50% probability of success. At the recognition range, the model predicts that a trained observer will correctly recognise the target about half the time. Not always. Half.
So a DRI figure is a statistical prediction, not a guarantee. “Identification range 900 m” doesn’t mean you’ll cleanly identify every target at 900 m; it means that, under the defined conditions, roughly half of attempts would succeed at that distance. Push in closer than the quoted figure and your odds climb well above half; sit right on it and you’re effectively flipping a coin. This is a huge part of why experienced shooters treat published identification ranges as an optimistic ceiling and do their real identifying much closer, where the probability is comfortably high rather than marginal.
Why the 50% matters in the field
A quoted identification range is the distance at which you’d succeed about half the time in ideal conditions. That’s nowhere near good enough to bet a shot on. Treat the published number as the furthest identification is even theoretically possible — then close the distance until you’re genuinely certain, not statistically hopeful.
What Actually Moves These Ranges
If identification range is really “how far can the sensor put 6.4 cycles on the target,” then improving it means one of two things: putting more cycles on the target at a given distance, or making each cycle easier to see. Every meaningful spec on a thermal optic works through one of those two levers, and understanding which is which tells you where your money actually goes.
Sensor resolution is the biggest lever, by far. More pixels across the sensor means more pixels — and therefore more cycles — land on the target at any distance. Stepping from a 384 to a 640 sensor puts noticeably more detail on the same animal at the same range, which is why resolution moves recognition and identification range more than almost anything else. If you care about the ranges that matter for a shot, resolution is where to spend.
The lens sets how those pixels are aimed. A longer focal length spreads the sensor’s pixels across a narrower slice of the world, so each pixel covers a smaller patch of the target — more cycles on target, longer identification range, at the cost of a narrower field of view. This is the trade at the centre of choosing a lens: reach versus how much country you can watch. Pixel pitch works alongside it — a smaller pitch packs more pixels behind the same lens, adding cycles without adding bulk.
Sensitivity makes each cycle easier to see. A low NETD (thermal sensitivity) means the sensor can pull faint temperature differences out of the scene, so the detail it does capture is clean rather than buried in noise. Resolution and lens decide how many cycles reach the target; NETD and processing decide how clearly you can actually see them — which matters enormously when thermal contrast is poor.
And then there are the things you don’t buy but can’t ignore. Target size: a big boar carries more cycles than a rabbit at the same range, so every DRI range is longer for large game and shorter for small. Thermal contrast: an animal much warmer than its background (a pig on cold ground) is far easier than one barely warmer (the same pig on a hot afternoon’s residual heat), and low contrast shortens all three ranges. The atmosphere: humidity, rain, fog and dust absorb and scatter infrared, so a muggy or dusty night quietly strips range off every figure on the sheet.
Everything that improves identification range does one of two jobs: it lands more cycles on the target (resolution, lens, pitch) or it makes those cycles clearer (NETD, processing). The third box is the weather and the animal — real factors that move your ranges every night, whatever you bought.
How to Read a DRI Figure on a Spec Sheet
Armed with all this, you can now read a detection/recognition/identification listing the way it’s meant to be read. First, if only a single “detection range” is quoted, mentally generate the other two: recognition is roughly a quarter of it, identification roughly a sixth. That instantly turns a headline number into the distances you actually care about.
Second, check the target the figures are based on. A DRI set quoted against a 1.8 m man-sized or 2.3 m NATO target tells you very little about your detection range on a fox — scale it down for smaller game. If two brands quote wildly different detection ranges, the first thing to suspect isn’t a better sensor; it’s a bigger assumed target. Cross-brand DRI comparison is only fair when both are quoting the same target size, and they frequently aren’t.
Third, remember the conditions baked in: these are fair-weather, high-contrast, 50%-probability numbers. On your actual worst-case night — humid, low contrast, a nervy target that won’t stand still — every figure comes in. Read DRI as an optimistic ceiling, not a working promise.
Why This Matters for an Ethical (and Legal) Shot
All of this stops being academic the moment your finger is near a trigger. The single most important rule in thermal hunting falls straight out of the Johnson Criteria: you may only shoot once you have positively identified the target and confirmed the range — not when you’ve merely detected it, and not even when you’ve only recognised its class.
Shooting at a detected-but-unidentified heat source is how non-target and protected animals get hit, how someone’s working dog or livestock takes a round, and how an unsafe shot goes off into country you couldn’t properly see. Recognition isn’t enough either: “it’s probably a pig” is a thought, not a target. You need identification — this specific, legal, safe-to-take animal, with a clear background — before the safety comes off.
That gives you a simple, unbreakable sequence for the field, and it’s the discipline that separates a hunter from a hazard:
Detect at distance, identify up close, range it, then decide. Detection and recognition are permissions to keep looking, never permissions to fire. The only rung that unlocks the trigger is positive identification with the range confirmed — and if you can’t reach it, the answer is simply no shot.
The field rule in one line
Detect at distance, identify up close, range, then decide. If you’ve only detected or recognised, you keep watching or you close the gap — you don’t shoot. A responsible shooter lets the animal come into identification range rather than gambling a shot at detection range.
Detection Sells the Scope; Identification Governs the Shot
The Johnson Criteria is really one idea wearing a lab coat: seeing that something is there and knowing what it is are different jobs, and the harder job needs far more detail — so it only becomes possible much closer. Detection, recognition and identification aren’t three words for the same range; they’re three shrinking distances, and the marketing department will always quote you the biggest one.
So the next time a thermal boasts a detection range measured in thousands of metres, you’ll know exactly what it’s telling you and what it isn’t. It’s telling you the range at which the scope can first notice a warm object exists. It is not telling you the range at which you can tell what that object is, and it is certainly not telling you the range at which you can take a shot. For those, quietly divide — and let the animal come to you.
Detection tells you something exists.
Identification tells you whether you should take the shot.
Thermal Academy
Keep going: 384 vs 640 and Pixel Pitch explain the resolution that puts cycles on target; Lens Size and Field of View cover the reach-versus-width trade that aims them; NETD and Image Processing explain how cleanly you see the detail you’ve got; and How Far Can a Thermal Scope Really See? puts real ranges by animal to everything above — the practical, field-interpretation companion to the theoretical model on this page.
Test your knowledge · 6 questions
Check what you've learned
Open book — scroll back up any time. Each question appears as you go; the test marks itself once all six are answered. Score 100% to claim your Thermal Academy certificate.
Enter your details and we'll email your personalised Thermal Academy certificate for Johnson Criteria Explained as a PDF.
Frequently Asked Questions
Is a thermal scope’s detection range its shooting range?
No. Detection range is only where the scope can first tell something warm is present — you can’t identify it there. Your usable shooting range is bounded by identification range, often a quarter to a sixth of the detection figure, plus your rifle, hold and the conditions. Never shoot at a signature you’ve only detected.
What’s the difference between detection, recognition and identification?
Detection is knowing something warm is there; recognition is telling what class of thing it is (a pig versus a deer); identification is being sure of the exact target (this boar, a legal target, not a non-target animal). Each needs more detail than the last, so each happens at a shorter range.
How much shorter is identification range than detection range?
Roughly a quarter to a sixth of the detection figure as a rule of thumb, though it varies with sensor resolution, lens, conditions and how big the target is. A 2,600 m detection scope might identify a target somewhere around 700 to 900 m on a good night — and less in poor conditions.
What improves identification range the most?
Sensor resolution has the biggest effect, because more pixels put more cycles of detail on the target. A longer lens and a smaller pixel pitch help by concentrating those pixels, and a low NETD plus good processing keep the image clean. They all work together, but resolution is the main lever.
Why does this matter for hunting?
Because you should only shoot once you’ve positively identified the target and confirmed the range. Shooting at a detected-but-unidentified heat source risks hitting non-target or protected animals and taking unsafe shots. Detect at distance, identify up close, range, then decide.
What is the Johnson Criteria?
It’s a 1958 model developed by John Johnson at the US Army Night Vision Laboratory that predicts how much sensor detail — measured in “cycles across the target” — is needed to detect, recognise or identify a target. It’s the reason a thermal has three different range figures rather than one.
What is a “cycle” in the Johnson Criteria?
A cycle is one line pair — one dark line and one light line — laid across the target. It’s the unit of detail the model counts. Because it takes at least two pixels to show one cycle, the number of cycles on a target comes down to how many pixels the sensor lands across it at that range.
How many cycles are needed for detection, recognition and identification?
The classic Johnson figures are about 1 cycle across the target for detection, about 4 for recognition and about 6.4 for identification (with orientation around 1.4). Identification needs roughly six times the detail of detection, which is why it happens at a small fraction of the range.
Are manufacturers lying about detection range?
Usually not — the detection number is generally technically true under ideal conditions on the target size they’ve chosen. They simply quote the largest of three real ranges, because detection is always the biggest. It’s not a lie, but it is the least useful number for a shooter, who needs identification range instead.
Why is detection range so much bigger than identification range?
Because detection needs only about one cycle of detail on the target while identification needs about 6.4, and the detail a sensor puts on a target shrinks with distance. Needing roughly six times more detail means identification happens at roughly one-sixth of the range. It’s geometry, not marketing.
Does the Johnson Criteria guarantee I’ll identify a target at that range?
No. The cycle figures are pegged to a 50% probability of a trained observer succeeding under defined conditions — so a quoted identification range is where you’d succeed about half the time, not always. Treat it as an optimistic ceiling and do your real identifying much closer, where the odds are comfortably high.
What target size are DRI ranges based on?
Often a 1.8 m “man-sized” target or a 2.3 × 2.3 m NATO standard target. Those are larger than a fox or rabbit, so your real detection and identification ranges on small game are shorter than the sheet suggests. Always check the assumed target before comparing figures between brands.
Can I compare DRI figures between different thermal brands?
Only if they use the same target size and conditions, which they often don’t. A brand quoting a bigger detection range may simply be assuming a bigger target, not building a better sensor. Check the target and conditions each figure is based on before treating a comparison as fair.
Does weather affect detection and identification range?
Yes, a lot. Humidity, rain, fog and dust absorb and scatter infrared, and low thermal contrast (a target barely warmer than its background) reduces effective detail. All three ranges — detection, recognition and identification — come in on a muggy, low-contrast or dusty night compared with the fair-weather spec figures.
Does a higher-resolution sensor increase identification range?
Yes — it’s the most effective way to do it. A higher-resolution sensor lands more pixels, and therefore more cycles of detail, on the target at any range, so it reaches the 4 cycles for recognition and 6.4 for identification at a greater distance. Stepping from 384 to 640 noticeably extends both.
What is the orientation task in the Johnson Criteria?
Orientation is an intermediate task, needing about 1.4 cycles, where you can tell which way a target is facing or its rough aspect before you can classify it. It sits between detection and recognition and is often left off spec sheets, which usually quote only detection, recognition and identification.
How far can a thermal scope actually identify a pig?
It depends on the optic and the night, but as a rule of thumb identification lands around a quarter to a sixth of the quoted detection range, and a bit further for large game like a boar than for small game. Match it to your real distances and confirm identification up close before any shot — see our range-by-animal guide for worked numbers.
Is recognition enough to take a shot?
No. Recognition only tells you the class of animal — “that’s a pig” — not whether it’s a specific legal, safe target with a clear background. A responsible shot requires positive identification and a confirmed range. If you’ve only recognised the target, close the distance until you can identify it, or don’t shoot.
Why do two scopes with the same detection range identify at different distances?
Because identification depends on cycles on target and image clarity, not on the detection number. A scope with higher resolution, a longer lens or lower NETD can put cleaner, finer detail on the target, so it reaches identification further out even if its headline detection figure matches a rival’s.
Should I buy for detection range or identification range?
Think in identification range, because that’s the distance at which the optic actually helps you decide and shoot. Estimate it as roughly a quarter to a sixth of the quoted detection figure, check the target size assumed, and choose the optic whose identification range comfortably covers the distances you really hunt.
The Ultimate Guide to Thermal Scopes in Australia
