How we compare
Where the numbers come from
Every published specification on this site was read off a manufacturer’s product page or a major retailer’s listing and recorded against that source; each unit’s page names the sources its figures were checked against. Where two sources disagreed the manufacturer’s figure won. Where a maker publishes nothing — NETD is the usual gap; 16 of 67 units here do not publish it — the field is blank rather than estimated. All 67 entries across 13 brands were recorded on 2026-09-20.
There is no test lab and no staff. Nobody here has shot every unit in the catalog. What the site adds is arithmetic applied consistently to every maker’s own numbers.
How the range estimates are computed
Two published numbers go in: pixel pitch (µm) and lens focal length (mm). Divide them and you have the angle one pixel sees, in milliradians — the IFOV. The Johnson criteria then say how many pixels across a target’s critical dimension an observer needs to perform each task, at 50 % probability: about 2 to detect that something is there, 8 to recognize what kind of thing it is, 13 to identify which one. For a standing human we use the 0.5 m width as the critical dimension, so range in metres for N pixels is 0.5 ÷ (N × IFOV) × 1000.
That is the entire model. It assumes clear air, a typical observer, and a target with normal thermal contrast against its background. It does not include the display, the eyepiece, the NETD, or the weather, all of which can only make things worse. The same function in one file produces every estimate on the site, so cards, tables, guides and descriptions cannot disagree with one another.
Why our numbers are lower than the box
Manufacturers almost always compute their detection range against the 1.8 m height of a standing person, at the point where that height spans about two pixels. That is a legitimate statement of the furthest distance at which a warm vertical object registers, and it is roughly 3.6 times further than a two-pixel criterion on the 0.5 m width. Across the 60 units here that publish a claim, the box figure averages 2.3× our detection estimate, ranging from 1.4× to 5.1×. 7 units publish no claim at all.
We print both, side by side, on every page. Neither is a lie; they answer different questions. Ours answers the one a hunter is actually asking, which is why the guides sort on the estimate and never on the claim. If you want to know which figure to trust before a shot, it is the recognition estimate, and even that is an upper bound.
Why there are no star ratings
Because there are none we can stand behind. An aggregate rating is a claim about what real owners think, and inventing one — or quietly reprinting a retailer’s — is the most common dishonest thing a comparison site does. The code gates every star on a cited, non-Amazon source; nothing currently clears that gate, so no stars render anywhere. For the same reason there are no owner reviews: we have not read owners of these units, and “owners report…” without owners is fiction.
Prices
Every price on the site carries a ~ because it is a snapshot — the manufacturer’s list price or a major retailer’s listed price on the day the catalog was recorded. Thermal prices move on closeouts and new generations constantly. The band is more durable than the number; treat the number as an order of magnitude.
Why some units have no buy link
Roughly one in ten plausible-looking Amazon product IDs resolves to the wrong thing — a 384 where the page says 640, a mount, a battery pack. A link is only published once what it actually opens has been confirmed against the unit name. 66 of 67 units currently have a confirmed manufacturer link; the rest show no link, which is the correct outcome: sending you to the wrong optic is worse than sending you nowhere.
How the lists are ordered
Not by opinion. Every list on the site says the figure it is sorted by — estimated detection or recognition range, weight, price, battery life, sensor pixels. Guides are filters over the same catalog, so a guide cannot contain a unit we have not recorded, and any guide that would qualify fewer than three units is not published at all. Hub pages need two.
What we never do
- Take money for placement. Nobody can buy a position in a list here.
- Write owner reviews for products whose owners we have not read.
- Print a rating, a review count, or a specification we cannot source.
- Present a derived number as a published one. Estimates are labelled estimates, everywhere.
- Publish a comparison page for a pairing nobody would actually cross-shop.
How this site makes money
Affiliate commission on retailer links, when there are verified links to publish. As an Amazon Associate, thermalranker earns from qualifying purchases. Commissions never affect what appears on a list or in what order — the orderings are computed from specifications, so there is no lever to pull even if we wanted one. Full details in the privacy policy.