Antique Identifier App: What They Do and How to Choose One

What an antique identifier app actually does

It compares a photograph against a large body of images, catalog descriptions and sale listings, and returns the closest reading it can support. It does not examine your object.

That distinction is the whole story of the category. An appraiser picks a piece up. Weight, the ring a ceramic body gives when tapped, the temperature of glass against a hand, the feel of a foot rim, the join under a chair seat, whether a screw was cut by hand: none of that reaches a camera. What does reach a camera is form, proportion, decoration, color, surface and wear, and those carry a surprising amount of an identification, because they are also what a specialist uses in the first three seconds. An app is working from a real but partial signal, and the useful ones are built as though that were true.

How photo identification works, end to end

Five steps: capture, separation, description, retrieval, and a range derived from what was retrieved. The two that decide whether the answer is any use are the last two, and they are the two least often described.

1. Capture. You photograph the object or choose a picture already in your library. Everything downstream inherits the limits of this frame, which is why a distant or soft photograph costs more accuracy than a cluttered background does.
2. Separation. The object is cut away from its background so the model is reading the piece rather than the room. RelicLens does this on the device, which means the picture that leaves your phone is the object and not your kitchen table.
3. Description. A model returns what the object appears to be: a category, a likely period, an origin, materials, a condition read. This is the step people imagine is the whole app.
4. Retrieval. That description becomes a query against listings and sale records, looking for pieces that resemble yours. This is where tools separate, and where an app either has real data behind it or does not.
5. Range. A low and a high estimate come out of the comparables that were found. A figure produced without comparables behind it cannot be checked by anyone, including whoever produced it.

A step by step walk through of a single photograph, from framing to finished report, is on the identify antiques by photo page, and how we build a range out of retrieved comparables is written up on the methodology page.

What separates a useful result from a worthless one

You can check a useful result. That is the entire test, and it is worth applying in the first minute rather than the first month.

Comparable listings you can open. A number on its own cannot be argued with. A handful of listings for similar pieces, each with a price and the marketplace it came from, can: you open them, read what the seller declared about condition, and decide how close a match each one really is. Sometimes you will conclude the estimate is optimistic. That is the tool working, not failing.

A range rather than a point. Antiques do not have prices, they have ranges, and the range moves with condition, completeness, provenance and who happens to be in the room on the day. A single confident figure is a stylistic choice, not a more precise answer.

Named fields rather than adjectives. A period, an origin, a materials read, a category and a condition grade are five separate claims, and you can check them one at a time against the object in front of you. A paragraph of warm description is one claim you can only accept or reject whole.

Evidence rather than a percentage. A confidence score is generated by the same process that produced the answer, so there is no way to check it from outside. Comparable listings are the better signal, because you can open them and judge the match yourself. That is the approach RelicLens takes: every report shows the pieces the range was drawn from.

The questions to ask before trusting one with something you own

Six of them, and if an App Store listing cannot answer them, that is the answer.

1. Does it show its evidence? Look for comparable listings you can open in a browser, not a screenshot of a chart. If the only output is a number, there is nothing to check.
2. Does it claim to authenticate? If an app says it verifies authenticity, or tells genuine from fake, stop there. Authentication is a physical and documentary process. An app claiming it is either using the word loosely or does not know what it means, and neither is a good sign.
3. What happens to your photograph? Whether the image leaves the device, what it is used for afterward, whether it is attached to an account, and where the finished report is stored.
4. What does free actually mean? Free covers at least three different arrangements: a free download with every result behind a subscription, a fixed number of results at no cost, and a trial that converts to a subscription. Find the sentence that says which one you are looking at before you photograph anything.
5. Can you get the result out? Whether a report can be shared, exported or copied matters the first time you want to show it to a relative or hand it to an appraiser.
6. Who publishes it, and on what? A named publisher, a reachable support address, a stated platform and a minimum operating system version. An app with no reachable human behind it is not one to lean on when it matters.

Where these apps help, and where they do not

They help when you have many objects, little time and no baseline. They do not help anywhere a document is required or a dispute is possible, which covers most of the situations where money actually changes hands.

The useful cases have a shape in common: the alternative is knowing nothing at all. Standing at a table of forty objects with ten minutes before someone else buys them, a rough read on four of them is worth more than a perfect read on none. Holding a piece nobody in the family can date, a period and a material give you the words to search with, which is often the real blocker. The unhelpful cases have a shape in common too: somebody other than you has to accept the number.

Are antique identifier apps legitimate?

The software is real and it does a real, narrow thing. The question worth asking is not whether the category is a scam, but whether the specific claim in front of you is one a photograph can support.

The category has earned some of its suspicion. Listings promise appraisals, certificates and authentication, none of which a photograph produces. Some apps charge before showing anything at all. Some return a figure and no evidence, which is unfalsifiable by design. And a value estimate is the kind of output people want to believe, which makes it the easiest thing in the category to sell badly.

What a well built tool does give you, from one photograph, is the family of object you are holding, roughly when and where it was made, what it appears to be made of, and what comparable pieces are being asked for. When the alternative is a guess, that is the difference between walking away from something good and recognizing it.

The practical test takes a minute. Run one object, then open the comparables. If they are pieces you would accept as similar, the tool is doing its job. If they are a random sweep of the category, you have learned what that tool is worth before trusting it with anything that matters.

RelicLens, which we publish

This page is published by Eversight Studio LLC, which makes RelicLens, so here is RelicLens held to the same questions as everything above.

What a report returns. From one photograph: a category, one of eight (Silver & Metalwork, Ceramics, Glassware, Furniture, Jewelry, Sculpture, Decorative Object, Collectible); the likely period, origin and materials as free text; a condition grade on five levels from poor to excellent; a rarity read on five levels from common to exceptional; a reference value with a low and a high estimate either side of it; and comparable listings with prices and the marketplace they came from, which open in your browser. Every report is saved to a collection on the device that keeps a running estimated total.

What one free report means. One full report, once, held against the device rather than an account, so reinstalling the app does not reset it. RelicLens includes one full report free. A subscription is what lets you run more of them; it does not add fields to a report or unlock anything in your collection.

What it does not do. It does not read text, so hallmarks, maker marks, signatures and inscriptions are not transcribed. It gives no confidence score and no verdict on whether a piece is genuine. It offers no care, cleaning or restoration advice. There is no export, and no way to search, tag or filter a collection. Identification runs on a server, so it needs an internet connection. And it runs on iPhone and iPad, iOS 17.0 or later, with no Android, web or desktop version.

Who to reach. The publisher is Eversight Studio LLC and support is one address, hello@eversight.io, read by a person. The App Store shows 4.6 out of 5 from 56 ratings. The official app page collects the store listing, the publisher and the support address in one place, which is how to tell the real listing from something copying it.

Further reading on this site

The pages below go deeper on the mechanics, on what a value actually means, and on the two places people most often get hurt: buying reproductions, and inheriting something with no paperwork.

• Identify antiques by photo, which walks one photograph through to a finished report.
• Methodology, on where a value range comes from and what it is not.
• Antique marks identification, a reference for the marks no app can read for you.
• How AI is used in antique appraisal, on what the technology does and does not change.
• How much is my antique worth, five ways to estimate a value, only one of which is an app.
• How to spot fake antiques, the physical checks a photograph cannot make.
• Inherited antiques and what they are worth, where an app is a first step and not the last one.