Everything, answered

No. Any USB microscope or webcam that Windows recognises will work, because NumismaLab uses the standard UVC camera interface rather than a manufacturer driver. If the camera appears in the Windows Camera app, NumismaLab can use it. A scope with adjustable focus and its own light ring will give you far better results than a fixed-focus one, but nothing is locked to a brand.

Windows 10 or Windows 11, 64-bit. 8 GB of memory is enough for everyday work and 16 GB is comfortable. A graphics card is optional: the 3D viewer, the depth map and the super-resolution upscale run faster on one, and everything else runs on the processor. The program has a built-in requirements check that tells you where your own machine stands.

Not at present. It is a native Windows application. It has been reported to run under Parallels and under Wine, but neither is supported and camera passthrough is the usual sticking point.

No. Examination, measurement, grading, the collection database and printing all work with no network at all. Only the optional assistant panels and the reference lookups need a connection.

Yes. Every analysis tool works on a saved photograph as well as on the live view. Open a folder or a coin from your collection and the tools read the selected picture.

It is a coin examination and collection program. It drives your microscope, renders the coin surface fifteen different ways so faults become visible, measures what you find on a calibrated scale, helps you place a grade, catalogues each coin with its photographs and value, prints 2x2 labels and flip cards, and includes a numismatic reference library. The microscope is the input; the program is everything that happens after the shutter.

Each one makes a different kind of fault obvious. Raking light swings a virtual lamp so you can see whether a line is raised or incuse by which way its shadow falls. Relief exaggerates height. Contour map draws lines of equal brightness, so doubling shows as a second contour running parallel. Crack scan, micro detail, edges, flow direction, heat map, depth map, negative, solarize, polarize and mono each isolate something else. They are ways of LOOKING, and the measurements are always taken from the real pixels rather than from a view.

Yes, and it runs two independent tests because doubling presents in two different ways. One looks for a separated ghost image standing off the primary, which is what lighter classes leave. The other looks for a second image running along the device outline, which is the signature of heavy hub doubling like the 1955 cent, where the two images have merged and no gap remains. The verdict goes to whichever test actually fires.

Only when the separated-ghost test is the one that found it, because that test was validated against known offsets. The device-outline test reports how much of the device carries the second image but deliberately prints no distance, because its offset figure was checked against a coin with an independently established separation and came back at roughly half the true value. A wrong measurement presented as a measurement is worse than no measurement, so it is withheld.

It carries listings for confirmed VAMs with their die markers described, so you can work through them against your coin. It does not automatically attribute a VAM for you. Attribution rests on marker positions that need a human eye, and a program that guessed would be wrong often enough to cost you money.

It will estimate a grade and show you exactly what that estimate rests on, which is the useful part. It scores wear, retained relief, lustre, contact marks, strike sharpness and hairlines, and prints each number beside the range the Sheldon scale expects for it, so you can see which measurements agree with the grade and which pull against it.

No, and you should not. A grade estimated from one photograph cannot see the coin tilt in your hand, which is how grading is actually done, and it cannot reliably tell a cleaned coin from an original one. Treat it as a second opinion that shows its working, and send anything valuable to a grading service.

Because the diagnostics do not always agree, and that disagreement is information. If wear says XF-40 while lustre says Fine-12, something is unusual about the coin or about the lighting, and knowing which number is the odd one out tells you where to look.

Lustre is measured as the fraction of the surface throwing a bright specular highlight, and that depends heavily on your lamp. A flat, diffuse light shows almost no lustre on a coin that has plenty. Trust the figure when you are comparing two coins shot the same way, and distrust it across different lighting.

Calibrate once, three possible ways: against a coin whose diameter you know, by dragging along a steel rule or stage micrometer, or by storing a scale for each magnification setting of your scope. After that the on-screen caliper reads in millimetres and microns, and diameter can be checked against the specification database.

The measurement is as good as the calibration and the focus. A stage micrometer will get you within a few microns; a catalogue nominal diameter will not, because the coin in your hand is not exactly nominal. The program reports what it measured, not what it hopes.

Relative relief, yes, from a focus series or a photometric light set where the lamp genuinely moved between shots. Those are real measurements. The single-photograph depth estimate is a neural network guess and is labelled as one, because it will produce a plausible landscape from almost anything, including a flat coin.

In ordinary folders on your own drive, inside a collection root you choose. Each coin gets its own folder, named so a set reads as a set in Explorer with no software at all, and the camera settings that took each shot are recorded alongside it.

They go into a Pending folder, one folder per coin, and nothing is filed or given a SKU until you say what the coin is. If you close the program part way through a coin, it comes back to that same coin next time you open it.

Photographs, year, denomination, series, variety, grade, cost, current value, provenance, notes, the physical storage location where the coin actually sits, and a SKU that never changes.

Photographs are written to a temporary name, flushed to the disk, and only then renamed into place. A crash mid-write leaves a partial file with a temporary name that is cleaned up later; it cannot leave you with a half-written photograph under a real name. A crash also never deletes pending work.

Yes. The photographs are standard image files and the records export as text. Nothing is held in a format only this program can read.

Yes, to a thermal label printer over USB or Bluetooth, and to any ordinary printer. The label carries the coin details, your own name or business name, the storage location, and the SKU as a scannable Code 128 barcode.

If Windows has it installed as a printer, yes. NumismaLab prints through Windows rather than talking to the hardware directly, so USB and Bluetooth are the same problem as far as it is concerned, and you pick the printer from a list.

Custom 2x2 inserts you design yourself, with your own graphics, any font on your machine, and whatever content you like on each side, for instance the coin history on one face and your contact details on the other. A whole page prints front and back with the two sides aligned, and a design can be saved against an individual coin.

It will draft listing copy from the coin record and the measurements you have taken. You edit it before it goes anywhere.

A reference library built into the program: lessons, a grading reference, several hundred coins worth knowing, confirmed VAM listings with markers described, and a hardware guide. It is reading rather than an instrument, so it opens in a window of its own.

No. Nothing is reproduced from PCGS, NGC or the variety sites, in image or in text. The material is written for this program.

Three ways to pay: $39.99 a month, $179 a year, or $299 once for a lifetime licence. Every tier includes every tool - the only difference is how long the licence lasts. Nothing renews by itself; when a term ends you decide whether to buy another.

The licence covers the machine on your bench. If you change computers you can release the licence yourself and activate it on the new one, without waiting on support.

No. The software is unlocked with a registration key, not a login. You may want a website account for support history and order lookup, but that is separate and a website outage never affects whether your software runs.

Nothing, for a long while. The licence is cached locally and the program keeps working offline. Past the grace period the gated tools stop, but capture, the photograph strip, the collection and export keep working, because your photographs are irreplaceable and holding them hostage over a network fault would be indefensible.

Yes, five days. It needs a free account on this site — that is how the trial is issued and how it stays five days rather than restarting on a reinstall — but it needs no card, and nothing is charged when it ends. It exists so you can check the program works with your own microscope before spending anything.

It is not tracking you - there is no telemetry and no analytics of any kind - but a licensed copy does contact the server, so here is exactly what for. It checks that the licence is still valid, sending your licence key and a fingerprint derived from your hardware. That is the whole of it. Your photographs, your collection, your measurements and the coins you own stay on your computer. It checks rarely rather than constantly, too: sign in once and it runs fully offline for a week at a time. The only other thing that ever leaves your machine is a bug report or a feature request, and that carries only what you typed into it.

No. Captures come out clean. Screen recordings of the program itself carry a mark, which is a different thing from a photograph of your coin.

Only if you deliberately use an assistant panel, which is a signed-in web view you drive yourself. The measuring tools are all local. The two features that use a model on your own machine, the depth map and the super-resolution upscale, are listed inside the program so the word AI never appears somewhere you were not told about.

Close anything else using it, since Windows gives a camera to one program at a time, then check the camera privacy setting under Windows Settings. The program also has a requirements check that reports what it can see.

It needs either a live camera or a photograph on the bench. Open a coin from your collection, or pick a photograph in the strip under the live view, then run the tool again.

A blurred photograph depresses the relief measurement and drags the estimate down with it, which is why the program warns you when strike sharpness falls below a threshold. Refocus and run it again before believing the grade.

Frame a single device tightly, a letter or a digit rather than the whole coin, and make sure the lettering fills a good part of the frame. The detection is blind below about six pixels of separation, so magnification matters. A crop dominated by the rim can also mislead it, because concentric arcs at constant spacing look locally like a second image.

Three things, in order of how much they matter. Light it from the side rather than from the front, because a shadow is what makes relief visible and a coin lit head-on looks flat. Get the coin square to the lens, or one side of it will be out of focus however carefully you focus the other. Then use the focus peaking and the focus meter: peaking outlines what is actually in the plane of focus, and the meter tells you when you have found the sharpest point of the knob's travel rather than a point that merely looks sharp on a small preview.

Use the quadrant sweep. It walks the coin in a grid so every area is covered, naming the frames so the set stays in order. For a general view, back the scope off to its lowest magnification first - a microscope at full zoom is for a device, not for a coin.

Focus stacking. Turn the focus knob slowly from one extreme of the coin's relief to the other while it records, and it keeps whichever frame was sharpest at each individual pixel. What comes out is one photograph sharp across the rim and the devices together, which no single exposure at that magnification can be.

The detail stack. It aligns each incoming frame to a reference and keeps a running average, so random sensor noise cancels out while real detail reinforces. Hold everything still and give it a few seconds.

Open the phone camera panel and scan the QR code it shows. The phone streams its camera to the program over your own Wi-Fi, with nothing to install on the phone. Both devices have to be on the same network, and your phone will warn you about the security certificate the first time - that is expected, because the certificate is generated on your own machine.

Connect it by USB and open the camera panel. If you would rather shoot the way you already do, point the folder watcher at wherever your camera software saves its files and every photograph that appears is filed into the open coin automatically, raw files included.

Watch the live histogram before you press the shutter. If it is piled against the right-hand edge the lustre has been clipped, and no amount of editing afterwards will bring it back, because the camera recorded nothing there. Lower the light or the exposure until the brightest part sits just short of the edge.

Photograph the doubled area at as much magnification as you can hold steady, then run the doubling tools. Two independent tests run and the verdict goes to whichever one fires, which matters because light doubling and heavy hub doubling look nothing alike. Then run the relief test, because that is what separates a doubled die from machine doubling: a struck secondary image stands at full height, and machine doubling is a flat shelf.

By height, not by shape. Doubling that was in the die was struck into the coin, so it has rounded relief of its own. Machine doubling happened as the die bounced coming out, so it is a flat shelf shaved off the edge of the device. The program measures the profile across the suspect edge and shows you the profile it read, so you can check the reasoning rather than take the verdict on trust. It will refuse to answer if the lighting is too flat to establish relief at all, which is a real answer and not a failure.

Whether it is raised or incuse, which the line tools report for every line they trace. A crack in the die leaves a RAISED line on the coin, because metal flowed into the crack. A scratch was cut into the coin afterwards, so it is incuse. Raking light shows the same thing by eye: swing the virtual lamp and a raised line throws its shadow the opposite way to an incuse one.

Run the cleaning scan. A wiped or polished coin carries fine scratches running in roughly one direction, because a hand moves in one direction. Be careful with the result in both directions: die polish is also directional and is not damage at all, and a skilfully cleaned coin can pass. It is a reason to look closely, not a verdict.

The tools do that for you now. On a well-framed shot the coin is about half the frame and the rest is mat, and a mat under a microscope lamp is often the strongest texture in the picture - which wrecks any measurement that sets its thresholds from the whole frame. The analysis tools isolate the coin first and say in the status bar that they have. If no round coin is found, the frame is used as it is, because a slab or a close-up of one device is what you meant to measure.

Alignment Lab, then Precision Compare. Alignment Lab lines the photographs up to sub-pixel accuracy inside a region you choose; Precision Compare then removes the things that masquerade as differences - lighting change, sub-pixel misregistration, sensor noise - before reporting anything. Comparing two raw photographs mostly measures your lamp.

Open it from the strip under the picture, or pick a coin out of the collection and put it back on the bench. Every tool works on a photograph exactly as it does on the live feed. If one photograph is selected in the strip that one wins; otherwise whatever is open in the reviewer is used; otherwise the live view.

Photograph it first and identify it afterwards. Shots taken before you have said what the coin is wait in a Pending folder, one folder per coin, and nothing is filed or given a SKU until you say what it is. That order is deliberate: at a show or on a roll search you photograph far more than you will keep.

It is in the collection tab, searchable by anything recorded about it - date, denomination, series, variety, grade, where it is stored, or its SKU. The program also reopens on the coin you last worked on, so a session interrupted halfway through picks up where it stopped.

A hundred. It used to be twenty, which was nowhere near enough - a zoom survey and a quadrant sweep together run to dozens of frames before anybody has photographed a variety marker.

Five ways, because they answer five different questions. CSV goes into somebody else's program. HTML goes to a family member or an insurer who will not install anything. Excel is what most collectors actually keep a ledger in. PDF is for printing. Or take a copy of the database itself. Nothing is locked in.

Use the database backup, and know that your collection, photographs and reports already live OUTSIDE the program folder by design - so uninstalling, reinstalling or letting a licence lapse never touches them. Keep one copy somewhere other than the same disk; that part is still yours to do.

The flip label prints to a thermal printer over USB or Bluetooth, or to any ordinary printer. It carries the coin details, your name or business name, the storage location, and the SKU as a scannable barcode so the shelf and the software always agree. You can have one print automatically at the moment you file the coin.

On the measurement tools. It only reads in millimetres once the scope is calibrated - before that every distance is in pixels, and a pixel is a property of your camera rather than of the coin.

In the measurement section, three ways: against a coin whose diameter you know, by dragging along a steel rule or a stage micrometer, or by storing a scale for each magnification setting of your scope so it is ready next time. A stage micrometer is worth the few pounds it costs if you intend to quote measurements to anybody.

In a folder per coin, outside the program directory, named so a set reads as a set even in Windows Explorer. File then Open captures folder takes you straight there.

In NumismaticsU, the learning window. Pick the date, pick the VAM, and the right-hand pane lists the diagnostic markers one at a time with where on the coin to look for each.

Under the three-dimensional tools. It needs either a focus series - the same view shot five to fifteen times while you turn the focus knob - or a photometric light set, where you move the lamp between shots. Those are measurements. The single-photograph depth estimate is in the experimental tools instead, because it is a guess and is labelled as one.

The Controls page has every slider, and there are sixty-nine of them. If that is not what you want, the preset adjustments are a short list of named things to do to a picture, each a toggle and a slider - the same work without needing to know what CLAHE tone mapping is.

Help, then About. Quote it in any bug report - it is the single most useful line in one.

Sign in to the website and open My Account, then Licences. It lists your key, when it expires and which machines are using it.

Each view carries its own legend saying exactly that, and hovering over a colour tells you what it represents in that particular view. They are not consistent between views on purpose: a heat map and a depth map are answering different questions and sharing a colour scale would imply they are comparable.

Each measurement is printed beside the range the Sheldon scale expects for it, so you can see which diagnostics agree with the estimate and which pull against it. The disagreement is the useful part - if wear says XF while lustre says Fine, something is odd about the coin or about the lighting, and that is worth knowing before you send it anywhere.

How unusual this coin looks, from 0 to 1, with the reasons behind it. It deliberately does not name the error - naming is your job. It is a reason to stop and look, never a finding, and a low score is not a clean bill of health.

That a round coin was found and the mat has been excluded before measuring. If it says the frame was used untouched, no round coin was found - a slab, a coin cut off by the frame, or a close-up of one device - and in each of those cases what you framed is what you wanted measured anyway.

Because it could not honestly reach one, and saying so is the design. A tool that always produces a number teaches you to trust numbers that are sometimes invented. The repeat-edge screen reporting nothing is it telling you it found no rhythm, not it failing.

Because the figure would be wrong. Where the outline test is the one that found the doubling, the separation was checked against a device whose true offset could be established three independent ways and came back at about half the real value - so it is withheld rather than printed.

Open a support ticket from My Account. It goes to a real person - to Jen - and you get an answer, not a tracking number. You can also email support@numismalab.com.

Because no customer can, and that is deliberate. An account here exists so you can manage your licences, your billing details and your support tickets, and all of that lives in My Account. The dashboard would expose things about the site that are nobody's business but ours, and closing that door is the simplest way to keep your details safe.

See and edit your billing and account details, read your licences and which machines are using them, release a machine you no longer have, look at your orders, raise and read support tickets, and request a feature. Beta testers also get the bug report form there.

Release the old machine from My Account, then sign in on the new one. If the old machine is gone or dead you can still release it from the website, because the release happens here rather than on the PC you no longer have. If you are stuck, ask - it takes a moment to sort out.

In My Account, under account details and payment methods respectively.

No, and it never will. Your photographs, database and reports live outside the program folder and are yours. When a licence lapses the program degrades rather than locks: capture, the filmstrip, your collection and export all keep working, and only the analysis tools stop. Coin photography is often unrepeatable once a coin is sold or slabbed, and holding that hostage would be indefensible.

By invitation. Each invitation is a single-use link that creates one account, and you download from that account and sign in with it from inside the program.

Thirty days, and then it ends permanently. That is not a nag screen you can dismiss - a beta build is expected to be replaced, not lived in.

No. Thirty days with every tool unlocked, and then a further five, would be a thirty-five day free run, and the beta was meant to end. A licence works immediately and picks up exactly where you left off - and if you tested and then bought, nothing about having helped counts against you.

Nothing at all. Your collection, database, photographs and reports live outside the program folder and are untouched.

Beta testers get a bug report form in My Account. Fill in as much of it as you can: the version, what you did step by step, and what you expected instead are the three things that decide whether a bug can be reproduced, and a bug that can be reproduced is usually a bug that gets fixed.

Raise a support ticket. You get a person and an answer, which is a better deal than a defect report - and it reaches the same place.

Yes, and please do - every customer and every tester can, from My Account. Most of what is in the program now started as somebody saying it should be.

No. The people most likely to notice what is missing are often the ones still deciding whether to buy.

The problem behind it. Describe what you are trying to do and why it is awkward now, not only the solution you had in mind - the underlying problem often has a better answer than either of us first thought of.

Every request is read, and its standing shows in My Account: received, triaged, planned, or not planned. "Not planned" is used honestly rather than left hanging, because a request that sits untouched for a year is worse manners than a straight answer.

Check it appears in the Windows Camera app first - if Windows cannot see it, neither can NumismaLab. Close anything else using the camera, because Windows generally gives a webcam to one program at a time. Then try a different USB port, preferring one directly on the machine over a hub.

That is probably the hardware, not the program. A USB microscope is a webcam, and most of them expose no real exposure or gain control at all - so what you are adjusting is the picture after the fact. The program says which controls are real for your device rather than pretending they all are. A DSLR or mirrorless body gives you genuine control.

Open Settings and check the AI tabs option. The panels normally run inside the program; there is an option to run them in a real Edge or Chrome window instead, which depends on Windows docking that window into the panel, and when that fails the tab looks empty and the browser window is left behind the program. Turning it off puts the panels back inside.

That is usually the embedded web engine failing on that machine. Contact support and it can be started with the assistant panels left out, which gets you running while the cause is found.

The program noticed the system clock moving backwards relative to when your sign-in was issued. Set the clock correctly - ideally to sync automatically - and sign in again.

Signing in once buys seven days of working with no network at all. Past that it needs to check in. Connect briefly and it will refresh in the background.

Check the calibration first - an uncalibrated scope reports pixels, and a nominal catalogue diameter is not the diameter of the coin in your hand. Then check the zoom: analysis starts from the whole frame, and if a tool has been handed a digitally zoomed picture it is measuring part of the coin and calling it the coin. Please report it either way.