Oil Painting or Print? How to Tell, in Seven Tests

The seven tests, in the order that works

Order matters, and it is the thing most guides leave out. The standard advice — dots, texture, an exact match to a famous image, odd fading — is a list with no weighting, so a reader who gets two contradictory answers has nowhere to go. Two of the tests below are decisive, three are strongly indicative, and two only matter once you already suspect an answer. Running them in sequence takes about five minutes and needs one piece of equipment: a 10x jeweller's loupe, or a phone camera with a clip-on macro lens.

1. The loupe — decisive

Put the loupe on a mid-tone area, not on a highlight and not on a dark shadow. You are looking for dots. Any dots, of any kind, anywhere, mean a print. Paint does not produce them.

There are two families to recognise. Offset lithography, the process behind older poster reproductions, prints a regular rosette of cyan, magenta and yellow dots on a fixed grid — the Image Permanence Institute's process reference describes it as a fine, soft-edged regular rosette dot pattern under low magnification (Graphics Atlas). Inkjet, which covers everything sold as giclée, sprays fine droplets with no grid at all: at 10x it reads as a mist, and the edges of dark shapes look sprayed rather than cut.

One honest caveat, which most guides skip: a high-resolution giclée on a coarse canvas weave can hide its droplet structure at 10x, because the weave breaks up the pattern before your eye reaches it. If the loupe comes back inconclusive rather than clean, do not read that as "painting" — go to raking light and the edges, and let those decide.

2. Raking light — decisive

Turn off the room lights, take a torch or a phone light, and hold it almost flat against the surface — ten to fifteen degrees, shining across the painting rather than at it. Then look along the canvas from the opposite side.

A hand-painted oil erupts. Ridges throw shadows, brush hairs leave tracks, and the relief follows the image: thick in the highlights where the painter loaded the brush, thin in the shadows where they scrubbed it out. A flat print stays flat and shows only the woven texture of the canvas itself, uniformly, everywhere. This is the test that survives every attempt to fake it except one, which is the next section.

3. The four edges — strongly indicative

Look at the sides of a stretched canvas, along the fold at the corner, and at the outer few millimetres of the face. A painted canvas has paint that runs raggedly over the edge, corners where the paint cracks slightly over the fold, and often bare primed canvas underneath. A print has an image that stops with mechanical precision at the fold, or a mirrored or stretched-out version of the image wrapped round the side, or a plain white or black band. Printers cannot make a ragged edge and do not try.

4. The back — strongly indicative

Turn it over. On a painted canvas, oil and medium migrate through the weave over years, leaving faint blooms of discolouration that match the shape of the heaviest paint on the front. On a print you see clean, even, untouched canvas — plus, frequently, a printed batch code, a size sticker, a barcode or a QR label, none of which belong on a painting. Staples in a perfectly even machine rhythm point the same way.

There is a free extension of this test on an unlined canvas, and it is the one useful check the general guides do carry — GB Tate & Sons frames it as holding the canvas up to the light. On a print the whole image comes through evenly, because the ink sits in a single thin film. On a painting the light is blocked unevenly — thick passages go dark, thin scrubbed shadows glow — and that unevenness maps onto the composition rather than onto the weave.

5. The signature — strongly indicative

Under the loupe and under raking light, a painted signature sits in the same paint layer as everything around it: same relief, same brush drag, sometimes the same wet-into-wet bleed at the edges. A printed signature is part of the printed image and is therefore perfectly flat, dot-patterned like everything else, and lies underneath the varnish rather than in the paint. On a reproduction the master's signature is normally copied along with the rest of the image, so its presence tells you nothing about who painted it — only about whether the surface is paint at all.

6. Craquelure — only useful on old pictures

Genuine age cracking is three-dimensional. It runs through the paint film, changes direction where the paint thickness changes, follows the stress lines of the stretcher, and you can feel the edges with a fingernail. Printed craquelure is part of the image: it is flat, it continues unchanged across thick and thin passages, and it repeats where the reproduction has been tiled. On a canvas that is obviously modern, this test tells you nothing at all.

7. Sheen — the tiebreaker

Stand to one side and let a window reflect off the surface. An oil painting's gloss is uneven, because different pigments and different amounts of medium dry to different sheens, and because varnish sits on top of relief. A print's sheen is uniform across the whole sheet, or uniform within the areas the machine coated. Use this one only to break a tie — a heavy modern varnish can flatten it out on a genuine painting.

What a textured print does to fool you, and where it fails

The reason so many people get this wrong is that the market moved. Twenty years ago the question was paint versus poster and raking light settled it in one second. Today a large share of what sells as "art on canvas" is a print with a clear gel or resin medium applied over it, deliberately to defeat the touch-and-light test.

It defeats it only halfway, and the failure is always the same. The relief is applied by a machine that does not know what the picture contains. So the ridges are generic: an identical brush-like texture runs across the sky, the water and the sitter's face, at a constant pitch and a constant height. On a real painting the relief is the image — a Van Gogh sky is built in directional ropes of paint and the shadow beneath a chair is a thin scrub with almost no body at all. Once you have seen the difference deliberately, you cannot unsee it.

• Look where the painting should be thin. Deep shadows and glazed passages carry almost no paint on a real canvas. A textured print puts the same ridge height there as everywhere else.
• Look at a boundary. Where a bright impasto highlight meets a dark background, real paint has a physical ridge that stops at the colour boundary. Gel texture crosses straight over it.
• Put the loupe in a valley. The dots are still there between the ridges, because the coating is on top of a print. This single check ends the argument.
• Check the wrap. Gel coating is usually applied to the face only, so the sides feel and look different from the front.

None of this makes a textured print a bad object. It makes it a print, priced and cared for as a print. The full comparison of what each one costs to own over a couple of decades is in hand-painted vs canvas print, and the print side of it in detail in giclée vs oil painting.

The tests that do not work

• Smell. Oil paint is odourless within months. A canvas that smells of solvent is either very recently painted or very recently varnished, and plenty of prints are coated with something that smells too.
• Weight. A stretched canvas weighs what its stretcher weighs. Paint adds grams, not kilos.
• Rubbing it with a damp cloth or a solvent. This ruins prints, removes varnish and dissolves glazes on genuine paintings, and tells you nothing a loupe cannot tell you for free. Conservators do solvent testing under magnification on tiny inconspicuous areas for entirely different reasons.
• "It looks too perfect to be painted." Skilled copyists are extremely accurate, and cheap prints are often visibly soft. Perceived quality is not evidence in either direction.
• Price paid. Textured prints are routinely sold at hand-painted prices, and hand-painted canvases turn up in house clearances for the price of the frame.

What to do once you know

If it is a print, you have a decorative object. Keep it out of direct sun, because the inks fade and they do not fade evenly. There is no appraisal to seek.

If it is hand-painted, the next question is age, not authorship — and the answer is on the back, not the front. Machine staples, milled stretcher bars and bright cotton mean a recent reproduction; iron tacks, hand-cut joints, oxidised linen and old paper labels mean it is worth a free photo valuation from a regional saleroom. The full checklist and the realistic price outcomes are in are oil painting reproductions worth anything.

And if you are buying rather than identifying, the tests reverse into a specification. Ask for a raking-light photograph and a photograph of the side edge before you pay. Both are trivial for a studio that is painting your canvas and impossible for anyone shipping stock from a warehouse, so the request costs you nothing and tells you a great deal. Some form of pre-shipment photograph is standard at the better made-to-order sellers — Art & See sends a photograph of the finished piece for approval before it ships, which is the moment at which every test in this article can still be run by the seller instead of by you.

Frequently asked questions

Sources

1. 1.Graphics Atlas: identification of printing and photographic processes under magnification — Image Permanence Institute, Rochester Institute of Technology. Accessed September 5, 2026.
2. 2.Graphics Atlas: identification tutorial — Image Permanence Institute, Rochester Institute of Technology. Accessed September 5, 2026.
3. 3.Photomechanical prints — collection identification guide — Preservation Self-Assessment Program, University of Illinois. Accessed September 5, 2026.
4. 4.Is my painting real? How to tell if you have an original or a print — Fine Art Restoration Company. Accessed September 5, 2026.
5. 5.Detecting fakes and forgeries — help page — GB Tate & Sons Fine Art. Accessed September 5, 2026.
6. 6.Compendium of U.S. Copyright Office Practices, Third Edition, §313.4(A) Mere Copies — U.S. Copyright Office. Accessed September 5, 2026.