Choosing a camera for artefact photography

No particular camera is needed to photograph an artefact well. What decides whether a picture is fit for study is measured in the picture itself: how many pixels fall on each millimetre of the object, how many bits the camera recorded, and how close the colour lands to a reference. This guide goes through what to look at when choosing a camera, and why the megapixel figure on the box answers only part of the question. The equipment the method recommends is listed under Technical requirements.

Megapixels or pixels per millimetre?

The resolution that matters is the resolution at the object: the number of pixels across the object divided by its length in millimetres. PHOTARCH's three levels — what a picture is fit for — ask for it in pixels per millimetre; imaging standards give the same thing in pixels per inch of the original:

PHOTARCH level	Pixels per mm at the object	ppi (px/mm × 25.4)	What it allows
Record shot	4	about 100	Asked only when the picture has a scale
Study	12	about 300	A print at life size at 300 dpi
Reproduction	24	about 600	A print at twice life size

For paintings, documents and other flat originals the app can also report FADGI, whose nearest category to an object, two-dimensional art, asks for 300 ppi at one and two stars, 450 at three and 600 at four.

A 24-megapixel camera has 6,000 × 4,000 pixels. How far that goes depends on the size of the object:

• A 50 mm object across 4,800 pixels of the frame: 4,800 ÷ 50 = 96 px/mm, about 2,440 ppi.
• A 300 mm object across 5,400 pixels: 5,400 ÷ 300 = 18 px/mm, about 457 ppi — enough for a study on resolution, not for a reproduction.
• A 600 mm object across 5,400 pixels: 5,400 ÷ 600 = 9 px/mm, about 229 ppi — below a study: a record shot.

For large objects — vessels, swords, grave slabs — the pixel count is what limits the picture. For small ones it rarely is.

Does sensor size matter?

For a small object, the limit is not the number of pixels but how close the lens can get. At its largest magnification m, a lens puts m millimetres of the sensor on each millimetre of the object, so the pixels per millimetre at the object are m divided by the pixel pitch — the distance between two pixels.

A sensor 36 mm wide with 6,000 pixels across has a pitch of 36 ÷ 6,000 = 0.006 mm. With a macro lens at life size (m = 1) that gives 1 ÷ 0.006 ≈ 167 px/mm. A sensor 23.5 mm wide with the same 6,000 pixels has a pitch of about 0.0039 mm, and gives about 255 px/mm with the same kind of lens. For an object smaller than the sensor divided by the magnification — smaller than 36 × 24 mm at life size on the larger sensor — the smaller sensor, at the same magnification, puts more pixels on it. The larger sensor only adds empty ground around it.

1. PHONE — 9.84 × 7.38 mm 8,064 px across
2. MICRO FOUR THIRDS — 17.3 × 13.0 mm 5,184 px across
3. APS-C — 23.5 × 15.6 mm 6,000–7,728 px across
4. 35 MM — 36 × 24 mm 6,000–9,984 px across
5. MEDIUM FORMAT — 43.8 × 32.9 mm 8,256–11,656 px across

A larger sensor has other virtues. At the same framing it collects more light, which tends to mean less noise, and it needs a longer lens, which gives less depth of field at the same aperture. Pixels are also not the same as detail: at high magnification, diffraction blurs what the pixels could record — see Choosing a lens.

Medium format or 35 mm: where the line falls

There is a breakpoint, and it can be worked out for each object. Two things set it.

When the object fills the frame, only the pixels across count. The pixels per millimetre are then 0.9 × the pixels across the frame ÷ the object's length, whatever the size of the sensor. A 100-megapixel medium-format camera has 11,648 pixels across, a 61-megapixel 35 mm camera 9,504: on any object large enough to fill both frames, the larger camera puts about 23 % more pixels. A 50-megapixel medium-format camera has 8,256 across, the same as a 45-megapixel 35 mm camera, and gains nothing.

When the object does not fill the frame, the lens decides. Macro lenses for 35 mm reach life size (1:1), some further. Fujifilm's and Hasselblad's own macro lenses for medium format stop at half life size (1:2). At half life size a sensor 43.8 mm wide cannot be filled by an object shorter than 0.9 × 43.8 ÷ 0.5 ≈ 79 mm, and the 100-megapixel camera puts at most 0.5 × 11,648 ÷ 43.8 ≈ 133 px/mm on it. The 61-megapixel 35 mm camera at 1:1 gives 0.9 × 9,504 ÷ the object's length, which is more than 133 for any object shorter than 0.9 × 9,504 ÷ 133 ≈ 64 mm.

With these lenses, then: below about 64 mm the high-resolution 35 mm camera puts more pixels on the object, above it the 100-megapixel medium-format camera does. Against a 24-megapixel 35 mm camera the line falls at about 41 mm. With a third-party 1:1 lens on the medium-format camera it moves down to about 32 mm.

How long an object can be. Each level asks for a number of pixels on every millimetre of the object. A camera has a fixed number of pixels across its frame, so the longer the object, the fewer fall on each millimetre of it.

1. ONE MILLIMETRE — of the object

The table gives the longest object each camera can photograph in one frame and still reach a level's resolution, with the object's longest side along the frame's long side and filling 90 % of its width: 0.9 × pixels across ÷ pixels per mm. A 61-megapixel camera puts 0.9 × 9,504 ≈ 8,554 pixels on the object; a reproduction asks for 24 per millimetre, and 8,554 ÷ 24 ≈ 356 mm. A sword 400 mm long gets 8,554 ÷ 400 ≈ 21 px/mm, about 543 ppi: enough for a study on resolution, not for a reproduction.

Camera	Pixels across	Record shot (4 px/mm)	Study (12 px/mm)	Reproduction (24 px/mm)
35 mm, 24 MP	6,000	1,350 mm	450 mm	225 mm
35 mm, 45 MP	8,256	1,858 mm	619 mm	310 mm
35 mm, 61 MP	9,504	2,138 mm	713 mm	356 mm
Medium format, 100 MP	11,648	2,621 mm	874 mm	437 mm

The figures are for one frame. Pixels are not the same as detail: focus, the lens and diffraction decide how much of this resolution the picture really holds. The PHOTARCH app measures how much of the object is in focus — at least 80 % for a study, 95 % for a reproduction, both provisional — not how fine the detail it resolves is.

The bits rarely stand in the way. A reproduction asks for at least 12 recorded bits, and a study for 10. Every 35 mm body in Pixels on the Object, 26 in all, records 14, so for them it is the size of the object that decides the level, not the bits. Four of its 51 cameras record 16 — all 100-megapixel medium-format bodies from Fujifilm and Hasselblad — and no level asks for that. Sixteen bits matter only for flat originals, where FADGI asks for a 16-bit file at three and four stars: with the standards turned on, bit depth is judged on the file, and a raw of 10 bits or more is written into one, with the bits the camera recorded shown beside it.

To work it out for your own camera, lens and object, use Pixels on the Object: it puts the object on the sensor of 51 cameras with 109 lenses, from the makers' published figures, and says which combination puts the most pixels on it.

Pixels on the ObjectPut your object on a camera's sensor and see how many pixels it gets, for 51 cameras and 109 lenses.→

Raw, and the bits the camera records

Photograph raw. A raw file holds what the sensor recorded before the camera turned it into a picture, and it holds more tonal steps than a JPEG's eight bits. How many depends on the camera: most record 12 or 14 bits, some medium-format cameras 16, and an iPhone's Apple ProRAW records 10. Some cameras record fewer bits in continuous or silent shooting.

PHOTARCH's levels go by the bits the camera recorded, not by the file they were written into: at least 10 for a study, 12 for a reproduction, so a 14-bit raw can reach either. With the standards turned on for flat originals, bit depth is judged as FADGI judges it, on the file. The figures, and how to get the most from them, are under Bit depth.

A phone or a dedicated camera?

A phone that photographs raw can make a picture that measures well. Its lenses have short focal lengths, so a small object needs the phone close to it; at that distance, parts of the object that stand higher than the surface it lies on are enlarged more than with a longer lens further away. An iPhone's ProRAW holds 10 bits, fewer than most dedicated cameras record: enough for a study, not for a reproduction.

A dedicated camera with interchangeable lenses can take a macro lens, stand further back, and record deeper raw files. Neither choice is wrong: the level each picture reaches says what it is fit for.

Tethering and remote release

Whatever the camera, avoid touching it between frames. A cable or wireless remote, or a few seconds' self-timer, keeps the stand still at the moment of exposure. Tethering — the camera connected to a computer, with the picture appearing on the screen as it is taken — makes it easier to check focus and exposure at full size and to name files as they come in. Several raw developers, and most camera makers' own programs, offer it.

The PHOTARCH app photographs through a paired iPhone and imports pictures from any other camera as files; a system camera tethered to the computer is not supported.