Photographs do not translate to relief printing by being traced. A photograph is continuous tone; a relief print is a small number of flat areas of solid ink. The work is deciding which of those millions of tones collapse together into each flat area — and that decision, not the carving, is what determines whether the finished print reads.
Start by choosing how many inks
Everything follows from this number, so make it a deliberate choice rather than something you discover halfway through.
- Two tones (ink and bare paper) is the classic single-block linocut. Unforgiving, and the most graphic. Everything depends on where the single threshold falls.
- Three or four is where most images start to hold together as pictures rather than as patterns. It is also about as many passes as a reduction print stays enjoyable for.
- Five or six gives real tonal range, at the cost of five or six registrations and five or six ink mixes that all have to be right.
Portraits and anything with soft modelling usually need more tones than you expect. Architecture, silhouettes and strong-light subjects often need fewer.
Where the thresholds go
Having chosen the number, something has to decide which tones fall on which side of each boundary. Doing it by eye is entirely legitimate and often better, but there is a well-founded automatic method worth knowing about.
Otsu's method, published in 1979, picks the threshold that best separates a greyscale image into two groups — it looks at the spread of tones and finds the cut that makes the two resulting groups as internally consistent as possible. It is the standard answer to "where should the line go" and it is what a good automatic conversion is doing under the bonnet.
For more than two tones, the equivalent job is clustering: grouping the image's colours into however many groups you asked for, and giving each group one flat colour. Done in a colour space that matches how the eye judges difference, this produces separations that look right rather than merely arithmetically even.
Neither method knows anything about your subject. They will happily put the threshold straight through the middle of a face. Treat the automatic answer as a strong starting point and then move it.
Use the paper as one of your colours
An easy gain. In relief printing one of your tones costs nothing to print, because it is simply the sheet showing through. On cream or buff paper that unprinted tone is not white — it is a warm mid-light, and it will sit differently against your inks than a paper white would.
Planning that in deliberately means a four-tone print needs only three passes.
The thing that will actually ruin the block: slivers
When a photograph is reduced to flat tones, the boundaries between those tones are ragged. They produce, all over the image, thin isolated fingers of material — sometimes a fraction of a millimetre wide — that look perfectly fine on screen.
Those will not survive. They crumble under the gouge, or they hold together through carving and then break off in the press, or they take ink so unevenly that they print as a smear. The time to find them is before you start cutting, not at the point when you have spent an evening carving around them.
Finding them is straightforward in principle: shrink every shape in the image slightly, then grow it back, and anything that disappears in between was too thin to survive. What is not straightforward is knowing where to set "too thin".
I could not find a published figure for the minimum width of lino that will reliably survive carving and printing. Not in the trade literature, not in the printmaking manuals.
The thresholds in my own tool are therefore my invention. They may well be sensible. They are not evidence-based, and they almost certainly should differ by material — grey battleship lino, Japanese vinyl, Softcut and Speedy Carve are meaningfully different in hardness, and the softer materials hold a thin even width considerably less well.
If you have a working figure for this, from your own bench, I would genuinely like to hear it. It is the single most useful thing anyone could tell me about these tools.
Given that uncertainty, the sensible bias is to over-warn. Being told about a sliver that would have been fine costs you a glance. Not being told about one that was not costs you the block.
What to do with the slivers you find
Three options, in roughly the order most printmakers reach for them:
- Absorb them. Let the thin finger merge into the neighbouring tone. Usually invisible in the print.
- Thicken them. If the detail matters, redraw it wider than the photograph made it. A print is not a photograph and does not have to pretend to be.
- Move the threshold. Often a sliver exists only because a boundary happened to fall through a gradual gradient. Shifting the threshold a little makes it vanish and changes nothing else you care about.
Block Plate takes a photograph and reduces it to a set number of printable tones, lets you hold one of them as bare paper, warns about slivers that will not survive carving, and gives you a wipe control so you can compare the result against the original.
It shows you what a picture reduces to. Separating that into blocks and working out the ink for each is Reduction Ladder's job.
Where this comes from
- Otsu, N. (1979), "A threshold selection method from gray-level histograms", IEEE Transactions on Systems, Man, and Cybernetics 9(1):62–66 — the thresholding method.
- Arthur, D. & Vassilvitskii, S. (2007), "k-means++: the advantages of careful seeding", SODA '07 — the clustering initialisation used for multi-tone separation.
- Jackson's Art, "Comparing lino, Softcut, Speedy Carve and Japanese vinyl" — the material differences that bear on how fine a sliver survives.
- On minimum sliver width: no source. See the note above.