Alexander Grabovetskiy, woodcarver, School of Woodcarving · Published 2022
How do you carve one shape passing under another?
Quick answer: An intersection in relief carving is two cuts working together. The upper shape keeps a wall standing dead upright, and the lower one drops roughly three millimetres into shadow before returning at the height it left. Tilt the blade on that wall and the effect disappears.

A crossing is made of two things: a wall that drops dead upright, and a shadow about three millimetres deep behind it. Get those right and the eye decides on its own that one shape travels beneath the other. Lean the tool while cutting that wall and the illusion collapses.
This workshop is where the panel starts telling a story instead of showing shapes. An acanthus arm runs into a volute, the volute runs into the ground prepared for a molding, and at every meeting one of them has to win.
I like this type of detail more than almost anything else in ornament. A line disappears under another line, comes out the far side, and carries on as if nothing happened.
The whole thing lives inside about five millimetres of relief, which is the joke of it. Something that looks like two objects lying on top of each other is really one board with a few very small decisions cut into it, which is the trick sculptors have used on flat panels for centuries.
Key takeaways
- A crossing is a vertical wall plus a shadow. Nothing more complicated than that.
- Roughly 3 mm of depth is what buys you the shadow.
- Never lean the blade while cutting that wall.
- Check the finished panel for slopes before you copy a shape onto fresh wood, because a slope you did not expect will wreck the height you were aiming at.
- Swap tools until one fits. A gouge that nearly fits is the wrong gouge.
- More than three tools loose on my bench means I am tired and should stop.
What actually makes a crossing read as a crossing?
Depth difference and a clean edge. The shape on top keeps its full height and its wall stands dead vertical; the shape underneath drops away, loses its edge into shadow, and reappears further along at the same level it left.
That reappearance is what sells it. If the far side comes back at a different height, the eye works out that the two shapes never met, and the whole detail reads as decoration lying flat.
So the two ends have to agree with each other, which is why I measure them rather than trusting them.

Why must the wall go straight down?
A leaning tool undercuts, and an undercut wall is one whose top line you can no longer see. Straight down means straight down, with no tilt toward you and none away, and the whole quality of the line depends on that one discipline.
It would be a big mistake to place the blade and simply push it over. I say that on camera more than once in this workshop and I meant it every time.
Later on, from underneath and off the board, I will undercut deliberately. That is a different operation with a different purpose, done at a different stage.

How deep does the shadow line go?
Three millimetres on this panel, which is about an eighth of an inch. Deep enough to throw a dark line under raking light, shallow enough that I am not eating into wood a later stage will need.
The cut goes inward first, then runs down to the floor. Walk around the shape, take the wall, slice all the way round, and the shadow arrives as a consequence rather than as a separate job.

A ridge runs alongside all of this and must survive untouched. Cutting the shadow too deep damages it, so the depth is pinned at both ends, dark enough to read and shallow enough to spare its neighbour. I did not damage anything on this one, and that is the whole report I want from a stage like this.
Which tool actually fits an inside curve?
The one that fits. My first choice here was too big, fine along the outside and useless on the inside, so it went back and a number six at twelve millimetres came out instead.
It dropped into the curve as though it had been ground for it. That feeling is the test, and a tool that nearly fits will leave a facet you spend twenty minutes chasing afterwards.

Eight or nine gouges came out before the afternoon finished. Sixes, fives in three sizes, a nine at ten millimetres and again at a quarter inch, a flat number one, a spoon gouge near four millimetres, a seven at fourteen, plus the smallest veiner I own. None of that is showing off. Every swap happened because the tool before it had stopped seating.

What does the heel of the blade do?
It reaches places the full edge cannot. On a curved tool the heel is the very back of the edge itself, and I lean on that alone when the work sits in a pocket too tight for anything else.
My Sekira tool earns its keep here. The shape lets me slice with the heel while the rest of the blade stays clear of the wood entirely.
It is a small habit and it saves a surprising amount of damage.


Why check the first panel for slopes?
A slope you did not expect changes every height downstream of it. Before repeating this shape on the new block I went back to the finished piece and looked for one, and there was none, which made the rest of the morning simpler.
I have nothing against slopes. I like them. What they do is introduce confusion into a stage where you are already tracking four heights at once, and confusion is how a panel ends up two millimetres wrong in a place nobody can fix.

Elsewhere on the panel slopes are deliberate and quite large. One section drops away five millimetres, which is a serious tilt for work this shallow. The rule is simply to know which you are dealing with before the tool goes in.
How do you know when to stop cutting?
You measure, and then you believe the measurement. The measuring arm comes back onto the bench several times in this workshop, and a drilled hole gives me a floor I can carve down to without thinking.
One reading here came back at six millimetres, a quarter inch. I drilled it, carved to it, and the needle dropped straight to the bottom of the hole afterwards. No mistake, which is a pleasant thing to be able to say out loud.

Another reading, further along, gave seven millimetres at a spot where a small volute has to sit very low indeed. Everything around it comes down to one level so the flower can be exposed above it later.
| The place where shapes cross | The depth | What it costs you when the depth goes wrong |
|---|---|---|
| The wall belonging to the upper shape | Full height, cut at ninety degrees, with no tilt in either direction | A soft line |
| Its shadow | About three millimetres, an eighth of an inch | The crossing flattens and reads as two shapes that were drawn beside each other |
| The far end coming back | Matched exactly | It looks as though nothing ever passed beneath anything else |
| Divisions and the eye | Just half a millimetre of veiner and nothing deeper than that | Detail vanishes |
| The ridge that runs along beside it | Left untouched entirely | A repair, and repairs on work this shallow are hardly ever invisible afterwards |
Why does the grain decide the direction?
Wood tears when you come at it the wrong way, and no amount of sharpness prevents that. On the left side of this detail the cut runs across the fibres, which means slow work and a very sharp edge.
One section could only be approached downhill. Coming at it from the other direction would have caught the grain and broken the piece off, so the tool went in straight from above and cut down until it met the line.
There are several ways to remove any given lump of wood, which is one of the first things woodworking teaches you. The grain then eliminates most of them before you have finished considering.

Two waste triangles came out with a plain vertical cut for exactly this reason. Tool upright, ninety degrees, straight down, then lift the waste clear.

What is the half millimetre tool for?
The line that says one shape goes beneath another. Run the smallest veiner I own tight under the edge of the upper shape, and the lower one is suddenly passing behind rather than stopping.
It is not a deep cut. Visible, yes, but shallow, because all it has to do is catch light differently from the wood on either side of it.
Later the same tool puts in the divisions and the eye on the leaf. Tiny cuts. They carry a surprising share of what anybody ends up noticing.


What does it mean to carry a line to the same point?
Where the molding finishes, its end has to sit at precisely the height of the concave next to it. Not near it. At it, so that both surfaces feel like a single movement rolling through to one spot and stopping there.
Miss it by a millimetre and each shape still looks perfectly good on its own. Side by side they read as two separate pieces of work that happen to share a board.

The ground for that molding gets opened early, with a quarter inch number nine, then flattened with a number one. Preliminary work, nowhere near finished, and the real shaping waits for its own workshop.

What can a phone show that the camera cannot?
You pick up a phone. The fixed camera sits in front of me, so you see this panel at an angle where a width reads as something other than what it is, and no amount of explaining fixes an angle.
So I hold the phone above the work and fly it round the detail. Shaky, and I apologise for that, but you get to see a crossing from the side where it makes sense instead of from the one where it does not.
Sometimes there is a second view in the corner of the frame as well, looking straight down. I am not going to overcomplicate the setup for its own sake, though while it works you may as well have it.

For the drawing itself the phone does something better. Photograph the finished panel, open the picture, flip it, and the mirrored design sits on the bench facing the way this second piece needs. Copying beats redrawing from memory every time.

Does any of this go faster if you hurry?
Hurrying a crossing is among the more expensive mistakes available on a panel this size. Every cut lands within a hair of something that must not move, so speed buys nothing and costs you the ridge beside it.
Careful is the word I keep saying out loud while I work. Not because it sounds wise. Saying it slows my hands, and slow hands are most of the technique in carving at this scale.
The bench looks like a mess while all of this is happening and I am comfortable with that. A stage that stays tidy the whole way through is usually a stage where somebody completed a detail before working out whether it belonged there.
When is it time to put the tools down?
When they start piling up. My own test is the bench: two or three tools lying out while I work is normal, and once they begin accumulating past that, I have stopped putting them away because I am tired.
Tired hands break things. A tool makes a particular sound just before it snaps a detail off. I heard that sound on this recording and did not enjoy it.
Walk away for a few minutes. That is the entire remedy, and it costs less than the repair.

By the end of this session one line runs under another and comes back out the far side at the right height, the ridge beside it is untouched, and the ground is open for a molding that has not been carved yet. Messy in places. Correct where it matters.
This is the fifth workshop of fourteen. Workshop three builds the volute from three curves and workshop four puts the whole thing at a measured depth, which is what makes this one possible. Workshop six shapes the leaf itself. The full course carries this panel through to the end, and if you would rather begin at the beginning, workshop one slices and outlines while workshop two grounds the background flat. The measuring arm has a workshop of its own, and picking gouges is covered there too. A dull edge tears across grain, so sharpening comes first. More crossings and leaves live in acanthus over moldings, modern acanthus and the Venice Room, while the reasoning behind the movement sits in the design course.
You have two lines that need to cross properly. Access to the school puts the whole panel in front of you.
How it's done
- Find the tool that drops into the curve. Try one and feel whether it seats. Mine was too wide on the inside, so a six at 12 mm replaced it. Anything that only nearly seats will leave a facet you chase for twenty minutes.
- Take the upper wall dead upright. No tilt toward you and none away. A blade that leans has already undercut, and an undercut edge hides its own top line from every angle a viewer might use.
- Sink the shadow to 3 mm. Work inward first, then down to the floor, going the whole way round before judging any of it. Raking light across the panel does the rest.
- Match the far end to what it meets. Where the lower shape reappears, its height has to equal the concave beside it. Out by a millimetre and both shapes still look fine alone, but they stop belonging together.
- Let the fibres pick the direction. Where coming across would tear, drop the tool in from above and cut downward to the line instead. Two waste triangles left the panel exactly that way.
- Save the smallest veiner for last. Tight beneath the upper shape, no more than half a millimetre deep, and shallow enough that light rather than depth does the work. That one line is what tells anybody looking that something travelled underneath.
Frequently asked questions
How deep should a shadow line be?
Three millimetres was what this panel needed, and it has held up. On deeper relief you can afford more. What decides it is the light in the room the piece will hang in, so a panel for a dim hallway wants a heavier shadow than one under a window.
Is undercutting always a mistake?
Only when it happens by accident while you are cutting a wall. Done on purpose, from the back, after the piece leaves the board, undercutting is how an edge gets air behind it and throws a proper dark line. Same word, two completely different operations.
Why photograph the panel and flip the picture?
A mirrored panel has to carry the mirrored design, and redrawing that from memory invites errors nobody notices until the two halves hang side by side. Photograph the finished one, flip the image, and you are copying instead of remembering.

About the author
Alexander Grabovetskiy
Master Wood Carver · Founder, School of Wood Carving
Alexander Grabovetskiy is a master wood carver, named International Woodcarver of the Year. He carves classical relief the old way, from acanthus leaves to full Baroque ornament. He started the School of Wood Carving, a nonprofit, to hand these skills to anyone ready to pick up a gouge.
