Portrait eye placement
The eye lands near a cell centre — the X marks it, where a plain grid would leave it floating.
Artist guide · the X grid · sub-cell precision
Cross-diagonal grid
When a feature lands near the middle of a cell rather than a corner, a plain grid leaves you guessing. The cross-diagonal grid draws an X through every cell, splitting it into four triangles and making the cell centre an explicit point — roughly twice the placement precision of a plain grid at the same cell count, without the clutter of doubling the mesh. It is the precision endpoint of the grid-method family: the tool for reproduction copying, photorealist transfer and any subject whose key feature sits dead-centre in a cell. Here is the math of sub-cell precision, the verified history, and where the X earns its density.
- Construction
- Transfer grid + X in every cell
- Per cell
- 4 triangles + centre point
- Lineage
- Dürer · Bargue · atelier
- Difficulty
- Intermediate
- Best for
- Reproduction & centre-of-cell features
- Sweet spot
- N = 6–10
See the cross-diagonal grid on five subject categories
‹›
On a portrait, the eye and the catchlight often fall near a cell centre — exactly where a plain grid is weakest. The X marks those centres explicitly and brackets each feature inside one of four triangles.
What the overlay shows
The cross-diagonal overlay draws the same labelled square grid as the proportional transfer grid, then adds both diagonals through every cell. The X divides each cell into four triangles, marks the cell centre as an explicit reference point, and creates four edge midpoints where the diagonals meet the cell edges. Those extra reference points also give the eye fixed angles and slopes to judge a contour against, not just a corner. The result is roughly twice the placement specificity of a plain grid at the same cell count.
The key gain is the cell centre. A plain grid is weakest exactly where features tend to land — near the middle of a cell — and the X puts a point there. Crucially this precision arrives without doubling the cell count: a 16×16 plain grid matches the per-cell precision of an 8×8 cross-diagonal grid, but with twice the cells and twice the lines to read. Cross-diagonal keeps the visual weight of a coarse grid while making finer placement available.
The math: sub-cell precision without doubling N
A plain grid bounds a feature only by the four corners of its cell, leaving roughly ten to fifteen per cent positional error inside each cell. The X tightens that — it turns the cell into a fine-measurement frame rather than an open square:
8×8 cross-diagonal ≈ 16×16 plain (same per-cell precision, half the lines)
- The centre point halves the error. Adding an explicit cell centre roughly halves the eyeball error, from ten to fifteen per cent down to about five per cent, because the worst-case feature — one near the middle — now has a point to register against.
- Four triangles localise the feature. The X brackets every feature inside one of four triangles, so you read which triangle it sits in and where within that triangle, rather than across an open square.
- It is the visual-weight trade. You get a finer grid's precision while keeping a coarse grid's readability — the reason it beats simply doubling N when the composition must stay legible under the mesh.
The trade reverses past N≈12, where the diagonals start competing with the subject's own lines. Try it in the live tool — set N around 6 to 10 and the X earns its density; push it higher and a plain finer grid reads better.
History — what is real and what is myth
Verified history (with primary sources)
Renaissance roots. Diagonal subdivision of a transfer grid descends from the workshop measurement tradition Dürer documented in Underweysung der Messung (1525); his manual already included diagonal variants of the gridded screen.1
19th century — the Bargue plates. Some original plates from Charles Bargue's Cours de Dessin (1866–71) used cross-diagonal subdivision for the more complex anatomical and ornamental studies, where corner reference alone was insufficient.27
20th century — atelier formalisation. The systematic naming of the cross-diagonal grid traces to atelier teaching, where it was set out as the precision endpoint of the grid-method family — an explicit extension of the single-diagonal croisillons variant. The Art Students League lineage of Bridgman and Hale carried the broader measurement discipline forward, the same observational training Ruskin had set out for the drawing student.348
Late 20th century — photorealist reproduction. Photorealist practice, exemplified by Chuck Close's very fine gridded transfers, made exact sub-cell placement a finished-work technique; the cross-diagonal cell is the moderate-density tool for the same precision goal at a readable grid size.6
Unverified claims that won't die
"More diagonals always mean more accuracy." Only up to a point. Past N≈12 the X stops helping and starts hiding the subject; the right move there is a finer plain grid, not more diagonals.
"Cross-diagonal replaces the plain grid." No. It is the precision step you add after blocking in with a plain grid, not a substitute for it — the corners still do the coarse proportional work.
"The X gives eight times the precision." Overstated. The realistic gain is roughly a doubling of per-cell specificity — useful, but not the order-of-magnitude sometimes claimed.
When to use it (and when not)
| If you want to... | Use cross-diagonal | Don't use it for... | Difficulty |
|---|---|---|---|
| Place a centre-of-cell feature precisely | The explicit cell centre fixes a point a plain grid leaves vague | Features that already sit on cell corners or edges | Intermediate |
| Reproduce an existing artwork exactly | Sub-cell precision at a readable grid size | Loose interpretive work where exactness isn't the goal | Advanced |
| Keep a coarse grid but gain precision | The X adds reference without doubling the cell count | Whole-canvas uniform fine work (double N instead) | Intermediate |
| Catch placement errors early | Two reference systems disagree when one is wrong | Quick studies where one rough anchor suffices | Intermediate |
| Snap a flowing curve | Use the single-diagonal cousin instead for readability | Curve-heavy organic subjects (use diagonal-cell) | Beginner |
Where cross-diagonal actually appears
Six places the X earns its density — wherever exact placement of a non-corner feature matters more than visual simplicity.
Reproduction copying
Exact copying needs precision the plain grid can't give without an unreadable fine mesh.
Technical & architectural detail
Where a window mullion or fitting sits mid-cell, the triangles localise it without doubling the grid.
Centred focal points
Any strong feature that lands in a cell's middle is exactly what the X is built to pin.
Error-catching redundancy
A feature centred by the corners but off by the X means one reading is wrong — the error surfaces early.
The four-triangle demonstration
One cell, both diagonals: four triangles, four edge midpoints and the centre — the sub-cell reference made visible.
Common mistakes
1
Pushing N too high
At N=12 and above the X competes with the subject's own contours and the grid becomes harder to read than the drawing, costing more than it gives.
Fix: keep cross-diagonal at N≈6–10; if you need finer whole-canvas precision, switch to a finer plain grid.
2
Skipping the plain block-in
Jumping straight to the X without first confirming the corner placement means sub-cell precision built on a coarse error — accurate detail in the wrong place.
Fix: block in with the plain grid first, verify the corners, then add the diagonals.
3
Using the X for curve-heavy organic work
Both diagonals on every cell is reproduction-grade density that buries soft organic forms; hair and drapery read better with a single diagonal.
Fix: for curves use the diagonal-cell grid; reserve the full X for exact reproduction and centre-of-cell features.
How different disciplines use it
For portrait painters
Faces put their most-watched features — the eyes, the catchlights, the corner of the mouth — near cell centres, the one place a plain grid is weak. Block in the head with a coarse grid, then switch on the X so those centred features have an explicit point and a triangle to sit in. Keep N around 6 to 10 so the diagonals stay legible against the form.
For atelier students
Cross-diagonal is the precision endpoint of the grid family you learn after the plain grid and the single-diagonal croisillons. Use it to understand why grid precision matters — the X is a visible reminder that a feature has a position inside the cell, not just within it — the shift from drawing what you expect to drawing what is there.5 Treat it as a checking tool: a feature centred by the corners but off by the diagonals tells you a reading is wrong.
For reproduction copyists
Matching an existing artwork exactly is where the X pays for its density. It delivers sub-cell placement at a grid coarse enough to read, so you can locate a non-corner detail precisely without dropping to an unreadable fine mesh. The redundancy of two reference systems — corners and diagonals — catches the small drift that ruins a faithful copy.
For technical illustrators
When a fitting, mullion or label lands mid-cell, the four triangles localise it without forcing the whole drawing onto a finer grid. For uniform whole-drawing precision, doubling the cell count is cleaner; for a few critical non-corner features on an otherwise coarse grid, the cross-diagonal cell is the efficient choice.
"Measurement is not the enemy of art; it is the floor it stands on. The more exactly you can place a point, the freer you are everywhere else."
George B. Bridgman, Constructive Anatomy (1920)3
Frequently asked questions
What is an X grid in drawing?
An X grid — another name for the cross-diagonal grid — is a plain transfer grid with both diagonals drawn through every cell, so each cell carries an X. The X-grid method splits each cell into four triangles and marks the cell centre as an explicit point, giving the sub-cell reference a corner-only grid can't. It is the precision endpoint of the grid-method family for accurate drawing and photorealistic transfer.
Is the cross-diagonal grid the most accurate grid for drawing?
For locating non-corner features it is the most precise grid in classical practice that doesn't require doubling the cell count: the grid method with two diagonals roughly halves the per-cell eyeball error versus a plain grid. For uniform whole-canvas precision a finer plain grid can match it with cleaner geometry, so the cross-diagonal grid is the grid for accurate drawing where a few features land mid-cell on an otherwise coarse mesh — including grid for photorealistic drawing work.
What is the cross-diagonal grid?
A standard square grid with both diagonals drawn through every cell, forming an X. The X subdivides each cell into four triangles and makes the cell centre an explicit reference point — about twice the placement specificity of a plain grid at the same cell count. It is the finest-resolution observational grid in classical practice that doesn't require doubling the overall grid.
How is cross-diagonal different from the diagonal-cell grid?
The diagonal-cell grid typically uses one diagonal per cell to snap curves on organic subjects, prioritising readability. The cross-diagonal grid always draws both diagonals for maximum sub-cell precision — twice the placement reference per cell — at the cost of visual density. Use diagonal-cell for curves, cross-diagonal for exact reproduction.
When is doubling N better than cross-diagonal?
When the whole canvas needs uniform fine precision — architectural reproduction, technical illustration — doubling the cell count gives the same precision with simpler line geometry that reads more cleanly. Cross-diagonal is better when you want the visual weight of a coarse grid so the composition stays legible, with finer precision available where you need it.
How many reference points does a cross-diagonal cell give?
Beyond the four corners, the X adds the cell centre and the four edge midpoints where the diagonals cross the cell edges, and it divides the cell into four triangles. In practice that roughly halves the eyeball error per cell, from about ten to fifteen per cent down to five per cent or better.
At what grid size does cross-diagonal stop helping?
Around N=12 and above. The diagonals start competing visually with the subject's own lines, and the grid becomes harder to read than the drawing. Cross-diagonal is most useful at N of about 6 to 10, where the cells are large enough that sub-cell reference genuinely helps.
Where does the cross-diagonal grid come from?
It is older than the name. Renaissance grid-method instruction such as Dürer's 1525 manual included diagonal variants, and some original Bargue plates used cross-diagonal subdivision for complex studies. The systematic naming traces to 20th-century atelier teaching, where it was formalised as the precision endpoint of the grid-method family.
Can cross-diagonal be combined with other overlays?
Yes. It is the precision step above the plain proportional-transfer grid and the curve-focused diagonal-cell grid, it scales onto walls through the mural-scaling variant, and for figures it pairs with fixed measurement so you can count head-heights alongside the sub-cell reference.
References
- Dürer, A. Underweysung der Messung (Course in the Art of Measurement) (1525). Trans. as The Painter's Manual, W.L. Strauss. Abaris (1977). ISBN 0-913870-23-5.
- Bargue, C. & Gérôme, J-L. Drawing Course (Cours de Dessin) (1866–71). Ed. G. Ackerman, ACR Édition (2003). ISBN 2-86770-128-6.
- Bridgman, G.B. Constructive Anatomy (1920). Reprint: Dover (1973). ISBN 0-486-21104-5.
- Hale, R.B. Drawing Lessons from the Great Masters. Watson-Guptill (1964). ISBN 0-8230-1401-4.
- Edwards, B. Drawing on the Right Side of the Brain. J.P. Tarcher (1979; rev. 2012). ISBN 0-87477-419-5.
- Storr, R. (ed.). Chuck Close. The Museum of Modern Art, New York (1998). ISBN 0-87070-068-8.
- Ackerman, G. Charles Bargue and the Art of Drawing. ACR Édition (2003). ISBN 2-86770-152-9.
- Ruskin, J. The Elements of Drawing (1857). Reprint: Dover (1971). ISBN 0-486-22730-X.
Other Artist Guide overlays
Notes from the studio · Three practitioners on the cross-diagonal grid
Illustrative composites of how the tool gets used in practice — not quotes from named individuals.
Eyes land in the middle of the cell every time. The X gives me the centre point a plain grid never does, and I keep it at eight by eight.
For a faithful copy I cross two readings — corners and diagonals. When they disagree, I've found my error before the paint is down.
Free and browser-only means I can dial N up and down on any machine until the X helps instead of hides the form.
Open the cross-diagonal overlay
Drop a reference image. The X applies through every cell in one click. Free, in your browser.
Launch Grid Maker Pro →