How the Grid Method Can Help You Break Down Complex Drawings

A complex reference is not harder to draw than a simple one. It is the same drawing problem repeated two hundred times, and the reason it feels impossible is that you are trying to solve all two hundred at once. A grid makes you solve them one at a time. This is how to choose the grid, where to make it finer, and how long the result actually takes.

If the method itself is new to you, the complete grid drawing guide covers it from the start, and the basics of grid drawing for beginners is the short version. Everything below assumes you already know how to lay a grid and is about what changes when the subject is dense. You can put one on your own reference with the grid drawing tool.

What the grid actually takes off your hands

The usual explanation is that a grid reduces cognitive load — that you can only hold a few things in your head at once, so you work on one square at a time. There is a real finding underneath it. Nelson Cowan's 2001 review put working memory at about four chunks, revising George Miller's famous seven, and Luck and Vogel had shown in 1997 that visual change-detection stays accurate up to three or four objects and falls off after that.

It is the wrong mechanism for drawing, though. Your reference is not in your memory; it is on the desk in front of you, and you look back at it every few seconds.

The better-fitting result is Cohen and Bennett's 1997 study, Why can't most people draw what they see. They separated the four things that could go wrong in a drawing — misperceiving the subject, deciding badly what to put down, poor motor control, and misjudging your own marks — and tested the last three. Motor control barely mattered. Decision-making barely mattered. Misjudging your own drawing was not a source of error at all. What was left, and what carried nearly all of it, was misperceiving the subject in the first place.

That is what a grid is for, and it is worth being precise about it: a grid does not teach you to see better. It shrinks the judgement until the seeing you already have is good enough. Deciding where an eye sits on a whole face is a judgement most people get wrong. Deciding where a line crosses a half-inch square is one almost nobody gets wrong.

Pick a cell size, not a number of cells

Everything else here depends on this one decision, and the useful way to make it is backwards: decide how much detail you are willing to judge inside one cell, then let that set how many cells there are.

Grid density is the number of cells covering the reference, written columns x rows, always in that order. An 8 x 5 grid is eight columns wide and five rows tall.

Work in cell size on the reference. If your printed reference is 6 x 4 inches:

• 1 inch cells give you 6 x 4 = 24 cells. Right for simple shapes — a single figure against a plain ground, a piece of fruit, a building with no ornament.
• Half-inch cells give you 12 x 8 = 96 cells. Right for most complex references: a street, a face, foliage with structure.
• Quarter-inch cells give you 24 x 16 = 384 cells. Only for the busiest patch, and only as a subdivision of the half-inch grid.

The check: hold a cell at arm's length and describe out loud what is inside it. If you need more than one short sentence, the cell is too big. If you find yourself saying "a bit of the roof edge", it is about right.

Then the paper. Whatever the cell counts came out as, the paper carries the same counts, at whatever cell size fits the sheet. A 6 x 4 reference gridded at half an inch is 12 x 8 cells; on a 24 x 16 inch sheet that is 12 x 8 cells of 2 inches. Same grid, four times the size. The counts are what transfers — never the measurements.

This is the mechanism the whole method runs on, and it is the one people break first. Reference and paper must carry the same number of cells. If they do not, cell B4 on the reference is not cell B4 on the paper, and every measurement you take is being applied to the wrong piece of the drawing.

Where to make the grid finer

Not every part of a complex reference needs the same grid. Sky, plain walls and out-of-focus background carry almost no information, and a fine grid over them is cells you have to draw for nothing. Put the fine grid only where the detail is. The rule that keeps this honest: the fine grid must be a subdivision of the coarse one — halve or quarter a cell, never start a second unrelated grid. A 6x4 base grid can carry 12x8 inside two of its cells because every fine line lands on a coarse one. A 6x4 next to a 5x5 shares no lines at all, and the two stop agreeing about where anything is.

The image at the top of this page is exactly this: an 8 x 5 base grid over the whole scene, halved to 16 x 10 across the two columns that hold the ornamented facade. Every fine line sits on a coarse one.

Deciding where the fine grid goes takes about a minute. Look at the reference and mark the areas that would embarrass you if they were wrong: the face, the ornament, the mechanism, the focal point. Those get the subdivision. Everything else keeps the base grid, including areas that look busy but carry no structure, such as gravel, leaf litter and distant foliage. Busy is not the same as detailed. Buildings are the case where this pays most, and using grids to draw architecture accurately goes through the perspective side of it.

The five steps

1. Grid the reference at the size you chose. Print it if you can.
2. Copy the same cell counts onto your paper, light lines, 2H or harder. Check: count the cells on both. If the counts differ, stop now.
3. First pass, whole drawing: mark only where each major shape crosses a grid line. Check: step back; shapes should read as the right subject at the right proportions from across the room.
4. Second pass, whole drawing, cell by cell but only to "the main thing in this cell, roughly right". Check: every cell has something in it before any cell is finished. This is the habit that prevents the patchwork look.
5. Third pass, detail, starting at the focal point. Check: if a line stops dead at a grid line you are drawing the grid instead of the subject — redo it in one stroke across both cells.

When it goes wrong, it nearly always goes wrong in the same place. You reach step 3, the shapes do not read from across the room, and the instinct is to start fixing details in the hope that the whole will come right. It will not. A first pass that is out is a first pass that has to be redone, and redoing it costs twenty minutes while carrying it through to a finished drawing costs the drawing. Erase the first pass and lay it again. If it is out a second time, the cause is almost never your eye: it is a mismatch in cell counts, or a reference that has been cropped since you gridded it.

Common pitfalls and solutions

Over-detailing the first sections

Problem: you perfect the first few cells while the rest of the sheet is empty. Time and energy drain before you reach anything that matters, and the finish level across the drawing ends up uneven.

Solution: the three passes above exist to make this impossible. No cell gets its third pass until every cell has had its second.

Grid dependency

Problem: you stop being able to draw without one. Quick sketching goes, and with it the practice that would eventually make the grid unnecessary.

Solution: shrink the grid's job over months rather than dropping it. Use it for the initial proportions only, then take the detail freehand. Sketch daily without one. Keep the grid for genuinely complex work, which is what it is for.

Losing the whole

Problem: cell-by-cell work turns into tunnel vision. Values stop relating to each other, and elements at opposite ends of the sheet drift apart.

Solution: step back every 20 minutes and break every hour. Photograph the drawing with your phone and look at the photo, which flattens it and shows value problems your eye has stopped seeing. Working standing up makes stepping back automatic.

Mechanical, segmented results

Problem: the drawing is accurate and looks assembled from tiles. Lines stop at cell boundaries, line weight never changes, and nothing flows.

Solution: treat grid lines as guides, not walls. A branch, a cable or the curve of a jaw is drawn in one movement across as many cells as it takes, then checked against the grid afterwards rather than during.

Uneven finish between cells

Problem: the cells you enjoyed are fully rendered and the ones you did not are still sketches. Nothing destroys a complex drawing faster, because the eye reads the unevenness before it reads the subject.

Solution: decide the finish level before you start and apply it to the dull cells first, while you still have the patience for them.

Working with grid coordinates

Once a drawing runs past a few dozen cells you need to be able to name them, if only to note where you stopped. Columns take letters and rows take numbers, so a cell is B4.

GridMyPic generates these coordinate labels automatically. The lettering is Excel-style, so it does not run out at Z: column 27 is AA, then AB, and so on for as many columns as the grid has. Turn them on in the grid maker.

Two conventions are worth fixing early. Label the reference and the paper identically, including the direction the numbering runs, because a reference numbered from the top and a sheet numbered from the bottom will agree everywhere except where it matters. And when you subdivide a cell, name the parts after the parent: B4 becomes B4a to B4d, not a new sequence. The name should tell you where you are without counting.

Crowd scenes and other extreme subjects

Some references go beyond the usual: a crowd of a hundred figures, a dense canopy, a coral reef. The grid handles them, but only if you decide in advance how much of each element survives.

Sort the figures into levels before drawing any of them. Foreground figures get full detail and individual character. Mid-ground figures stay recognisable as specific people but lose their features. Background masses become shapes and value patterns with no faces at all. The far crowd becomes a rhythm of heads and shoulders.

Then the grid follows that hierarchy rather than the subject's own density: subdivided cells over the foreground figures, base cells over the mass. What defeats people in crowd scenes is giving the hundredth figure the same attention as the first, which is both exhausting and wrong, because a crowd where every face is equally sharp does not look like a crowd. Where a face in the foreground does need to hold up, how grids can help you create accurate graphite portraits covers the placement of features specifically.

Street corners, curved roads and buildings at different angles put several vanishing points into one scene. The grid does not resolve them, and it does not have to: it holds placement steady while each area follows its own local perspective. That is a separate problem, worked through in understanding perspective with grid drawing.

How long it will take

Time three cells of average difficulty, multiply by your cell count, then add half again. That is close enough to know whether this is a weekend or a winter.