The Foundation Most People Skip
You don't start a caricature by exaggerating the nose or the jawline. You start by understanding what makes a face recognizable in the first place. I learned this the hard way after wasting three months drawing silly faces that nobody could identify. A caricature is a distortion of specific features relative to the underlying structure, not just a cartoon with big eyes and a small mouth. If you skip the structural analysis, your work looks generic no matter how much you exaggerate. Here's the practical method I actually use, not the textbook version. First, pick a reference photo where the subject is looking straight at the camera with neutral lighting if you can manage it. Side profiles and heavy shadows make feature mapping unnecessarily difficult. Take that photo and overlay a basic head form — an egg shape with a center line and horizontal guides for the eyes, nose, and mouth. I usually sketch this lightly in my drawing software, though some people still do it on paper with a tracing pad. The key insight that beginners miss: measure the distances between features, not the features themselves. How far apart are the eyes relative to the width of one eye? Where does the brow sit relative to the hairline? These ratios are what carry identity. Once you have your measurement map, you decide which features to push and which to hold steady. This is where the art actually happens. A portrait artist renders everything equally; a caricaturist picks the top two or three characteristics and amplifies them. Everything else stays closer to reality. I've seen too many beginners blow up every feature at once, which just creates chaos rather than a likeness. The face becomes unreadable because nothing anchors it to the original person.
Let me give you a specific example from a job I did last year. A client sent me a headshot of a tech CEO — distinctive wide-set eyes, a recessed chin, and unusually thick eyebrows. My instinct was to exaggerate all three equally. Instead, I dialed the eye spacing out to about 1.5 times the natural width, kept the chin relatively grounded, and pushed the brows into a heavy, expressive shelf. The resulting drawing looked exactly like that person, just funnier. When I showed it to a colleague, they identified him immediately from a single glance. That's the metric that matters: can someone still tell who it is? One edge case that always trips people up is working from group photos or candid shots where everyone is turned at different angles. I ran into this with a sports magazine piece — they wanted caricatures of five players from game action shots, none of which were clean headshots. The workaround I settled on was pulling still frames from broadcast footage, enlarging each to find the dominant features, and cross-referencing with one official team photo per player for structural accuracy. It added about forty minutes to what would normally be a two-hour session, but it prevented me from guessing at facial geometry. Don't skip the reference check just because it slows you down. Another counter-intuitive thing: less exaggeration on the eyes often produces a stronger likeness than more. Beginners want to make the eyes huge because that's what cartoons typically do. But the eyes carry so much identity already that pushing them too far makes the face look like a generic illustration rather than a specific person. I keep the ocular region relatively restrained and redirect the distortion budget toward ears, jaws, foreheads, or noses depending on the subject. A pronounced ear shape or an unusual brow ridge will do more work for readability than oversized eyes ever will.
Tools matter but they don't make the drawing. I work primarily in Clip Studio Paint on a Wacom tablet because the brush engine responds predictably and the layer system lets me lock in the structural sketch before committing to linework. Procreate works fine too, and honestly the best tool is whatever lets you draw without fighting the software. Some professionals still draw on paper and scan, which is perfectly valid. The workflow I'd recommend for getting started is straightforward. Sketch the head form and placement guides first. Map the actual features on top of those guides. Lock that layer, create a new one, and redraw the features with your chosen exaggerations. Clean up the linework on a third layer, add minimal shading if needed, and step back to check whether the subject is recognizable at thumbnail size. If you can't read the face when it's small, the exaggerations are either too scattered or not pointed enough. There are real limitations to this approach that nobody talks about enough. Caricature relies heavily on facial features, so subjects with very symmetrical faces or those whose identity lives more in body language and posture tend to be harder to capture effectively. A comedian's entire persona might come from their physical presence and gesture, which doesn't translate well to a head-only drawing. In those cases, the best move is often to include a simplified body element or to acknowledge upfront that the piece will lean more toward impression than direct likeness. Being honest about what the medium can and cannot do saves you from delivering work that feels incomplete. If you want to practice without a live subject, grab any celebrity headshot and try the ratio mapping exercise I described. Spend ten minutes just measuring distances before you commit to a single exaggerated line. Most people rush past that step and spend the rest of the session fixing a proportion that was wrong from the beginning. A proper measurement phase usually cuts the overall process down from around two hours to roughly forty-five minutes for a polished piece, assuming you're working at editorial quality. The time investment upfront pays for itself in revision reduction.
Get the Full Details
