D&D Character Portrait Prompts for Tabletop RPG Art

Head-and-shoulders character portrait of a [race] [class], [age/build], [skin tone] skin, [hair color and style], [eye color] eyes, [1-2 signature features like a scar, tattoo, or heterochromia], wearing [armor/clothing] and carrying [weapon or focus], [mood/expression: grim, kind, cunning], [style lock], soft key light, plain [color] background, centered composition.

It forces the six choices that actually define a character, race, class, features, gear, mood, style, so the model has no room to invent generic filler.

Take any portrait prompt and add this to the end: personality reads as [two contrasting words like weary optimist or polite killer], expression matches.

A single mood tag changes the eyes and mouth more than any gear description, which is what makes a face feel like a person instead of a mannequin.

Append to any prompt: no text, no watermark, no logo, no extra limbs, no weapon clipping through the body, single character only, no modern clothing.

Portrait models love to smuggle in gibberish runes and a phantom second sword, so naming them kills them before they render.

STYLE: painterly fantasy oil portrait, visible brushstrokes, warm rim light, rich desaturated palette, in the tradition of classic tabletop rulebook cover art, dramatic chiaroscuro, matte finish.

The rulebook cover art cue pulls the model toward the hand-painted look most players picture for a hero, without naming a living artist.

STYLE: photoreal cinematic portrait, 85mm lens, shallow depth of field, realistic skin texture and fabric weave, moody practical lighting like torchlight or overcast sky, film grain minimal.

The lens and skin-texture callouts push it past the plastic AI sheen into something that looks like a real actor in costume.

STYLE: bold comic book ink portrait, clean black linework, flat cel shading, limited high-contrast palette, graphic novel cover framing, crisp edges.

Ink and cel-shade styles read cleanly when a portrait gets shrunk down to a tiny table token, where painterly detail turns to mud.

Head-and-shoulders portrait of a battle-scarred half-orc fighter, broad jaw, lower tusks, close-cropped hair, weathered green skin, dented steel plate armor with a fur mantle, one hand resting on a greatsword hilt, steady unafraid gaze, [style lock], plain stone-gray background.

Concrete gear and a single steady expression give a martial class weight, so it looks like a veteran rather than cosplay.

Head-and-shoulders portrait of an elderly gnome wizard, big curious eyes, wild white eyebrows, ink-stained fingers, patched star-patterned robe, brass spectacles pushed up, a tiny arcane spark hovering at one fingertip, wry knowing smile, [style lock], deep-blue background.

One small magical detail like the spark sells spellcaster far better than a generic glowing aura swallowing the whole frame.

Head-and-shoulders portrait of a lean half-elf rogue, sharp cheekbones, subtly pointed ears, hood half up casting a shadow over one eye, dark leather armor with hidden buckles, a dagger just visible at the shoulder, sly guarded half-smile, [style lock], muted alley-brown background.

Partial shadow and a half-drawn hood say sneaky visually, doing the storytelling that the word rogue alone cannot.

Head-and-shoulders portrait of a stout dwarf cleric, thick braided beard threaded with a holy symbol, kind tired eyes, engraved chainmail under a tabard, a warm glow rising from a pendant, resolute compassionate expression, [style lock], warm candlelit background.

Pairing kindness with tired eyes gives a healer real depth, someone who has patched up too many wounds to look fresh-faced.

Add to any portrait prompt: framed for a round token, face and shoulders fill the center, key features stay inside a central circle, high contrast against a simple background, readable when shrunk to thumbnail size.

Virtual tabletops crop portraits to a circle, so anything important pushed to the corners gets sliced off unless you center it on purpose.

Add to any portrait prompt: three-quarter view, from mid-chest up, even flattering light, full costume detail visible on the collar and shoulders, vertical portrait orientation suited to a character sheet slot.

A three-quarter angle from mid-chest shows off armor and gear that a tight face crop would hide, which is what a sheet portrait is for.

Head-and-shoulders portrait of a dragonborn paladin, bronze-scaled draconic head with a proud brow ridge, glowing golden eyes, ornate holy plate armor at the shoulders, a faint sacred light behind, noble unwavering expression, framed for a round token with the face centered, [style lock].

It shows the framing rule and the library rule working together, so you can see exactly where the token crop clause slots in.

Reuse this exact block in every image of the character: SAME CHARACTER, copper-red braided hair, three freckles under the left eye, a jagged scar across the right brow, amber eyes, a chipped front tooth when smiling. Only the pose, lighting, and scene change.

Locking three to five oddly specific features and repeating them word for word is what makes the model draw the same person twice instead of a lookalike.

When your tool allows it, keep the same seed number and feed your favorite portrait back in as an image reference at medium strength, then only edit the prompt words for the new pose or outfit.

The seed keeps the underlying face math stable while the reference image carries the likeness, so small prompt tweaks no longer reroll a whole new stranger.

Using your locked feature block, generate the same character as: a calm neutral portrait, an angry battle-ready shot, and a relaxed off-duty look, keeping every listed feature identical across all three.

Asking for a set in one go with the anchor block gives you a mini reference sheet, which is far more consistent than trying to match a face weeks apart.