Styles
Browse a style. Copy the prompt.
Every style comes with its prompt and the model I recommend for it. Click one to see more examples and copy the prompt.
- Styles
- 8
- Source
- Generated stills
No styles in that family.
Illustrated
Satin Noir
Night scenes in monochrome. Elongated Art Deco figures in flat shapes, grey-pink skin, fine grain, and a few polished white highlights on deep black.
Prompt · for GPT Image 2.5 Sunburst
Swap the highlighted scene for your own. The rest is the style, so leave it as it is.
Monochrome vintage editorial illustration, flat vector shapes finished with fine grain. Scene: [Your scene in a few plain sentences. E.g. a couple dancing close in an Art Deco ballroom, her head thrown back, his hand at her waist.] Shapes: every tone is a separate flat cut-out shape with a crisp edge, three or four tones layered like paper. No soft 3D shading, no drawn lines for wrinkles or folds, no outlines anywhere. Shape edges end in rounded, melted, teardrop-like tips; nothing actually drips. Some highlight lines are broken into dashes and dots. Faces and figures: simplified into a few large flat planes, sculptural, Art Deco. Tall elongated skulls, long column-like necks, small facial features set low in the face, jaws cut as straight angled edges. Lowered eyes, each a single thin curved stroke, no irises. Eyebrows are thin high arcs that thicken at one end. Hair is a black mass held close to the head, with thin white highlights along its edge. Poses tilted to extremes. Hands: every visible hand is anatomically correct, with exactly five digits, one thumb and four fingers, each finger its own shape separated by a thin dark gap, posed simply and clearly. Light: one hard light from above. Lit surfaces are neutral grey with a faint pink tint, never brown or sepia. The shadow side of everything has no edge and dissolves into the black background, so forms exist only through their lit planes. A few crisp pure-white highlights as sharp ribbons and teardrops on rims and polished surfaces; everything looks polished like satin. No light greys between the midtones and the white. Texture: soft gradients inside the shapes made of fine speckled grain like airbrush spray; sparse white specks across the black like dust. Colour: near-black charcoal fills about three quarters of the frame, grey-pink midtones, pure white only for highlights. No colour at all. Composition: subjects cropped by the frame edges, strong diagonals, dark areas merging into the background, large areas of empty black. Mood: hushed, glamorous and nocturnal. Avoid: outlines, realistic or 3D-rendered faces, open detailed eyes, cartoon proportions, smooth gradients without grain, sepia or brown tones, any colour, busy detailed backgrounds, extra or missing fingers, fused fingers.
Model recommendation
- I recommend GPT Image 2.5 on Sunburst, its higher-quality version. Every image here was made with it, from this prompt alone, with no reference image.
- It works well on other models too. I tested it on eight, and Seedream, Nano Banana Pro and Mystic also hold the look.
- On Nano Banana Pro or Seedream, delete the line about melted, teardrop-like tips, since those models take it literally.
- Keep the Hands paragraph. It keeps fingers clean and easy to read.
- Use it on any subject. The scenes here are just examples.
- My settings: Sunburst, high quality, 3:4 at 1536 × 2048.
Illustrated
Block Party
Busy city scenes seen from high above, in clean ink lines and flat colour, crowded with people living their small everyday moments.
Prompt · for GPT Image 2.5 Sunburst
Swap the highlighted scene for your own. The rest is the style, so leave it as it is.
Isometric editorial comic illustration: clean ink lines, flat colour and a crowded, lived-in world. Scene: [Your scene in a few plain sentences: the place, the city and the time of day; what stands along the back, with sign names in quotes; what is in front; what people are doing and where. E.g. the Union Square Greenmarket in New York on a weekday. Along the back, farm stalls stand under white tents with signs reading "HUDSON VALLEY FARM" and "BREAD & BUTTER". In front is the open plaza. Shoppers fill bags at the stalls, a chef wheels a hand cart of produce and a busker plays guitar by a bench.] View: everything is drawn from one high isometric angle. No perspective: vertical lines stay vertical and parallel lines stay parallel. Close enough to read every small action. Anything taller or longer continues past the edges of the frame at its true size. No sky. Composition: tall forms stand only along the back of the frame, facing the viewer. The front of the frame stays open and low, so nothing blocks the view. Figures: small, simple and whole, each absorbed in their own everyday moment, facing the way they move and resting naturally on whatever is under them. The frame feels busy and crowded. Details: generous lived-in clutter and small real-life touches throughout. Lettering: the only words are the ones in quotes in the scene, each spelled exactly. Other signs show pictures instead of words. Line and colour: thin, even, dark brown ink outlines. Flat fills with one darker flat shadow per surface, no gradients. Colour follows the time of day in the scene, daytime if none is given: - Morning: fresh pastels of mint, powder blue, butter yellow and soft coral on cream. - Daytime: warm, dusty sand, beige, cream, sienna and faded mauve, with plum-grey shadows. - Sunset: ochre, apricot, terracotta and honey, with long violet shadows. - Night: indigo and blue-grey, lit by warm yellow pools of light. Mood: warm, curious and gently funny. Avoid: 3D rendering, photorealism, vanishing-point perspective, sky, extra text.
Model recommendation
- I recommend GPT Image 2.5 on Sunburst. Every image here was made with it from words alone, with no reference image.
- The colours follow the time of day. Write morning, sunset or night into the scene, or leave it out for daytime.
- Name a real city or neighbourhood in the scene. The streets, crossings and buildings follow how that place really looks.
- Say what people are doing and where, like "reading on a bench" or "taking photos from the railing". It keeps every pose believable.
- Put your main sign names in quotes in the scene and they come out spelled exactly. Smaller signs stay as little pictures.
- For a grittier street, add "litter and stains everywhere" to the scene.
- In my model tests, Flux 3 came closest after GPT Image 2.5.
- Use it for any busy place. The scenes here are just examples.
- My settings: Sunburst, high quality, 3:4 at 1536 × 2048.
Illustrated
Sticker Street
Worn everyday street objects drawn like screen-printed comic posters, in heavy black ink, hot pink and silver, covered in stickers, scuffs and tags.
Prompt · for GPT Image 2.5 Sunburst
Swap the highlighted scene for your own. The rest is the style, so leave it as it is.
Inked comic-poster illustration: heavy black brush ink with a screen-printed feel, led by hot infrared pink and silver on a pale silver ground. Scene: [Your subject in a few plain sentences: what it is and where it stands, how it is seen, and what is stuck and written on it, with any words in quotes. E.g. a battered auto-rickshaw parked at the kerb, seen from the front in three-quarter view from slightly above, so its headlight and windscreen face the viewer. Peeling stickers, tape and tags cover the front panel, and a sticker across the top of the windscreen reads "NO HURRY".] Subject: the hero of the image, filling most of the frame, with parts of it running past the edges. It is drawn true to how it is really built, slightly chunky, used and lived-in, with dents, scuffs, tape and worn edges. Background: one flat pale silver-grey field, cool and slightly blue, like silver paper. No setting and no horizon, at most one small hint of place. Line: bold black brush-ink outlines, thicker on the outer contour and thinner inside, slightly hand-wobbly. Every part is outlined. Shadows: no gradients and no soft 3D shading. Each colour has two flat tones, a lit tone and a shadow tone. Deep shadows, folds and gaps are solid black shapes, like comic spot blacks. One hard light from the upper left throws hard-edged solid black shadows, with one long shadow falling across the silver. Colour: silver and hot infrared pink lead the image. Painted surfaces are infrared pink, with deep raspberry shadow tones and pale blush lit planes; metal is silver-grey; the darkest materials are black. Everything else keeps its real colour as a small accent, with tiny pops of acid yellow, mint green and electric blue. Highlights: glossy surfaces get two or three crisp white streaks and at most one small four-point sparkle. Matte surfaces get none. Texture: a fine speckled print grain over the whole image, visible as tiny light specks in the black areas, like a risograph print. Surfaces: layered with peeling stickers, tape, tags and scrawled notes that belong to the subject's own world, each one different and drawn in the same ink style. Every design is original and made up for this picture, with no logos, brand names, mascots or characters from games, films or cartoons. The only words are the ones in quotes in the scene, hand-scrawled and spelled exactly; everything else carries pictures. Mood: gritty, loud, funny and lived-in. Avoid: gradients, 3D render look, photorealism, busy backgrounds, clean new surfaces.
Model recommendation
- I recommend GPT Image 2.5 on Sunburst. Every image here was made with it, from words alone, with no reference image.
- Say how your subject is seen. From above, straight on and looking up all hold the look.
- Put the words you want on it in quotes in the scene. In my tests, every quoted word came out spelled exactly.
- Name a few stickers in your scene, like "a coin and a clothes peg". It keeps each image different.
- Use it on anything worn and lived-in. These are just examples.
- My settings: Sunburst, high quality, 3:4 at 1536 × 2048, or square at 2048 × 2048.
Illustrated
Big Kicks
Athletes in streetwear in bold flat colour, with a giant sneaker or ball right up front and swirling pastel skies behind.
Prompt · for GPT Image 2.5 Sunburst
Swap the highlighted scene for your own. The rest is the style, so leave it as it is.
Flat editorial illustration: bold flat colour shapes, extreme forced perspective and swirling pastel backgrounds. Scene: [Your scene in a few plain sentences: who, doing what, from where we see it, and what is nearest the camera. E.g. a young skateboarder in mid-air above the camera, seen from below and slightly to one side. The underside of the board is huge in the foreground, both sneakers stand on top of it, and the body shrinks up toward a small head against the swirls.] Perspective: exaggerated, like a wide lens held close. The nearest thing is drawn very large, and everything further away shrinks smoothly in true proportion. Any ground line curves gently. Shapes: everything is built from clean, flat colour shapes with crisp, smooth vector edges, like cut paper. No outlines; a change of colour makes the edge. The nearest thing is the exception, drawn in detail with thin dark lines. Faces and figures: stylised, long-limbed and confident. Skin has one lit tone and darker flat shadow shapes. Faces are simple: the eyes and brows are thin dark strokes, the eyes half-closed, with a small shadow for the nose and a simple mouth. Hair is a solid mass with a shaped edge. Bodies: anatomically correct under the style. Every limb has all its parts in the right order and proportion: thigh, knee, shin, ankle and foot; upper arm, elbow, forearm and wrist. Limbs bend only at real joints, loose clothes still show where those joints are, and every pose has believable weight and balance. Hands: every visible hand has exactly five digits, one thumb and four fingers, posed simply and clearly. Clothes: loose shapes in solid colour blocks, original designs with no logos, brand marks or writing. Background: big flowing swirls and rounded cloud shapes in soft pastels, flat and layered, with one large flat circle. No real place, at most one small hint of it. Colour: pastel backgrounds of blush pink, cream, peach and periwinkle, with punchy, fully saturated accents of lime, vermilion, orange, cobalt blue and butter yellow on the subjects. Navy black only for the darkest darks. Finish: clean digital flat colour, with no grain, noise or paper texture. Mood: energetic, playful and confident. Avoid: outlines around figures, 3D render look, photorealism, smooth gradients, painterly brushwork, text.
Model recommendation
- I recommend GPT Image 2.5 on Sunburst. Every image here was made with it, from words alone, with no reference image.
- Whatever is nearest the camera comes out huge. For giant sneakers, put the camera low and close to the feet, and say so in the scene.
- Keep the Bodies paragraph. It keeps anatomy solid at extreme angles.
- Say how the body sits in the frame and where the feet are, like "both sneakers stand on top of the board". It keeps the pose believable.
- Give each person their own space and one clear action.
- Use it for any active scene. These are just examples.
- My settings: Sunburst, high quality, 3:4 at 1536 × 2048, or square at 2048 × 2048.
Illustrated
Face Value
Portraits built from flat, hard-edged shapes of colour, with fine hatching, heavy brows and one bold accent on a plain background.
Prompt · for GPT Image 2.5 Sunburst
Swap the highlighted scene for your own. The rest is the style, so leave it as it is.
Stylised digital portrait: a face built from hard-edged flat planes of colour, with heavy brows and a long neck, on a plain muted background. Scene: [Your person in a few plain sentences: who they are, what they wear, how the head is turned and where they look, and where the light comes from. E.g. a man in his seventies with a deeply lined face and a short white beard, in a dark work jacket, turned three-quarters to the left and looking straight at the viewer, lit from the side by warm window light.] Planes: the skin is built from flat, hard-edged shapes in three to five tones, some angular and some blocky like overlapping rectangles, with crisp edges across the forehead, the bridge of the nose, the cheekbones and the jaw. Paler patches of light sit on top as clean blocks. Little soft blending. Age shows as more planes and sharper creases, never as smooth skin. Features: the brows are heavy, angular slabs with fine hair strokes along their edges. The eyes are long and heavy-lidded, with a dark upper lash line running out into a thin wing and one small highlight. The nose is two or three flat planes. The lips are in three flat tones with a crisp outline. The jaw is cut in straight lines. Proportions: stylised but believable. A slightly large head, a narrow chin and a long, column-like neck that joins the shoulders properly. The features all follow the turn of the head. Any visible arm bends only at real joints, and any visible hand has five digits, one thumb and four fingers. Hair: a solid mass with blocky patches of lighter colour and fine parallel hatch lines, with a few thin, wiry ink lines tracing its outer edge and straying from it. Clothes: simple flat shapes with a little hatching, original designs with no logos or writing. Light: one clear direction. The lit side is pale, warm blocks and the shadow side is deeper flat tones. Colour: natural skin tones in warm terracotta, peach and deep umber, with one saturated accent colour. Texture: mostly clean, with small patches of halftone dots and fine hatching in the shadows. Background: one plain muted field, softly graded. No place and no objects. Composition: head and shoulders, close, filling most of the frame. Mood: cool, composed and quietly intense. Avoid: 3D render look, photorealism, soft airbrushed skin, visible brushwork, a busy background, real or famous faces, text, signatures.
Model recommendation
- I recommend GPT Image 2.5 on Sunburst. Every image here was made with it, from words alone, with no reference image.
- Keep the words "flat" and "hard-edged". They hold the look together.
- Say where the light comes from, like warm side light or stove flames from below. It decides where the pale and dark shapes fall.
- Give each person one strong colour, like a cobalt coat or a red bandana. It pops against the warm skin.
- Give people one memorable detail, like big gold hoops or a scar through an eyebrow.
- For a hand in frame, say how the arm gets there, like "the forearm rises from a bent elbow at the bottom of the frame".
- Use it for anyone. These people are just examples.
- My settings: Sunburst, high quality, 3:4 at 1536 × 2048, or square at 2048 × 2048.
Illustrated
Bold Print
Mid-century screen-print posters that tell one clever idea with a few big flat shapes, in two or three bold inks on cream paper.
Prompt · for GPT Image 2.5 Sunburst
Swap the highlighted scene for your own. The rest is the style, so leave it as it is.
Mid-century screen-print poster: one clever idea told with a few big flat shapes, in a few bold inks on cream paper. Scene: [Your idea in a few plain sentences: the main object, figure or place, what makes it clever, which colours lead, and any words in quotes. E.g. a giant black fountain pen stands upright on a signal red field, with its nib at the bottom. A pool of black ink spreads out from the nib and turns into the silhouette of a city skyline with tiny cream windows. Below it, the words "THE LAST WORD".] Idea: one single image carries the whole poster and stands for something bigger. Nothing is there just to fill space, and large areas stay empty. Shapes: everything is built from big, flat, hard-edged shapes with simple curves and straight lines, as if cut from coloured paper. Empty space does some of the drawing, so one shape can read as two things. Small details are repeated marks and thin parallel stripes. Colour: at most three flat inks on warm cream paper, drawn from signal red, near-black, mustard ochre, dusty rose and deep oxblood. No gradients, no shading and no extra colours. Faces and figures: stylised, with small heads and straight, tailored shapes. Each is a flat shape with very few features. Bodies stay in proportion with limbs that join properly, and any visible hand has five digits. Light: none is drawn. Any shadow is one flat, hard-edged shape in a darker ink. Words: only the words quoted in the scene, spelled exactly, in bold condensed capitals, with thin straight lines near them. Otherwise no text. Every design is original, with no logos or brand names. Texture: a fine, dry screen-print speckle across the inks, small flecks where the ink missed and slightly rough paper. The edges stay clean. Layout: a wide cream margin around a field of colour, with one clear focal point and bold, balanced empty space. Flat and frontal, with almost no depth. Mood: dry, witty, confident and a little ominous. Avoid: 3D render look, photorealism, gradients, soft shadows, painterly brushwork, outlines around shapes, clutter, tiny detail, real or famous faces, signatures.
Model recommendation
- I recommend GPT Image 2.5 on Sunburst. Every image here was made with it, from words alone, with no reference image.
- Start with the idea, then describe one big shape that tells it, like one red heel wedged between closing doors.
- Name your colours in the scene, like mustard, dark red and cream. It keeps each poster to a few inks.
- Put any words in quotes, and end the scene with "No words." if you don't want any.
- It can add small details you didn't ask for, so check for them.
- Use it for any idea, from a product to a film. These posters are just examples.
- My settings: Sunburst, high quality, 3:4 at 1536 × 2048, or square at 2048 × 2048.
Illustrated
Save Point
Early-1990s point-and-click game scenes in chunky, hand-painted pixels, with moody light and rich, detailed worlds.
Prompt · for GPT Image 2.5 Sunburst
Swap the highlighted scene for your own. The rest is the style, so leave it as it is.
Early-1990s point-and-click adventure game scene: moody, hand-painted worlds in chunky low-resolution pixels. Scene: [Your scene in a few plain sentences: the place, the time of day and the light, who is there and what they are doing. E.g. an adult woman with a grey bun, in a long yellow raincoat, stands knee-deep in the flooded lobby of an abandoned grand hotel at night, holding up a lantern. Moonlight falls through tall arched windows onto the dark water, a chandelier hangs above it, and a half-sunken grand piano lies at the foot of a curving staircase at the back.] Pixels: the whole image is drawn on one very low-resolution grid, about 240 pixels wide, then enlarged with hard edges. Every pixel is a clearly visible square of exactly the same size, including in the smallest details. No smoothing, no blur, no soft edges and no pixels of any other size. Painting: every surface is painted by hand, pixel by pixel, in clusters of colour that follow the form, like brushwork made of squares, and is rich in texture. Smooth changes of tone use dithering, with scattered and checkerboard patterns of two neighbouring colours. No black outlines around shapes. Colour: a limited palette of about 256 colours, with strong, slightly saturated steps from light to dark. One mood colour dominates each image. The deepest shadows are near-black. Light: clear sources, painted as pools and bands of light that step from colour to colour. Faces and figures: in true scale with the world around them, with natural proportions, simple clear silhouettes and faces of only a few pixels. World: believable, with real perspective and many small details, seen by a fixed camera from a wide angle, slightly above eye level. Every design is original, with no logos or brand names. Composition: only the game scene, filling the whole frame. No menus, inventory, icons, cursor, dialogue, captions or any other text. Mood: curious, quiet and mysterious, like the moment before something happens. Avoid: modern clean cartoon pixel art, isometric view, smooth gradients, blur, mixed pixel sizes, soft painterly edges, 3D render look, photorealism, scanlines, real or famous faces, characters from games or films, signatures.
Model recommendation
- I recommend GPT Image 2.5 on Sunburst. Every image here was made with it, from words alone, with no reference image.
- Say who is in the scene and what they are doing, like "an adult woman holding up a lantern".
- Pick one mood colour for each scene, like deep indigo night or hot red alarm lights. It sets the whole image.
- Say that signs are blank and gauges have no numbers, and end the scene with "No words."
- Give small figures something to hold, like a lantern or a flag. It keeps their hands easy to read.
- Use it for any place and time. These scenes are just examples.
- My settings: Sunburst, high quality, 3:4 at 1536 × 2048, or square at 2048 × 2048.
Illustrated
Full Volume
Flat 2D animation stills in loud tangerine, violet and hot pink, with thin ink lines, one-colour worlds and glowing gold.
Prompt · for GPT Image 2.5 Sunburst
Swap the highlighted scene for your own. The rest is the style, so leave it as it is.
Flat 2D cel animation still: simple, clean shapes in loud flat colour, thin black line, and one single colour for the whole world behind the figures. Scene: [Your scene in a few plain sentences: the place and which colour fills it, where we see it from, who is there and what they are doing, and any words in quotes. E.g. an ice-cream truck parked at a kerb, seen in three-quarter view from pavement height, so it looms overhead and leans, with apartment buildings behind it tilting away. The setting is tangerine orange, and the truck is electric violet and white with a hot pink awning. The vendor, a big man in his fifties with a grey moustache, a patterned short-sleeved shirt and a gold chain, leans far out of the serving hatch, grips a cone below the ice cream and holds it down toward a small dog, which sits on the pavement staring up at it. His other hand grips the edge of the counter. Painted along the truck's side in hot pink, the words "SOFT SERVE".] Setting: one flat colour from edge to edge, sky and ground included. Everything in it is drawn on that colour in thin black line, like a line drawing on coloured paper, and nothing in it keeps its natural colour. Some of its shapes are filled with hard flat blocks of the other two main colours, so colour punches through the line drawing. Colour: three loud colours lead: warm tangerine orange, electric violet and hot pink, with near-black and clean white. Clothes and objects use only these. Skin, in warm natural tones, and polished gold are the only exceptions. Drawing: every form is simplified into a few clean shapes with smooth edges and very few inner lines. Surfaces are smooth and unmarked. Hair and fur are solid shapes with a few flowing lines. Line: clean, thin black ink line of nearly even weight on every form. The solid blacks carry the weight. Blacks: deep shadows, gaps and undersides are solid near-black shapes with hard edges. Shading: the setting and objects have no shading, only hard black shadow shapes. Figures have one soft airbrushed shadow over their flat colour, and skin has a warm blush across the cheeks and nose. Shine: polished surfaces are saturated gold with hot white highlights and a soft glowing halo, with small four-point white sparkles beside them. They reflect nothing. Pattern: large surfaces carry crisp graphic patterns that follow their perspective: checkerboard, halftone dots and small repeated motifs. Air: open space carries thin white wavy lines and a few smooth floating blobs, thinly outlined and split into curved bands of orange, violet and pink, with no white in them. Perspective: a wide lens, theatrical and slightly tilted, with strong converging lines, so the place leans around the action. The nearest things run off the edges of the frame. Faces and figures: stylised, with long limbs, strong curves, small heads and real attitude. Faces have natural, handsome proportions, a thin contour, a nose drawn as one looping line, heavy black upper eyelids, thin arched brows, a scatter of freckles and lips in one dark flat tone with one small white highlight. Limbs join at real joints, and every visible hand has five long, slim digits. Words: only the words quoted in the scene, spelled exactly, in chunky hand-painted sign letters with a hard black drop shade. Otherwise no text. Every design is original, with no logos or brand names. Finish: clean, smooth animation colour. Mood: sunny, proud, funny and full of swagger, like a hot afternoon with music coming from somewhere. Avoid: 3D render look, photorealism, comic-book inking, cross-hatching, painterly brushwork, surface texture, real or famous faces, signatures.
Model recommendation
- I recommend GPT Image 2.5 on Sunburst. Every image here was made with it, from words alone, with no reference image.
- Name the colour of the setting in your scene, like "the setting is hot pink". Each image here has its own.
- Give the scene one small human moment, like a whisper behind a hand. This style lives on attitude.
- Give people something polished to wear, like gold hoops or round glasses. The glow shows best in a close-up.
- Put any words you want in quotes in the scene. Without quotes, the image stays free of text.
- Use it for any place and any people. These scenes are just examples.
- My settings: Sunburst, high quality, 3:4 at 1536 × 2048, or square at 2048 × 2048.