Beginner's Guide to AI Image Prompting

Master the art of AI image prompting with our beginner-friendly guide! Learn the do's and don'ts for better, more accurate results every time.

Note:

The GPT Image and GPT Image HD models strictly do not allow generating images of real people (including public figures like “the president of the USA”) or licensed properties (like “a Disney character”). Even broad prompts referencing these can be rejected automatically—these models are especially strict about these limitations.

Dos and Don'ts of Text to Image Prompting

DO ✅	DON'T ❌
✅ Describe what you want to see directly Good: "snow-covered mountain peak at sunrise" "a red rose in a crystal vase" "man in a business suit"	❌ Don't use commands or instructions in natural language Bad: "generate a snow-covered mountain peak at sunrise" "create a red rose in a crystal vase" "make me a picture of a man in a business suit"
✅ Completely omit unwanted elements or use alternatives Good: "professional headshot of a middle-aged man" "beautiful landscape in summer" Exception: In the SD mode, you can input a negative prompt, separate from the main prompt, which guides the AI on what not to generate.	❌ Don't include negation words such as "no", "not" and "without" Bad: "man without beard" "landscape not in winter"
✅ Take advantage of AI Improve Tap the "AI Improve" button under the prompt box to instantly turn even a short and simple description into a professional-grade prompt.	❌ Don't spend hours struggling to create a perfect prompt Creating good prompts can be challenging and time-consuming, especially in the beginning. AI Improve feature saves time and lets you see gaps in your descriptions, helping you achieve better results in the future.
✅ Be clear and specific Good: "a sunset over rolling green hills, with a lone tree silhouetted against the sky" "a young woman with curly red hair, wearing a yellow dress, smiling while holding a sunflower bouquet"	❌ Don't be vague Bad: "a nice view" "a person smiling"
✅ Use adjectives and style references Good: "a futuristic knight in sleek, metallic armor, standing on a dusty Martian landscape with a glowing red sunset" "serene fairytale castle, softly lit by golden light, surrounded by misty forests, in the style of an Impressionist painting"	❌ Don't overload your prompt with random elements Bad: "a futuristic knight in medieval armor, on Mars, with unicorns, robots, and dragons, during a thunderstorm" "a fairytale castle surrounded by forests, with aliens, a futuristic city skyline in the background, a vintage car parked outside, and an astronaut floating nearby"

DO ✅

DON'T ❌

✅ Describe what you want to see directly

Good:

"snow-covered mountain peak at sunrise"

"a red rose in a crystal vase"

"man in a business suit"

❌ Don't use commands or instructions in natural language

Bad:

"generate a snow-covered mountain peak at sunrise"

"create a red rose in a crystal vase"

"make me a picture of a man in a business suit"

✅ Completely omit unwanted elements or use alternatives

Good:

"professional headshot of a middle-aged man"

"beautiful landscape in summer"

Exception: In the SD mode, you can input a negative prompt, separate from the main prompt, which guides the AI on what not to generate.

❌ Don't include negation words such as "no", "not" and "without"

Bad:

"man without beard"

"landscape not in winter"

✅ Take advantage of AI Improve

Tap the "AI Improve" button under the prompt box to instantly turn even a short and simple description into a professional-grade prompt.

❌ Don't spend hours struggling to create a perfect prompt

Creating good prompts can be challenging and time-consuming, especially in the beginning. AI Improve feature saves time and lets you see gaps in your descriptions, helping you achieve better results in the future.

✅ Be clear and specific

Good:

"a sunset over rolling green hills, with a lone tree silhouetted against the sky"

"a young woman with curly red hair, wearing a yellow dress, smiling while holding a sunflower bouquet"

❌ Don't be vague

Bad:

"a nice view"

"a person smiling"

✅ Use adjectives and style references

Good:

"a futuristic knight in sleek, metallic armor, standing on a dusty Martian landscape with a glowing red sunset"

"serene fairytale castle, softly lit by golden light, surrounded by misty forests, in the style of an Impressionist painting"

❌ Don't overload your prompt with random elements

Bad:

"a futuristic knight in medieval armor, on Mars, with unicorns, robots, and dragons, during a thunderstorm"

"a fairytale castle surrounded by forests, with aliens, a futuristic city skyline in the background, a vintage car parked outside, and an astronaut floating nearby"

Tip:

Want to know even more? Check out our advanced prompting guide for additional tips and tricks!

Essential Keywords for Text to Image Prompts

General

Category	Keywords
Mood and atmosphere	uplifting, eerie, serene, dramatic, melancholic, joyful, mysterious, peaceful, intense, whimsical, foggy, dreamy, surreal, smoky, vibrant, nostalgic, tranquil, ominous, cinematic, magical
Textures	smooth, rough, gritty, soft, shiny, matte, glossy, velvety, coarse, delicate
Composition	balanced, asymmetrical, minimal, crowded, dynamic, layered, symmetrical, negative space, rule of thirds
Action	still, fast-paced, frozen, dynamic, flowing, chaotic, peaceful
Colors	vibrant colors, muted, warm tones, cool hues, pastel shades, neutrals, earthy colors, monochrome, rich tones, bold contrasts, soft gradients, deep saturation, subtle highlights, faded palette, vibrant colors

Photography

Category	Keywords
Resolution and quality	UHD, 8K, 16K, RAW, high detail, hyper-realistic, photorealistic, finely detailed
Lighting	cinematic lighting, soft lighting, golden hour, backlit, natural light, studio lighting, dramatic shadows, rim lighting, volumetric lighting
Lens	wide-angle, ultra-wide, macro lens, telephoto lens, zoom lens, fisheye lens, prime lens, portrait lens
Effects	lens flare, motion blur, tilt-shift, color-graded, HDR, long exposure, candid, bokeh effect, black and white, vintage film, sepia
Shot types	close-up, extreme close-up, medium shot, long shot, wide shot, extreme wide shot, over-the-shoulder shot, bird’s-eye view, worm’s-eye view
Device	shot on DSLR, captured on iPhone 15 Pro, taken with a mirrorless camera, photo from a GoPro, drone photography, shot on DJI Mavic, Polaroid photo, webcam photo, captured on security camera
Camera settings	f/1.8, f/2.8, f/5.6, f/11, f/16, ISO 100, ISO 400, ISO 800, ISO 1600, ISO 3200, 1/60 shutter speed, 1/250 shutter speed, 1/1000 shutter speed

Cartoons

Category	Keywords
Styles	flat design, cel-shaded, 2D animation, rubber hose style, comic book art, children’s book illustration, Western cartoon style, retro cartoon style
Line work	bold outlines, thick black lines, clean line art, sketchy strokes, hatching, minimal line work, smooth curves
Character designs	exaggerated expressions, simple geometric shapes, anthropomorphic animals, comically large features

Anime

Category	Keywords
Styles	chibi, shonen, shojo, seinen, isekai style
Line work	fine details, clean line art, dynamic motion lines, delicate contours, soft shadows with hatching, stylized hair strands
Character features	large expressive eyes, spiky hair, intricate costumes, flowing hair, exaggerated emotions, detailed armor, sleek mecha designs

Text

Category	Keywords
Style	Classic: serif, Times New Roman, traditional, elegant, formal Modern: sans-serif, sleek, minimalist, geometric, clean Vintage: distressed, retro, weathered, 70s-inspired, grunge, old-school Decorative: ornate, embellished, script, calligraphy, swirly Futuristic: tech-inspired, glitchy, sci-fi, sharp edges, holographic Artistic: abstract, brush strokes, graffiti, doodle-style, watercolor Gothic: blackletter, medieval, sharp serifs, dramatic
Size	Large: oversized, billboard style, bold, statement text, towering letters Small: delicate, fine print, subtle, compact, tiny letters Dynamic: fluctuating size, exaggerated perspective, tapering edges Blocky: uniform size, block lettering, monospaced, heavy-weight
Material	Metallic: gold, silver, bronze, chrome, reflective steel, iron Organic: woodgrain, stone-carved, mossy, clay, leaf-textured Liquid: water droplets, melted wax, ink splashes, flowing lava Glass: frosted glass, transparent, stained glass, shattered effects Fabric: stitched, embroidered, denim-textured, patchwork Plastic: shiny, translucent, neon acrylic, molded plastic Natural elements: sand, ice, fire, clouds, smoke
Effects	3D: extruded letters, beveled edges, dynamic shadows, layered depth Shattered: fragmented, cracked, broken pieces, jagged shards Glitch: distorted, pixelated, fragmented lines, digital noise Textured: rough, grainy, embossed, tactile patterns

Traditional art

Category	Keywords
Style	lifelike, geometric patterns, dreamlike, cubist, angular shapes, layered perspectives, ornate, high contrast, classical proportions
Texture	visible brush strokes, thick brush strokes, layered paint, smooth blending, rough lines, smudged edges, gradient washes, bleeding edges, coarse canvas texture, visible fibers, stretched fabric
Medium	oil paint, acrylic, watercolor, gouache, pencil, ink, charcoal, pastel, mixed media, woodcut, lithography

‎ Now that you've learned the fundamentals, it's time to start experimenting with your own prompts. Feel free to explore and refine your approach!

Dos and Don'ts of Text to Image Prompting
Essential Keywords for Text to Image Prompts