GPTPrompts.AI

SD
Prompts.

Own the local model. Our master guide reveals the advanced Stable Diffusion prompts that unlock unprecedented control over photorealism, artistic style, and anatomical accuracy.

Recommended toolInVideoTurn photos, scripts and ideas into finished videos with AI.Make a video

Affiliate link, we may earn a commission at no extra cost to you.

The Open Source Edge: Why Stable Diffusion Matters

Unlike DALL-E or Midjourney, Stable Diffusion gives you the keys to the engine. It's a tool for creators who demand absolute control. However, that control comes with a steep learning curve. Mastering Stable Diffusion prompts requires an understanding of token weighting, negative embedding spaces, and how different checkpoints interpret your words.

This guide explores the specific syntax that sets SD apart. We cover how to use parentheses for emphasis, the critical role of the negative prompt box, and how to integrate LoRAs to achieve consistent characters and styles across multiple generations.

ControlNet Mastery

Structural Sovereignty

ControlNet is the ultimate plugin for Stable Diffusion. It allows you to use reference images to define the exact pose, depth, or edges of your generation.

Canny Edge Detection:

"Use a sketch of a building as a Canny reference. Prompt: 'Hyper-modern glass skyscraper, sunset lighting, architectural photography, 8k.' The model will follow the lines of your sketch exactly."

OpenPose:

"Upload a photo of a person dancing. Use OpenPose to extract the skeleton. Prompt: 'Cyberpunk cyborg performing a ritual dance, neon trails, cinematic smoke.' The cyborg will match the dancer's pose."

Advanced Photorealism Syntax

To achieve 'Pore-level' detail in Stable Diffusion, you must prompt for the process of photography, not just the result.

The Realism Stack:

"(RAW photo:1.2), portrait of a [SUBJECT], (highly detailed skin:1.3), 8k uhd, dslr, soft lighting, high quality, Fujifilm XT4, 85mm lens, f/1.8, (subtle skin imperfections:1.1), (natural lighting:1.2)"

Note: The weights (e.g. :1.3) are critical for preventing the 'Plastic AI Skin' look.

50+ Ready-to-Run SD Prompts

Copy these directly into Automatic1111, ComfyUI, Forge, or Fooocus. Weights and LoRA placeholders use common SDXL / SD 1.5 syntax. Adjust CFG 6–9 and 28–35 steps for most.

Photorealism, portraits

(RAW photo:1.2), half-body portrait of a 32-year-old woman with freckles and auburn hair, (detailed skin texture:1.3), overcast studio lighting, Fujifilm XT4, 85mm f/1.8, shallow depth of field, color-graded teal-and-orange

cinematic close-up of a weathered fisherman, wet salt-sprayed skin, backlit golden hour, volumetric sea mist, Kodak Portra 400 film grain, (skin imperfections:1.15), 8k

editorial portrait of a dancer mid-leap in a black studio, rim light from behind, long exposure motion blur on fabric, (sharp face:1.2), Hasselblad medium format look

(polaroid photo:1.2) of a teenage skateboarder on an empty suburban street, 1998 aesthetic, slight light leaks, washed cyan shadows, grainy film stock

African elder in traditional indigo robe, studio black backdrop, three-point Rembrandt lighting, ultra-sharp 105mm portrait lens, (dignified expression:1.2)

corporate headshot of a product manager, neutral grey seamless, soft key + fill, natural smile, sharp focus on eyes, shot on Sony A7R V, professional retouch

low-key portrait of an elderly jazz pianist, single Fresnel spotlight, smoke curling, 80s noir aesthetic, shallow DOF, tri-X film grain

Photorealism, scenes & product

top-down flat lay of a minimalist breakfast spread, linen napkin, matte ceramic, morning window light, (sharp highlights:1.1), 50mm food photography

architectural photograph of a concrete brutalist library, dramatic side light, long shadows, wide-angle 16mm, Fuji Velvia saturation

product shot of a black leather wallet on polished marble, reflective surface, softbox above, (crisp edges:1.3), commercial catalog style

aerial drone photo of a winding coastal road at dawn, turquoise water, limestone cliffs, slight haze, Mediterranean palette

macro photograph of morning dew on a spider web, backlit sunrise, golden bokeh, 100mm macro lens, (extreme detail:1.3)

hyperreal render of a sneaker floating in mid-air, studio white cyc, hard rim light, crisp reflections, commercial campaign feel

Anime & illustration

1girl, silver-haired mage in twilight forest, casting blue runes, anime key visual style, detailed eyes, Makoto Shinkai lighting, volumetric god rays, <lora:niji-style:0.7>

shonen-style battle scene, two swordsmen mid-clash, impact lines, dust explosion, ink-brush texture, (dynamic pose:1.3), cel shading

soft pastel slice-of-life, girl in summer dress watering plants on a balcony, Ghibli-inspired, warm afternoon light, muted color palette

cyberpunk anime cityscape, neon kanji signs, rain-slick streets, chromatic aberration, 90s OVA aesthetic, <lora:akira-style:0.6>

chibi sticker-style mascot of a frog wizard, thick outlines, flat vector look, transparent background, mobile game UI asset

watercolor anime portrait, gentle gradient washes, visible paper texture, dreamy pastels, minimal line art

Fantasy & sci-fi concepts

epic matte painting of a floating island city above a storm, dragons circling, god rays through clouds, Greg Rutkowski style, trending on ArtStation, 8k

dark fantasy knight standing before a giant cathedral at dusk, volumetric fog, intricate armor engraving, painterly brushstrokes, Frazetta-inspired

alien jungle biome, bioluminescent flora, massive coiled creature sleeping, atmospheric depth, concept art for film, wide cinematic frame

derelict space station interior, flickering emergency lights, zero-gravity debris, tense horror mood, Alien (1979) film aesthetic

elven archer on a cliff at sunset, flowing cape, ancient forest valley below, (hyper-detailed armor:1.2), fantasy book cover composition

cosmic horror deity emerging from a black lake, impossible geometry, Lovecraftian, deep blues and sickly greens, oil painting on canvas

Architecture & environments

isometric miniature village, tilt-shift lens effect, soft daylight, toy-like scale, matte finish, pastel palette

Japanese ryokan courtyard in autumn, koi pond, raked gravel, maple trees at peak red, soft overcast, (serene atmosphere:1.2)

Scandinavian cabin interior, wool throws, wood stove, falling snow outside, warm amber interior glow, cozy lifestyle photography

futuristic Tokyo street at night, translucent holographic ads, reflective puddles, Blade Runner 2049 palette, cinematic lens flare

desert Bedouin tent at dusk, warm lantern glow, star-filled sky, Milky Way visible, (long exposure:1.1)

abandoned Victorian greenhouse, overgrown with ivy, shafts of light through cracked glass, painterly mood

Portraits, stylized

renaissance oil painting of a modern programmer holding a laptop, Rembrandt lighting, ornate gilded frame composition, rich umber palette

vaporwave portrait of a 1980s secretary, pastel pink + cyan gradient, grid horizon, statue bust fragments, retro-futurist

Art Nouveau poster of a botanist, Alphonse Mucha style, ornate floral border, flat pastel color fills, decorative typography space

low-poly 3D render of a medieval blacksmith, flat shaded, pastel palette, isometric angle, indie game concept art

pencil sketch portrait on toned paper, loose gestural hatching, white chalk highlights, classical art-school study

ukiyo-e woodblock print of a modern skateboarder, Hokusai style, flat color fields, bold line work

Negative-prompt templates

NEGATIVE: (worst quality, low quality:1.4), jpeg artifacts, blurry, out of focus, deformed, extra fingers, extra limbs, mutated hands, bad anatomy, watermark, signature, text

NEGATIVE (portrait): plastic skin, wax figure, over-smoothed, airbrushed, doll-like, crossed eyes, lazy eye, bad teeth, uncanny valley

NEGATIVE (anime): bad hands, extra fingers, fused fingers, missing fingers, mutated, distorted face, low detail, pixelated, bad eyes, asymmetric eyes

NEGATIVE (realistic): cartoon, anime, illustration, painting, 3d render, cgi, over-saturated, HDR halo, low resolution, noise

NEGATIVE (product): clutter, busy background, multiple subjects, text, logo, reflection of photographer, lens flare, motion blur

LoRA + ControlNet combos

portrait of a cyberpunk ronin, <lora:cyberpunk_edgerunners:0.8> <lora:detail_tweaker:0.4>, ControlNet: OpenPose from reference image, Canny weight 0.5

isometric city block, <lora:pixel-art-xl:0.7>, ControlNet: Scribble from hand sketch, perfect tile boundaries

fashion editorial with specific pose, <lora:add_detail:0.5>, ControlNet: OpenPose + Depth, high fidelity to source

architectural rendering, <lora:archviz:0.6>, ControlNet: MLSD to preserve geometry from line art reference

character sheet with consistent face, <lora:consistent_character:0.7>, ControlNet: Reference-only + IP-Adapter Face

Specialty / experimental

double-exposure photograph of a wolf silhouette filled with a snowy forest scene, high contrast, fine grain, monochrome

cross-section architectural illustration of a submarine, technical blueprint lines, annotations, white-on-blue, vintage engineering poster

liquid mercury sculpture in the shape of a running horse, studio white background, hyper-reflective, (sharp caustics:1.2)

storybook illustration of a fox family on a picnic, gouache texture, cozy warm palette, thick white border

surreal Dali-inspired landscape, melting clocks over a desert, elongated shadows, oil on canvas

Stable Diffusion FAQ

Don't stop here

What to read next

Hand-picked guides our readers explore right after this one.