The best Stable Diffusion prompts: 46 ready-to-use examples

The best Stable Diffusion prompts: 46 ready-to-use examples

Table of Contents

Stable Diffusion belongs to “text2image” neural networks—in other words, networks that generate images from text. Such text-based tasks for AI are called “prompts.” On this page, you will learn everything about Stable Diffusion prompts and also find 46 of the most popular ones, which you can use to create amazing images.

What is a prompt

A prompt for a neural network is a short text description of exactly what the AI should depict. Based on this text description, the neural network understands your request. Prompts can be partially compared to a search query, except that with AI tools, you describe the task in greater detail using a specific structure. For example:

barack obama, wearing gray suit, forest in the background, hyper realistic painting

This example shows the basic structure:

  • Who
  • What they are doing and where

  • The style of the image

    In addition to positive statements, you can interact with a neural network in reverse—that is, by specifying what should not appear in the image. This is called a negative prompt (Stable Diffusion Negative Prompt). It works in SD 1.5, SDXL, and most fine-tuned models, while Flux does not use negative prompts. For example, if you want to generate a photorealistic image, add unwanted styles to your Stable Diffusion negative prompts:

computer graphics (2D, 3D)

Stable Diffusion is not unique in the way it handles tasks—most other neural networks also work with prompts, meaning text-based tasks. However, the rules for writing prompts differ between neural networks, since there is no single universal format. To ensure that the neural network understands you as accurately as possible, you need to describe the task in as much detail as possible. For example, if you simply write “

Car

,” the neural network will have too much room for creativity, and you will get a completely random car. The term “Prompt” is an English loanword and is written in the original as “Prompt.” You can remember this term because it will help you find ready-made prompts in the English-speaking segment by using the query “Stable Diffusion Prompts.” An alternative to prompts is image-based tasks. In other words, this is when AI generates an image from another image rather than from a text description. In this case, the neural network is classified not as text2image (like Stable Diffusion), but as image2image.

We have collected the most popular prompts that will help you not only generate images but also learn how to write your own prompts—with the same logic and level of detail.

Important: the same prompt does not generate identical results every time. Your results may differ from ours. Moreover, if you are not satisfied with the result, you can always regenerate the image using the same settings or change the parameters and generate it again.

Which models are they for: the prompts below are written in the comma-separated tags and key phrases format familiar to SD 1.5 and models based on it. They also work with SDXL and its finetunes (Juggernaut XL, RealVisXL), but the result will be different. Pony, Illustrious, and NoobAI require their own tags, while SD 3.5 and Flux understand regular sentences better. Read the article about prompt generators to learn how to write prompts for each model, and the model guide will help you choose a model.

1. Photorealistic fashion art image

Photorealistic Fashion Art image Create a photorealistic magazine-cover image using the following description:

Prompt**:** beautiful face, beautiful model body, intricate, sharp focus, natural lighting, RTX, professionally color graded, professional photography, masterpiece, ultra-detailed, high quality, best quality, 4k, 8k, RAW, fashion, high fashion, haute couture

Negative prompt**:** ugly, duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, bad anatomy, bad body proportions, extra limbs, cloned face, disfigured, gross body proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, mutated hands, fused fingers, too many fingers, long neck, noise, artifacts, dithering, poorly drawn hair, watermark, signature, amputee, low quality, low-res, low-poly, worst quality, nude, naked, nsfw

2. Realistic low-poly 3D render

Realistic Low-Poly 3D render

  If you need to generate an image in the style of a realistic 3D render, you can use this example prompt in a low-poly style.

Prompt**:** Realistic 3D render of an isometric hospital in a low poly art style, medical workers, beds and equipment

3. Portrait of an eastern beauty

Portrait of an Eastern beauty You can create a realistic portrait of an Eastern beauty using the following prompt:

Prompt**:** a beautiful Middle East girl side view, daylight, vibrant colors, photographic image, high-quality portrait, detailed photograph

4. Girl with her tongue sticking out

For this to work correctly, you will need the Awaiting Tongue add-on, since the base model does not always render a sticking-out tongue correctly. Use

Sampling method – DPM++ 2M Karras

and the following description:

Prompt**:** (awaitingtongue-sd-v1-800:1.1), photo of a young woman, (extreme closeup mouth shot) of a young woman in robes (with her eyes closed:1.2) with short hair, National Gallery of Art Highlights, sunset, staring at the viewer, antview, bokeh, closeup of face, highly detailed, 8K, stunning, hdr, subsurface scattering, global illumination, film still, Film-like, bokeh, 3d, vivid color/golden hour/raytraced/warm colors, (caucasian woman), (perfect teeth), (extreme closeup of face:1.4)

5. Beautiful girl with devil horns

Beautiful girl with devil horns To generate an image of an attractive red-haired girl with devil horns, use the following:

Prompt**:** masterpiece, best quality, upper body, 1girl, looking at viewer, red hair, medium hair, purple eyes, demon horns, black coat, indoors, dimly lit

6. Neon-punk city

Neon-punk city Create a breathtaking neon-punk cityscape with the following prompt:

Prompt**:** Beautiful city after rain, water over the roads, street lights fading away, street view, highly realistic and aesthetic view, beautiful landscape, wallpaper, high definition image, high quality

7. High-resolution realistic photo

High-resolution realistic photo You can generate a super-realistic high-resolution photo by adding the words “

Raw photo

” at the beginning of the prompt. You can adapt our example by changing the description:

Prompt**:** RAW photo, a dragon girl European appearance, full height, (high detailed skin:1.2), dslr, 8k uhd, soft lighting, high quality, film grain, Fujifilm XT3

8. Disney-style animated character

Disney-style animated character Have you always dreamed of creating a Disney-style character? The following prompt will give you a unique character:

Prompt**:** a wholesome animation key shot of a band behemoth performing on stage, medium shot, pixar and disney animation, studio ghibli, 3 d, sharp, rendered in unreal engine 5, anime key art by greg rutkowski, bloom, dramatic lighting

9. Old European city

Old European city Create an authentic high-resolution image of an old European city. The resulting image can also be used as desktop wallpaper:

Prompt**:** Old beautiful city after rain, yellow street lights fading away, rain water over the roads, aesthatic vibes, beautiful top street view, midnigh vibes, empty roads and yellow street lights, high quality high definition image, highly realistic aesthatic and viby

10. Realistic cinematic characters

Realistic cinematic characters Create a unique cinematic image featuring various realistic characters:

Prompt**:** cinematic, cinematic lighting, 8K raw photo, best quality, masterpiece, ultra-high res, realistic, digital painting, photography, vibrant, intricate details, high-definition, detailed, sharp focus

Negative prompt**:** text, extra fingers, fewer fingers, strange fingers, bad anatomy, missing fingers, bad hands, extra hands, more than two hands, more than two legs, unclear eyes, missing arms, mutilated, extra legs, extra limbs, cloned face, broken hand, fused fingers, paintings, sketches, worst quality, low quality, normal quality, low-res, skin spots, acne, skin blemishes, age spots, glans, mutated hands, distorted hands

11. Computer game-style characters

Computer game-style characters Here is another prompt for creating characters, but this time not in the Disney style, but in the style of computer games. Use the following description:

Prompt**:** epic video game key visual, RPG style, masterpiece, illustration, best quality, detailed, intricate, 8k, HDR, wallpaper, cinematic lighting, sharp focus, fantasy,

12. Futuristic kitchen interior

Futuristic kitchen interior Create a unique futuristic kitchen image using the following prompt:

Prompt**:** beautiful open kitchen in the style of elena of avalor overlooking aerial wide angle view of a solarpunk vibrant city with greenery, interior architecture, kitchen, eating space, rendered in octane, in the style of Luc Schuiten, craig mullins, solarpunk in deviantart, photorealistic, highly detailed, Vincent Callebaut, elena of avalor, highly detailed, –ar 16:9

13. Cat girl

Cat girl The following description will generate an image of a cat girl in the Lushill style:

Prompt**:** lushill style, mystical, transparent, ghost cat of the milky way, Trending on Artstation, {creative commons}, fanart, AIart, {Woolitize}, by Charlie Bowater, Illustration, Color Grading, Filmic, Brenizer Method, Side-View, Perspective, Depth of Field, Nikon D750, Field of View, F/2.8, Lens Flare, Tonal Colors, 8K, Full-HD, ProPhoto RGB, Perfectionism, Rim Lighting, Soft Lighting, Accent Lighting, Natural Lighting, Diffraction Grading, With Imperfections, insanely detailed and intricate, hypermaximalist, elegant, ornate, super detailed

14. Cyberpunk

Cyberpunk Anyone who appreciates Cyberpunk and images in this style should try the following prompt, which creates stunning images:

Prompt**:** wide angle action shot of a purple skyline in beautiful cyberpunk city, highly textured, highly detailed lofi anime comic art, cyberpunk 2077, steampunk elements, dark art, ethereal art, dramatic ray-traced lighting, cyberpunk environment, lofi art v4

15. Photorealistic portrait of a beautiful girl

Photorealistic portrait of a beautiful girl Stable Diffusion is excellent at generating photorealistic portraits of girls when given the right prompt. For example, you can use the following:

Prompt**:** photograph portrait of a MilBob3vLR, textured skin, brown wavy hair, sports bra, cleavage, leggings, clit, morning sunlight, perfect eyes, kodakvision color, shot on Arricam LT Camera, sharp focus on subject, bokeh, shot by Nick Brandt

Negative prompt**:** 3d render, long neck, out of frame, extra fingers, mutated hands, monochrome, ((poorly drawn face)), ((poorly drawn hands)), (((mutation))), (((deformed))), ((ugly)), blurry, ((bad anatomy)), (((bad proportions))), ((extra limbs)), (((long neck))), (child), (childlike)

16. Ariel image (animation)

Ariel image (Animation) To create different images of Ariel (the animated character), you will need

LoRA – Ariel_character

as well as the following prompt:

Prompt**:** Masterpiece, high quality, high resolution, (Very Feminine detailed face, detailed eye:1. 2), Disney little mermaid, 17 years old, Ariel from little mermaid, full body, sexy pose. Swimming, under the sea background, (mermaid:1. 27), <lora:Ariel_character:0. 5>, 16k, UHD, HDR, (Masterpiece:1. 5), (best quality:1. 5),

17. Korean girl with cat ears

Korean girl with cat ears For this to work correctly, you will need

LoRA – NijiExpressV2

after which you can create an image of a beautiful Korean girl with cat ears:

Prompt**:** (1girl, full body), beautiful Korean woman, (Cute Loose Bob hair:1. 5), (cat ears:1. 5), (red lips), (small breasts), (eyelashes, aegyo sal), (flowing hair, bright smile), (bedroom, immersive background), (volumetric haze), (global illumination, dynamic lighting, best illumination, volumetric lighting), (detailed face, beautiful detailed eyes, symmetrical eyes), (hyperrealism, photorealism, realistic, photo-realistic), (photon mapping, physically-based rendering),

18. Portrait of a couple in love

Portrait of a Couple in Love If you want to create a photo of a couple in love, use the following:

Prompt**:** Couple portrait, (man and woman:1. 3) in rags, masterpiece, high quality, solo focus, whole body, at a high angle, (perfect face: 1. 1) complex (great detail: 1. 1) body, closed navel, closed, extremely detailed eyes and face, eyes up, piercing gaze, big eyes, forest in the background, dense forest, butterflies, fireflies, Nikon d850 4 kodak portra 400 camera, f1. 6 lens, rich colors, hyperrealistic texture, dramatic unreal engine lighting is trending on Artstation Synestill 800 <lora:more_details:0. 2> <lora:epi_noiseoffset2:0. 3>

19. Birds

Birds The neural network works well not only with people, but also with birds. With the following prompt, you can generate a colorful bird image.

Prompt**:** a close up of a bird on a branch, he wears a blue shirt, cryptidcore, inspired by Louis Anquetin, beautiful oval face, with a big parrot, featured on pexels, outrageously fluffy, long nose, blue tinted, adoptable, cages

20. Abandoned city

Abandoned City You can create a frightening image of a little girl in an abandoned city using the following prompt:

Prompt**:** Ruined city, nuclear power plant, (creepy atmosphere:1. 32), single_girl, standing with (her back to the viewer:1. 0), head looking at the viewer, (dark atmosphere:1. 32), yellow jacket with a hood, hood behind her head, hunting backpack, radiation fund, 8k, UHD, HDR, (Masterpiece:1. 5), (best quality:1. 5) trending on artstation, volumetric lighting, octane render, Photograph, 8k, UHD, HDR, (Masterpiece:1. 5), (best quality:1. 5),

21. Nordic blonde girl

Nordic Blonde Girl Create a beautiful blonde aged 18–22 using the following prompt:

Prompt**:** A fairy trapped in an empty jar of jam, (photo realistic), extremely detailed and delicate illustration, beautiful and aesthetic, official art, single girl, ((glowing light purple and light pink theme, white theme, fairytale theme)), dream-like, (big jar is on the table, tiny fairy is in the jar), bioluminescent, glowing moon flowers, firefly, lowkey, dark shot, depth, detailed shade and lighting, Fairy wings are in the shoulder blade position, (effect: glowing firefly neon lights) [[(Masterpiece, Best Quality, High Quality:1. 4), very pretty look face, very pretty eyes, lovely images, cute images, (whole full body), solo, beautiful, front, lovely, adorable, pale skin, (18to 22 years old German girl, (full body:1. 4), glowing eyes, look beautiful German girl and blue eyes or green eyes with platinum blonde hair color), Nordic German young girl, (massive chest:1. 4), (wide hips:1. 4), professional drawings, extremely detailed CG unity 5669k80p Ultra QD_OLED wallpaper clear png image)]]

22. Young Samurai

Young Samurai Anyone interested in Japanese culture and samurai will appreciate the result generated by the following prompt:

Prompt**:** Young Samurai with long black hair and ponytail, wearing samurai armor, standing in a Japanese temple, cherry blossom petals falling, illuminated lanterns, glowing sunrise, tranquil lighting, ethereal lighting, delicate shadows, ((beautiful detailed eyes, symmetrical eyes), dramatic lighting, (photorealism:1. 5), (photorealistic:1. 4), (8k, RAW photo, masterpiece), High detail RAW color photo, a professional photo, realistic, (highest quality), (best shadow), (best illustration), ultra-high resolution, highly detailed CG unified 8K wallpapers, physics-based rendering, photo, realistic, realism, high contrast, hyperrealism, photography, f1. 6 lens, rich colors, hyper-realistic lifelike texture, cinestill 800)

23. Space rangers

Space Rangers You can create a beautiful girl in the style of space rangers with the following prompt:

Prompt**:** An intergalactic cute girl space ranger, space clothes, (Seductive Alluring expression:1. 1), Perfect eyes, Perfect Iris, another Galaxy. Sci-Fi, Retro, Acrylic art by Artgerm and Gediminas Pranckevicius, Laura Hollingsworth. intricate, highly detailed, crispy quality, dynamic lighting, hyperdetailed and realistic. , 8k, UHD, HDR, (Masterpiece:1. 5), (best quality:1. 5)

24. Cat

Cat AI can create a cat of any breed, such as a British Shorthair, using the following prompt. You can replace the breed in the first word.

Prompt**:** British Cat, masterpiece, (photorealistic:1.3), best quality, ultra-high res

25. Prehistoric world

Prehistoric World Create a realistic image of the prehistoric world with a prompt like this:

Prompt**:** real photo of prehistoric ocean, (Liopleurodon:1.2), Ammonite, Nothosaurus, Trilobites, Anomalocaris, Opabinia, Australia Hornstone, a lot of small fish, coral, seabed, (with sunlight streaming through the water:1.5), soft light, by Chen Daofuby, dramatic lighting, highly detailed

26. Realistic landscapes

Realistic Landscapes The neural network also handles nature well. For example, with the following prompt, you can generate a photorealistic landscape illuminated by sunset rays.

Prompt**:** Forest with green leaves at sunset, snowy mountain on background, high-fantasy. professional canvas by greg rutkowski, joe madureira, jody bergsma and boris vallejo, triadic colors, deep color, dynamic lighting, volumetric lighting, shading with dark edges, richly detailed, matte background, enameled glass, style octane render

27. Sea

Sea Stable Diffusion does not always create the sea realistically, but after several generations, you can obtain a photorealistic image. Example Stable Diffusion prompt for creating the sea:

Prompt**:** a painting of a beach with waves crashing on it and a sunset in the background with a house on the hill, Alena Aenami, extremely detailed oil painting, an oil painting, american scene painting

Negative prompt**:** people, human, signature, watermark, sign

28. City

City You can create a romantic image of any city by replacing the name in our example with your own. The result will be an atmospheric image of an autumn city in rainy weather.

Prompt**:** autumn in London, city light, market, beautiful atmosphere, rain by thomas kinkade

29. Past centuries

Past Centuries The neural network works not only with contemporary images but also creates historical ones. For example, you can create a realistic image of an 18th-century city using the following prompt:

Prompt**:** town architecture, river, 18 century, highly detailed, hyper-detailed architecture, photorealistic, octane render, unreal engine, digital painting hyperdetailed 8k

30. Futurism

Futurism Just as it handles past centuries, the neural network can also create cities of the future. The following prompt creates an image of a futuristic city:

Prompt**:** Ethereal gardens of marble built in a shining teal river in future city, gorgeous ornate multi-tiered fountain, futuristic, intricate elegant highly detailed lifelike photorealistic realistic painting, long shot, studio lighting, octane render, by Dorian Cleavenger

31. Male portrait

Male Portrait Statistics show that AI is used more often to create female characters, but you can also generate a male character. Use a prompt like this:

Prompt**:** ((realistic)), cinematic shot, movie theater, photo referenced, highest quality, high quality, (detailed face and eyes:1.1), sinny goth guy, victorian clothes, suspenders, short hair, makeup, black lips, glasses, inside an old mansion, subsurface scattering, intricate (high detail), ((cute)), (detailed skin:1.1), shiny skin, professional photo, photorealism, volumetric lighting

32. Interior

Interior You can create an image of any room, such as a bedroom, living room, bathroom, and so on, by replacing “ bedroom ” in our prompt with the name of the room you need:

Prompt**:** a futuristic bedroom with large curved ceiling, high windows, (futuristic decor: 1), looking out to a far future cityscape, sunny outside, scifi, (neon cyberpunk light:1.0), blade runner, (hologram:1.4)

33. Manga

Manga Stable Diffusion can generate various types of anime, including manga. The following prompt will produce 2D anime:

Prompt**:** 8bit, 8-bit, pixel art, pixel, 1boy, 20 years old, 2d, ((manga)), ((anime)), wearing teal sunglasses, red ombre short messy hair, side of head shaved

34. Anime with girls

Anime with Girls The previous example creates anime, but if you specifically want anime featuring girls, state this directly in the prompt, for example:

Prompt**:** Anime key visual of a girl with black hair and glasses, official media, trending on pixiv

35. Aerial photography

Aerial Photography To get a shot taken from an airplane, start with “ Aerial photography ” and then describe exactly what should appear in the frame, such as a spring forest:

Prompt**:** Aerial photography of a winding river through spring forests, with vibrantgreen foliage

36. Still life

Still Life Create a still life themed around Thanksgiving celebrations—you can replace it with another holiday—by describing what should be on the table:

Prompt**:** still life, thanksgiving turkey meal, fancy candle, wine, gravy, pumpkins, realistic photography, decorative, grapes

37. Fantasy

Fantasy The neural network can also handle nonexistent objects, including fantasy. Here is an example of a working prompt for a fantasy image:

Prompt**:** (extremely detailed CG unity 8k wallpaper), RAW photo, (sci-fi lab with lots of testing equipment and beakers and test tubes in background:1.1), robotic woman with electrical wiring for hair, highres, HDR+++, by Jeremy Lipking, by Antonio J. Manzanedo, raytracing, reflections, glowing eyes, (beautiful lighting:1.1), extremely detailed and intricatetechwear clothing, smug, super color range

38. Architecture

Architecture The examples above feature interior design, but you can also use a neural network to create architectural shots. For example, you can create an image of a modern house on a hill:

Prompt**:** House on a hill, modern architecture, house concept, sunset, morning, stary sky, rtx lighting, cloudy sky

39. Vintage

Vintage You can create a vintage photo by adding the term “

Vintage-style

” at the beginning of the prompt, for example:

Prompt**:** Vintage-style travel photography of a train station in the United States, with passengers and old luggage

40. Space

Space Neural networks are also excellent at generating space scenes. For example, the following prompt will produce an image of an astronaut with a reflection of Earth in the spacesuit.

Prompt**:** Digital painting of an astronaut floating in space, with a reflection of Earth in the helmet visor

41. Apocalypse

Apocalypse You can easily create an image of the end of the world and the apocalypse using the following prompt:

Prompt**:** Concept art for a post-apocalyptic world with ruins, overgrown vegetation, and a lone survivor

42. Pixel art

Pixel Art You can create an image in the Pixel Art style, that is, a “pixelated” image like those in retro games. Use the following:

Prompt: Pixel art of an 8-bit video game character running through a retro platformer level

43. Geometry

Geometry Realism, anime, and fantasy are not all a neural network is capable of. SD can also create abstract art and geometric shapes. For example, the following prompt will let you create an abstract colorful panel:

Prompt**:** A minimalistic digital artwork of an abstract geometric pattern with a harmonious color palette

44. Steampunk

Steampunk Just like cyberpunk, you can create an image in the Steampunk style, for example, using the following task description:

Prompt**:** Steampunk digital art of an inventor’s workshop, with intricate machines, gears, and steam engines

45. Celebrities

Celebrities You can create an image of any celebrity, either photorealistic or in another style. To get a photorealistic image, use a combination like this with a negative prompt.

Prompt**:** Joe Biden, shot on Sony A1

Negative prompt**:** computer graphics (2D, 3D)

46. Post-apocalypse

Post-apocalypse You can create a post-apocalyptic image by using the term “

postapocalypse

”, and if you want an image in a “survival” style, add “

warrior

”. Here is an example of a ready-made prompt:

Prompt**:** postapocalypse, warrior, Knight warrior helmet skyrim mask elder scrolls v nordic armor bethesda adam adamowicz illustration character design concept, unreal 5, daz, hyperrealistic, octane render, cosplay, rpg portrait, dynamic lighting, intricate detail, harvest fall vibrancy, cinematic volume inner glowing aura global illumination ray tracing hdr

Which prompt to use

Now you know how prompts are structured: they should be as detailed as possible, and you also have 46 popular prompts from our article at your disposal. Use them to work with Stable Diffusion. If you do not speak English even well enough to formulate tasks for AI, you can use DeepL or any other similar online translator. And if you do speak English but run into difficulties because the AI does not understand you, prompt generators and neural-network assistants (ChatGPT, Claude) can help: they will generate a properly formatted text prompt for the required model. To make the style or character consistent across different images, add LoRA, and define the pose and composition via ControlNet.

Author: Ai Pedia

I study AI not as magic but as a tool. I love order in data and complex systems.