How to get the most out of your prompts to create AI generated images?
Are you struggling to use stock photos and clip arts that you try to fuse to make a unique image? Do you find yourself amazing if a design or art can truly capture what you want to say and how you want the viewers to feel when they look at it?
It is really possible and easy to do with AI-generated images. How do you generate them? That’s why you should read this article.
Image Generation with AI is the latest technological advancement in the artificial intelligence industry. With it you’ll be able to create original and high-quality AI-generated images in seconds, and all you need is your imagination and a single sentence.
The top sites of AI-generated images are the DALL-E 2, Pictory, Midjourney, Neuroflash and Stable Diffusion. In case of AI advertising, Adcreative AI is ranked. All are using CLIP diffusion, a transformer model from Open-AI that matches text embedding with image embedding to form a new and unique image.
What are the benefits of using prompts to create AI-generated images?
What is the meaning of the prompt?
Prompt is a text that contains predetermined details which can help an AI generator to create an image.
The main benefit of the prompt is the ability to create a unique image. That is very important in many ways.
1. It is your own image and you can use it for commercial purposes (That depends on the conditions and criteria of AI latforms. For example Stable Diffusion and DALL•E2 allows the use of the images in a commercial way but Midjourney limits it to paid members only.)
2. No hard and fast rules for using the prompts. With your creativity and imagination, millions of prompts can be created with free hand.
3. Every image generating platform has a set of pre-defined styles that can be assisted for your prompt using to create the image which is really storming in your mind.
4. The prompt, that is correctly defined, can create an AI-generated image which has been really drawn in your mind. Then the final outcome is a very awesome unique image that is completely royalty free. Unique or original images are very important to SEO your blogs.
On May 18, 2021, Google announced the gradual implementation of its Google MUM (Multitask Unified Model) technology, which understands and extracts information directly from images, videos, and audio files.
“Make sure that your visual content is relevant to the topic of the page. We suggest that you display images only where they add original value to the page.
We particularly discourage pages where neither the images nor the text are original content.” As Google stated
What are the best prompts for creating AI art?
There is no particular best type of prompt for creating AI-generated images. All AI generating tools use the same ways to generate an AI art through a prompt that has been customized based on the imagination of what will be the final. The prompt actually is used according to the user requirement. A number of predefined Stiles, Modifiers and Parameters will be mixed up to get the final outcome.
How can you create AI-generated images using prompts?
Prompt practices are very simple. Whatever you think that can be generated with just writing a sentence.
Basic mode
Now imagine
1. A mango that looks like camera
Image is created by the author – DELL-E 2
2. An elephant is dancing on a stage
Image is created by the author - Midjourney
3. Gandhi is waiting at bus queue
Image is created by the author - Midjourney
Descriptive mode
In descriptive mode, you can add more details to your sentences by using adjectives, Art styles, modifications and various other parameters. In descriptive mode, be more careful about the words and its meanings
Concrete words and Abstract words.
If you are totally a beginner to AI creating, you are to start with “Concrete Words”. When you imagine what you want to see, remember that concrete things are easier for the AI to represent than abstract words.
Following are examples of concrete words: Ball, peacock, axe, beetle tire, tomato pitchfork, mousetrap, lantern, sled, board act.. So if you work with concrete words, you can get more predictable images.
If we create a prompt like “a pen is on a book” it will show the below image.
Abstract words, on the other hand, lead to more diverse results.
Examples for abstract words: Hope, sympathy, worthy, point of view agnosticism adequate, sufficient progressive.
Modifications, Style, Artists and Media Types
Exactly, AI-generated images can become really beautiful and realistic when you add more information about the resolution and the rating result. For your reference, here are a number of styles, artists, and mediums you can try to positively impact your results.
Here are some modifications that can be applied to creating an image and get an idea about how the final output displays after doing them.
○ “Detailed”
Adds more precise details to the output, instead of simple art, but can also make the art overwhelming/over the top in small details.
○ “Award-Winning Art”
Images in the dataset with captions like “Award-Winning Art” are usually extremely creative and original, so using this modifier can greatly improve the quality and inventiveness of your generations.
○ “Trending on ArtStation”
This modifier will sample extra training data from the most-liked artwork from the website ArtStation. Images which trend on ArtStation are usually very visually-appealing as it means the ArtStation community enjoys those images, so filtering the data to produce images similar to those will greatly increase the quality of the generated art.
○ “Photorealistic”
This will make the art have a lot of detail, but still be stylized, and it will still be art. Do NOT use this if you want to create a prompt which looks like a real photo, as the term “photorealistic” is used to describe an artificial image that looks real, not a real photo.
○ “Unreal Engine”
This modifier works similar to “photorealistic”, however it will produce slightly faker video-game style images with smoother edges, and it can be clearer and have sharper detail than the “photorealistic” modifier.
○ “Fanart” (Ex. “markiplier fanart”)
This gives the generation a cute young amateur graphic design feel, adding hearts to the image and so on.
Image Quality
○ “4K/8K”
Most images in the dataset with the caption “4K/8K” are of high production value therefore will look more professionally photographed if you add this modifier.
○ “15mm wide-angle lens”
Very wide image with lots of information in the image.
○ “35mm lens”
Reasonable amount of background blur, reasonable zoom level.
○ “85mm lens”
Quite zoomed in photo, a lot of background blur and detail on subject.
○ “200mm lens”
Extremely zoomed in photo, tons of background blur, & will look like it was photographed from a far distance and then zoomed in a lot (good for photos of flying birds, small animals”.
○ “Bokeh”
Enforce a large amount of background blur with clear outer bands, this can be used as a replacement for the “mm lens” prompts. Also could cause the subject to be closer to the camera.
○ “Award-Winning”
Will greatly increase the odds of the image being of high production quality and more aesthetically pleasing.
○ “Tilt-shift photography”
○ “Cinematic Movie Photograph”
Adds a very atmospheric movie-like feel to the image, with great color tones and image composure, and can also add nice background blur and pretty camera angles.
○ “Macro”
Will give a very close-up image with lots of background blur (good for small objects, animals).
Lighting
○ “Cinematic Lighting”
Movie-like imagery with dramatic shadowing and very strong vibrancy, it also seems to add sun rays whenever it can.
○ “At/During Golden Hour” or “Golden Hour Sunlight”
The hour just after sunrise or just before sunset when the natural light is soft and warm. Increases the temperature of generations.
○ “Ambient Lighting”
○ “Studio Lighting”
Dark/light background is imposed behind the subject, lighting accentuates details of the figure in the foreground.
○ “Lens Flare”
Adds a streak of light onto an image generation, creating the appearance of a bright light source being just outside of the frame.
● Situational
○ “Nature Photography”/”National Geographic”
Photographs in the dataset with these captions tend to showcase animals/nature in extraordinary positions and situations, works similarly to “Award-Winning” but is only for nature. This will also make animals/nature look more real and accurate.
Other
○ “Infrared Photography”
○ “Ultraviolet Photography” (2)
○ “Pinhole Photography”
○ “First person GoPro image”
Creates an image from the point of view of a human, usually showing arms and sometimes legs, works best in exploration context.
○ “Knolling”
Breaks down object(s) and rearranges them into an organized array, viewed from a top-down angle.
○ “[specific type of] film”
Applies an off-white/hue atmosphere to a photograph, with added grain, further enhancing the authenticity of a generation
You can apply the following common Styles for your images and test the results will change.
Art styles: Abstract, Abstract Expressionism, Academicism, American Realism, Anime, Art deco, Art Nouveau, Arts and Crafts, Atompunk, Baroque, Bauhaus, Biopunk, Classical Realism, Clockpunk, Conceptual Art, Cubism, Cybernoir, Cyberpunk, Dark Fantasy, Decopunk, Dieselpunk, Digital Art, Expressionism, Fantasy Realism, Flowerpunk, Fine Art, Forestpunk, Futurism, Gothic, Harlem Renaissance, High Fantasy, Impressionism, Installation Art, Manga, Modern Art, Modernism, Neoclassicism, Neo-Impressionism, New Realism, Op Art, Photorealism, Pixel Art, Pop Art, Post-Impressionism, Postmodernism, Precision Art, Realism, Rococo, Romanticism, Socialist Realism, Steampunk, Surrealism, Synthwave.
Painting Types: Acrylic Paint, Airbrush, Canvas, Cave Painting, Chinese Painting, Coffee Paint, Color Field Painting, Dripping Paint, Fine Art, Stained Glass, Gouache, Graffiti, Hard Edge Painting, Hydrodip, Wall Painting, Oil on Canvas, Oil Paint, Painting, Paper Marbling, Puffy Paint, Rock Art, Scroll Painting, Splatter Paint, Spray Paint, Still Life, Street Art, Tempera Paint, Tibetan Painting, Watercolor, Wet Paint.
Print styles: Advertising, aquatint, banner, barcode, block printing, blueprint, brochure, business card, collage, coloring book, comic, cyanotype, election photo, election poster, etching, graphic novel, halftone, illuminated manuscript, illustrated brochure, instruction manual, intaglio, linocut, lithograph, logo, magazine, “Magic the Gathering” card, manuscript, map, mezzotint, monoprint, movie poster, newspaper, newspaper printing, photo collage, photography, stamp, poster, product photo, propaganda poster, QR code, schematic, signage, silver gelatin, sticker, storyboard, storybook illustration, tarot card, ukiyo-e, visual novel, wall sticker, woodblock printing.
Adjectives: strange, ancient, angelic, angry, anxious, athletic, award-winning, simple, beautiful, chaotic, cheerful, clean, cold, colorful, confusing, cozy, creepy, cute, depressing, detailed, dirty, disgusting, dreamy, dry, ecstatic, older, ethereal, evil, excited, expensive, fancy, bold, flat, flat design, flat shading, fluffy, friendly, furry, blurry, gloomy, good, gorgeous, creepy, hairy, happy, very detailed, huge, hyperrealistic, impossible, incoherent, complicated, complicated maximalist, joyful, large, solitary, clear, luminous, massive, massive scale, mature, gentle, micro, mini, minimalist, moody, morbid, speckled, muted, nano, nervous, OCD, old, squiggly, otherworldly, photorealistic, simple, powerful, pretty, priceless, psychedelic, calm, rainy, realistic, refreshing, sad, simple, eerie, sleepy, smooth, ghostly, strong, surface detail.
Lighting: Accent lighting, , artificial lighting, backlighting, beautiful lighting, blue hour, bright lighting, lit by candlelight, Christmas lighting, cinematic lighting, backlighting, twilight, dark lighting, dawn, daylight, daytime, subdued lighting, , twilight, evening, film noir lighting, lit by firelight, flickering light, floodlight, fluorescent light, front lighting, global lighting, colored lighting, dramatic lighting, golden hour, semi-dark lighting, afternoon, halogen light.
Time periods: , Ancient Greece, Ancient Rome, Antiquity, Assyrian Empire, Aztec, Benin Kingdom, Bronze Age, Byzantine Empire, Carolingian Empire, , Edwardian Age, Elizabethan Age, Georgian Age, Gilded Age, Great Depression, Heian Period, Inca, Industrial Revolution, Iron Age, Maori, Mayan, Middle Ages, Meiji Period, Mid-Century, Middle Ages, Ming Dynasty, Minoan, Babylonian Empire, Modern, Moorish, Dark Ages, Mughal Era, Nasrid, Navajo, Neolithic, Olmec, Ottoman Empire, Paleolithic, Ancient Egypt, Persian Empire, Pre-Columbian, Prehistoric, Qing Dynasty, Regency, Renaissance, Retro, Shang Dynasty, Songhai, Stone Age, Sumerian, Tokugawa Shogunate, Tudor, Victorian, Viking, World War I, World War II, Zhou Dynasty, Zuni Pueblo, 1100s, etc.
Decorative art: 3D printing, applique, balloon modeling, balloon turning, bas-relief, bead embroidery, blown glass, bone china, carved, carved ivory, carved lacquer, carving, kneading, cloisonne, crochet, Cross stitch, Diorama, Embroidery, Enameling, Felting, Fretwork, Glass mosaic, Ice carving, Impressionist mosaic, Marquetry, Inlay, Puzzle, Crochet, Lacquer, amigurumi, Lampwork, Lath art, Leather work, Marble, marquetry, micromosaic, miniature painting, Leather carving, modular origami, mosaic, needlework, origami, paper model, paper cut, papier mache, pietra dura, porcelain, pottery, doll, puzzle, pysanky, Quiltwork, Quilting, Relief Carving, Repousse, Origami, Sand Art, Scrimshaw, Sculpture, Stained Glass, Statue, String Art, Tapestry, Tattoo, photographic mosaic, Tattoo Art, Venetian Glass, Weaving, Wet Folding, Carving, Wood Burning.
Rendering techniques: 3D Model, 500px, based renders, Arnold Render, ArtStation, Blender Render, CGsociety, Cinema4D Render, CryEngine, Cycles Render, Daz 3D, DeviantArt, DirectX Render, Doughy Render, Houdini Render, KitBash3D, Luxcore Render, 3ds Max, Marvelous Designer, MentalRay Render, OctaneRender, Optix Render, Photobashed, Photoshop, physically Pixia, Quixel Megascans, Raylectron Render, Redshift Render, Sketchfab, Substance 3D, Terragen, Unreal Engine, Vray Render, Weta Digital, Zbrush Render, Infini-D Render.
Photography Styles: Daguerreotype, Tri-X, Kodachrome, Slide Film, Portra 800, Natura 1600, Ilford Delta 3200, Polaroid, Hasselblad, Double Exposure, Multiple Exposure, Large Wide Angle Lens, Fisheye Lens, Tintype Anamorphic, Lensbaby, Telephoto, Prime Lens, photo shooting, commercial, thermography, x-ray, infrared ,Film Negative, Tilt Shift Lens, Format Camera, f1.8, f2.8, f4, f11, f16,
Artists: William Logsdail, , Roy Lichtenstein, Richard Corben, Michaelangelo, Gerhard Richter, Bjarke Ingels, John Berkey, George Inness, J.M.W.. Turner, Todd McFarlane, Caravaggio, Atey Ghailan, Hirohiko Araki, , Ray Caesar, Takeshi Obata, Antoine Blanchard, Diego Peter Andrew Jones, Velázquez, Huang Guangjian, Romero Britto, Beatrix Potter, Guido Borelli da Caluso, Lucas Cranach the Elder, Nele Zirnite, Bob Ross, Zdzislaw Beksinski, Glen Fabry, Jane Graverol, Krenz Cushart.
Colors: black, gray, white, red, purple, fuchsia, green, lime, olive, yellow, navy blue, blue, maroon, aquamarine, silver, aquamarine.
Common expressions that can increase results: high quality, trend on artstation, dynamic, trend on pixiv, vivid, geometric, intricate , Masterpiece detailed.
Practices
01. A military bladerunner man in an abandoned warehouse, moody, cinematic, RAW photo, extremely detailed, 8k resolution, photo by gregory crewdson, neon lights in background, --ar 2:3 --v 5 --s 750 --q 2
Image is created by the author - Stable Diffusion
02. surrealist photorealistic portrait of a swamp creature, eyes looking at camera, skin like a crocodile, glistening in the moonlight, sad expression on his face -v 5
Image is created by the author - Stable Diffusion
03. Characters, photorealistic life like Lord Buddha with a serene expression and surrounded by a lush forest and beautiful flowers and birds. It emits a warm light and a mystical aura, suggesting divine power. The image is a beautiful reminder of the peace and happiness found within and beyond our everyday experience. Projecting its divine love into the universe, uplight --ar 809:500 --v 5 --q 2
Image is created by the author - Stable Diffusion04. female Portrait mythical character Aquarius, blonde hair, red lips, blue eyes, full body, extreme beauty, in armor, smile, sparkles, dramatic lighting, cinematic, cinematic astrology, zodiac sign, exteme beauty, perfection, attention to details, intricate details, cinematic, cinematic light, fantasy astphotorealistic In a large reverse reflective white opposite luminous palace museum room full of picture and frankfurter sausage bananas with plastic vinyle gorgeous fashion top models plus size perfect faces dance and jump frantically in the air to attack the glossy colored giant mushrooms bones and hairy chimera animals with scales, by Damien Hirst, Jake et Dinos Chapman, Jheronimus Bosch, Dali, Alice, William Klein, jean paul goude, pinocchio, Bioman, thierry mugler, shining, Chris Cunningham, david lachapelle, Jheronimus Bosch, wim delvoye, hyperrealism photography, extremely detailed, art design photography, cinematic lighting, intricate, 8k, HD, cinematography, photorealistic, Unreal Engine, Cinematic, portrait Photography, hyper - detailed, insane details, intricate details, Cinematic, Editorial Photography, Photography, Photoshoot, DOF, White Balance, 32k, Super - Resolution, Megapixel, ProPhoto RGB, Halfrear Lighting, Backlight, Natural Lighting, Incandescent, Optical Fiber, Moody Lighting, Cinematic Lighting, Studio Lighting, Soft Lighting, Volumetric, Lighting, Accent Lighting, Global Illumination, Screen Space Global Illumination, Ray Tracing Global Illumination, Optics, Scattering, Glowing, Shadows, Rough, Shimmering, Ray Tracing Reflections, Lumen Reflections, Screen Space Reflections, Diffraction Grading, GB Displacement, Scan Lines, Ray Traced, Ray Tracing Ambient Occlusion, Anti - Aliasing, FKAA, TXAA, RTX, SSAO, Shaders, OpenGL - Shaders, GLSL - [...]
Image is created by the author - Stable Diffusion
The same image is created at Midjourney
In Midjourney, the image gets more realistic look to the given prompt than Stable Diffusion
05. Albert Einstein, detailed, 4K, 8K intricate details, cinematic, cinematic light, photography, extremely detailed, art design photography, cinematic lighting, intricate, 8k, HD, cinematography, photorealistic
Image is created by the author - Midjourney
06.Sri Lankan fisherman, super details, portrait, golden ratio, 8kHD
Image is created by the author – Midjourney
Tips for quality AI image
01. Always use positive terms
That means don’t use negative terms to describe your AI-generated image. Terms such as “not,” “but,” “except,” and “without” are not suitable to describe your image prompt.
For example if you prompted as the “car that is not red”, the final output may be red or something other colour. It does not give a clear idea about the colour of the car. All AI-generated image formation platforms are running on LLM (large language model). They are trained for catching positive terms only.
02. Avoiding of future assumption or forecasting
All AI applications are trained to describe the things that happened in the past. They have no sense to look for future.
For example, it will not be able to provide you with a depiction of London in 3000 A.D.
But you can provide some directions to AI as to what you want to see especially and you can try “Futuristic city home architecture at Downing Street in London” alternatively.
Image is created by the author - Stable Diffusion
03. Favoring of singular nouns or specific number
In general we are using plural forms of words to explain what we want. But on the way of AI image generation, it may cause some confusion.
If you want to generate a photo of Persian horses, it could include two or three or more horses. For example, “Black Persian horses with iron shoes”
Image is created by the author - Stable Diffusion
The best option is to use the word’s singular form or to indicate exactly how many horses you want in the picture by being precise in the description. Just as “ Four black Persian horses with iron shoes ”
Image is created by the author - Stable Diffusion
Conclusion
AI-generated images have revolutionized the world of design and art creation. With the help of prompts, anyone can create unique and high-quality images that accurately capture their imagination and ideas. The use of AI-generated images is especially useful in commercial settings, where unique and original images are essential for SEO and other marketing purposes. By using prompts, users can easily create concrete or abstract words, add modifications, and choose from various styles, artists, and media types to create their perfect image. The possibilities for AI-generated images are endless, and with the continued advancement of AI technology, we can expect to see even more incredible and unique art in the future.
Waruna Chanaka
0 Comments