True-to-Life AI Photos: Mastering Midjourney for Unbelievably Realistic Imagery
A Picture’s Worth a Thousand Doubts
Gone are the days of using photos as definitive proof. Thanks to advancements in AI image generation, telling fact from fiction just became a whole lot harder. ICYMI: Just within the last few weeks the public has been bamboozled by several viral photorealistic images created using the text-to-image AI generator, Midjourney. A few examples include the photo of the Pope in a puffer jacket, the fabricated image of Donald Trump being arrested, and an AI-generated image that fooled judges at the 2023 Sony World Photography Awards. With fake photos going viral left and right, we all need to make sure to be more critical of images online and work to educate others, especially those who might not be as tech-savvy.
AI-Generated Images and the Future of Media
AI-generated images that look like photos are undeniably captivating but can also be risky- fake images can spread misinformation, influence public opinion, or even ruin someone’s reputation. The bots have been trained on a vast amount of celebrity photos, making it easier to create convincingly realistic images of them. We all need to get better at spotting the signs of AI-generated images, double-checking sources, and thinking twice before sharing something.
Following the viral Pope and Trump photos, Midjourney discontinued access to their free trial. Midjourney’s CEO, David Holz, attributed the decision to overwhelming demand and abuse of the free trial system.
What is Midjourney?
Midjourney is one of the most popular AI image generators. Using artificial intelligence, the Midjourney Bot is able to transform a text description into an image. The program is accessible via Discord, a social communication platform with communities called “Discord Servers”. Midjourney’s newest version, V5, launched in March 2023 and has been gaining widespread recognition for how impressive its photorealistic capabilities are.
What Makes a Photo Look Realistic?
The best photographs have an incredible level of detail, spot-on color reproduction, and a sense of depth. In creating photorealistic images, the trick is crafting prompts to generate an image nailing these key features, like:
📸 Sharpness and Clarity: This is all about the overall level of detail included in a photograph, especially around the edges of objects and textures.
📸 Color and Contrast: How well the image shows true-to-life colors and appropriate levels of contrast.
📸 Depth of Field: This is when a specific area of a photo is in focus while blurring out others. For example, focusing on someone’s face in a portrait photo while blurring out the background helps to create depth and dimension.
📸 Lighting and Shadows: An accurate representation of light sources and the way they cast shadows helps to make the image even more authentic looking.
Expert Tips for Prompting
As mentioned earlier, Midjourney uses text prompts to generate images. A prompt is a text phrase that the bot will use for inspiration to create the masterpiece image. The words and phrases from the prompt are broken down into smaller bits, called tokens, which help it reference its training data when generating images. A well-constructed prompt can make all the difference in getting unique and high-quality images.
Midjourney has a great doc on prompts on their website. I’ve also summarized it for you below:
💛 Don’t make your prompts too short or too long.
💛 Stick to the main ideas you want to see in your image and make each word count.
💛 The bot isn’t a grammar nerd. It doesn’t understand grammar, sentence structure, or words like humans.
💛 Focus on the essence of your prompt and specific word choice.
💛 Focus on what you want in the image. If there’s something you don’t want, use the “––no” parameter for a better chance of keeping something unwanted out.
💛 Be clear on the details and context that are important and remember, anything you leave out is going to be randomized.
💛 Think about the subject, medium, environment, lighting, color, mood, and composition when prompting the bot.
And a few more expert tips I’ve learned along the way:
⭐ Everything in your prompt has a default weight of 1. That said, Midjourney prioritizes what’s first in your prompt.
⭐ You can add weight or use negative weights. For example “woman driving car::2” would give it a weight of 2.
⭐ Check your settings (/settings)! Make sure you’re using version 5 for the most realistic-looking images. I like to set mine to high quality and style med.
⭐ Upload a photo and use the /define feature to get prompt inspiration! The Midjourney explore page is also a great place for prompt inspo!
⭐ Save time with permutations! Test variations of a prompt with a single command using the /permutations feature.
⭐ Experiment with cameras and film types for a different look/feel to your image.
⭐ Explore different types of angles and perspectives.
⭐ If there’s a style or subject you like, react with the envelope (✉️) and the Midjourney Bot will message you with the seed number which you can use at the end of your prompt to get the same style/subject (––seed 1234567)
Terms to inspire your prompts
I’ve had quite a bit of time to test creating photo-realistic images using Midjourney (6,888 jobs in the last month 😬) and I’ve found that some terms work better than others. Below are some terms to help inspire you to create realistic images. My favorite terms to use are in bold! I’ve also provided 3 sample prompts for each category so you can see how I structure my prompts!
💠 Realism and Details 💠
◼Detailed Texture
◼Blink-and-you-miss-it-detail
◼Clear edge definition
◼Crisp Detailing
◼Hyper-realistic portraiture
◼Realistic Facial Expression
◼Photo-realistic techniques
/imagine prompt: full body blonde woman in a white dress next to a palm tree, dewy face, sunrays shine upon it, crisp detailing, no haze, modern dress, tropical, contax tix, cabincore, 8K, intricate, blue eyes --v 5 --q 2
/imagine prompt: a woman with glasses in front of a house, in the style of pop inspo, eye-catching resin jewelry, green academia, fluid and loose, retro, clear edge definition, exaggerated features --ar 93:122 --v 5 --q 2
/imagine prompt: a model walks the runway in a magenta blouse and skirt, in the style of 1990s, light aquamarine, photo taken with ektachrome, blink-and-you-miss-it-detail, swiss style --ar 2:3 --v 5 --q 2
💠 Camera and Film Types 💠
◼Diana F+
◼Redscale Film
◼Film Photo
◼Canon EOS R3
◼Minolta Riva Mini
◼Photo taken with Provia
◼Photo taken with Fujifilm Superia
◼Kodak Portra
◼In the style of Kodak Aero Chrome
◼Photo taken with Ektachrome
◼Rollei Prego 90
◼Ferrania P30
/imagine prompt: a woman wearing a metallic outfit at night in New York City, flash photography, in the style of diana f+, glitter, i can’t believe how beautiful this is, 1970s --v 5 --q 2
/imagine prompt: one woman on the stage walking in a short dress and tight outfit, in the style of redscale film, glittery and shiny, made of feathers, sparkling water reflections, blink-and-you-miss-it-detail, 1970-present, y2k aesthetic — ar 43:140 --v 5 --q 2
/imagine prompt: the fable of the rose: the cowboy and sassy fashionista, in the style of fujifilm fujicolor c200, gritty hollywood glamour, street scenes with vibrant colors, sculptural costumes, light red and light pink, pop inspo — ar 31:41--v 5 --q 2
💠 Lighting and Composition 💠
◼Light over her face
◼Volumetric Lighting
◼Play of light and shade
◼Warm face light
◼Neon lighting
◼Dynamic Lighting
◼Low-angle
◼Bird’s-eye view
/imagine prompt: a woman wearing a unisex white shirt and red dress, in the style of gritty hollywood glamour, neon lighting, photo taken with fujifilm superia, dark amber and red --ar 31:39 --v 5 --q 2
/imagine prompt: a blue photo of a woman with light over her face, in the style of bold fashion photography, volumetric lighting, intense shadows, close up, aurorapunk, pop art influencer --ar 31:39 --v 5 --q 2
/imagine prompt: the woman’s face in the street, low-angle, clear edge definition, bright daytime lighting --v 5 --q 2
💠 Styles and Aesthetics 💠
◼Snapshot Aesthetic
◼Lomography
◼Vintage photos
◼Nostalgic imagery
◼Surreal fashion photography
◼Minimalist photography
◼Pop Art
◼Street photography
/imagine prompt: woman is in white dress, sleepycore, carpetpunk, nostalgic imagery, playful poses, rollei prego 90 --ar 46:41 --v 5 --q 2
/imagine prompt: paris fashion week: lady gaga makes style statement, in the style of 1970s, glittery and shiny, multiple filter effect, shot on 70mm, nocturne, cabaret scenes, womancore --ar 43:140 --v 5 --q 2
/imagine prompt: a woman with a red dress is shopping at the supermarket, in the style of snapshot aesthetic, glitter, picassoesque, goosepunk, cosmic inspiration, i can’t believe how beautiful this is, effortlessly chic --ar 93:40 --v 5 --q 2
💠 Subject and Context 💠
◼Candid celebrity shots
◼Captivating documentary photos
◼Glamorous Hollywood Portraits
◼Model from magazine
◼Night photography
◼Self-portraits
◼Webcam Photography
◼Iconic Album Covers
/imagine prompt: photo shoot, in the style of red and maroon, sabattier filter, wearing gown, night photography, flash photography, life in New York City, sitting at bar, understated elegance, 1970s, dreamy --v 5 --q 2
/imagine prompt: an oscar winning actress in a red latex outfit, in the style of vintage aesthetics, candid celebrity shots, post-’70s ego generation, dark white and dark orange, body extensions --ar 46:75 --v 5 --q 2
/imagine prompt: a woman wearing a pink dress and scarf, in the style of Mediterranean-inspired, tabloid photography, effortlessly chic, pictorial fabrics, light orange and red, calm waters, densely patterned imagery --ar 62:79 --v 5 --q 2
AI-Generated Images and Their Role in Modern Photography
Many photographers see tools like Midjourney as exactly that- a tool. And some have even gone on to say that it elevates their craft, as they can create mood boards, and experiment with angles, subjects, lighting, and more, without needing a full array of equipment or a photography budget. Additionally, AI-generated images can inspire photographers to think differently about their work and encourage them to find new ways of storytelling and conveying emotions. Right now, AI-generated images don’t have to overshadow traditional photography but can instead act as a springboard for innovation and fresh artistic expression. Ultimately, it’s the connection with the viewer that remains at the core of all artistic endeavors, no matter if they were created through conventional methods or AI-generated processes.
Try Midjourney for Yourself
If you’re eager to try Midjourney, you’ll need a Discord account. Go to Midjourney.com and click “Join the Beta” to receive an invitation to the Midjourney Discord. You can choose between three subscription plans: Basic at $10/mo, Standard at $30/month, and Pro for $60/month, each granting different levels of access to the image generator. If you know you plan to use the generator for an extended period of time, they also offer annual billing at 20% off the monthly pricing. Each plan comes with a specific number of “fast hours,” which allows for high-priority processing to generate or upscale your images quickly.
Text-to-image generators like Midjourney have unlocked a whole world of possibilities for artistic expression, but also, for deception. As we embrace new tech tools, it’s important that we keep questioning what we see and educate ourselves and others about the potential downsides of AI-generated images. At the same time, we can celebrate the power of human creativity and appreciate how these new tools can enrich and inspire our artistic journeys. It will be exciting to see how photography and AI-generated images coexist and mingle in the creative world.
Thanks for reading! Please feel free to share your AI-generated images with me, I’d love to see them! For more, you can follow me on Twitter @brittanynpotter or contact me at workwithpotter@gmail.com.