ChatGPT Meets Midjourney: A Guide on How to Use ChatGPT to Generate Midjourney Prompts

We can creates impressive images with Midjourney using just a few words and phrases. However, its prompts aren’t as natural as those of ChatGPT, which mimic human speech. So, is it possible to use these two AI tools together for enhanced power?

In this article, we’ll show you two ways. The first is quick and simple, but offers less customization and is for Plus users only. The second, available to everyone, allows you to use or tweak my prompts as you wish.

Let’s begin!

Use Photorealistic ChatGPT Plugin

ChatGPT is a master of grammar, even able to work as a foreign language instructor to pinpoint your grammar errors. This makes interaction with it feel as natural as talking to a human.

Unlike ChatGPT, Midjourney doesn’t grasp grammar and sentence structure. Often, more precise synonyms lead to more pleasing images. For instance, using “petite”, “compact”, “diminutive”, or “tiny” could be more effective than the broader term “small”.

Moreover, to better control Midjourney’s output, it’s crucial to add parameters like “- – ar 16:9” to specify an image aspect ratio of around 1.75, not the default 1:1. This resembles coding and lacks intuition. As Midjourney evolves, the growing number of parameters may intimidate occasional users.

The Photorealistic ChatGPT plugin was designed to solve this problem. Available free for ChatGPT Plus users on the ChatGPT Plugin Store, it converts simple sentences into detailed Midjourney prompts.

To use the plugin, activate the ChatGPT plugin feature first.

Next, go to the ChatGPT plugin store, search and install the plugin.

Look for the “photorealistic” keyword and hit “Install”.

After installation, switch to ChatGPT4 before typing any prompts.

Hover over the “GPT-4” button, revealing a menu. Click “Plugins” in the menu.

Locate the “Photorealistic” plugin you just downloaded below the ChatGPT4 button and check its box.

Now, you’re ready to enter your prompt, for instance:

Generate an image of a dense forest during autumn, with trees displaying vibrant colors of red, orange, and yellow, and a clear stream running through.

ChatGPT then generates two prompts for us.

Prompt1:

Capture the essence of autumn in a dense forest. The trees are ablaze with vibrant hues of red, orange, and yellow, creating a natural tapestry that is both breathtaking and serene. A clear stream cuts through the forest, its waters reflecting the fiery colors of the foliage. The image should be hyper-realistic, with every leaf and ripple in the water rendered in exquisite detail. Use a high-resolution 16k camera with a wide-angle lens to capture the grandeur of the scene. The lighting should be soft and diffused, illuminating the forest in a warm, golden glow. The color palette should emphasize the rich, warm tones of autumn, with a focus on deep reds, bright oranges, and golden yellows. The composition should be balanced, with the stream leading the viewer’s eye through the image. –ar 16:9 –v 5.1 –style raw –q 2 –s 750

Prompt2:

Imagine standing in the heart of a dense forest during the peak of autumn. The trees are a riot of red, orange, and yellow, their colors so vibrant they almost seem unreal. A stream runs through the forest, its clear waters sparkling in the sunlight. This image should be rendered in a hyper-realistic style, with every detail from the texture of the tree bark to the shimmering surface of the water captured in high-resolution 16k. Use a DSLR camera with a telephoto lens to create a sense of depth and intimacy. The lighting should be natural and bright, highlighting the brilliant colors of the leaves and the clearness of the stream. The color scheme should be warm and saturated, with a focus on the fiery reds, oranges, and yellows of the autumn leaves. The composition should be dynamic, with the stream acting as a natural focal point. –ar 16:9 –v 5.1 –style raw –q 2 –s 750

Submit these to Midjourney to create an autumn forest image (Don’t forget the ‘/image’ command before it).

Let’s write another prompt:

Create a photorealistic image of a steampunk city, with steam-powered machinery, airships in the sky, and people dressed in Victorian-era clothing

I submitted the first prompt to Midjourney. Check out the result:

In summary, the Photorealistic plugin by default generates two Midjourney prompts based on our prompts, using the same parameters (–ar 16:9 –v 5.1 –style raw –q 2 –s 750), but you can adjust these or the prompt content as needed.

This plugin makes generating Midjourney prompts a breeze, but it requires a ChatGPT Plus account. For those without one, let’s explore another method.

Train ChatGPT into A Midjourney Prompt Helper

The previous method is efficient and great for realistic photographic style images. However, it’s accessible only to Plus members. Can ChatGPT, known for its versatile skills, assist us in creating Midjourney’s prompts?

The answer is yes, but with challenges.

Firstly, ChatGPT isn’t familiar with Midjourney as its training data is prior to September 2021. Even if it had networked knowledge of Midjourney, it would struggle to master the Midjourney prompting skills with limited information.

Secondly, while ChatGPT excels at generating natural language and code, Midjourney’s prompts follow a specific non-natural language format. They also need code-like parameters at the end of the prompts to control the output, but the prompts aren’t in a programming language.

To have ChatGPT compose Midjourney prompts, you must train it on Midjourney, the art of crafting its prompts, and ideally provide examples. This is the basic premise of training ChatGPT, with customization of the process based on your needs. This way, ChatGPT can stay current on writing Midjourney prompts.

Here’s a prompt I created for training ChatGPT. It can serve as a guideline, or be copied into ChatGPT. If too long, it can be divided and submitted to ChatGPT or used directly on FlowGPT.

https://flowgpt.com/prompt/-9ErLmO8Z-x0Z9-wojGLB

I’d like you to help me create Midjourney prompts. Let me first explain what Midjourney is and how we’ll generate prompts for it. We’ll also go through 20 examples to ensure you understand.

Midjourney is an AI tool that makes images from user’s input, similar to DALL-E.

The key part of the prompt are words or phrases that describe the image you want. More adjectives and specific descriptive nouns create unique images. On the contrary, basic nouns or adjectives make plain images. Keep in mind, Midjourney doesn’t understand grammar. So, very long prompts may not work well. When creating prompts, remove any unnecessary words. Fewer words give each word more importance, ensuring the image aligns with your theme.

For instance, “illustrate for me a beautiful sunset over a serene ocean, make the colors warm and soothing, and render it in an impressionistic style.” This prompt has words that Midjourney might not understand or work with. Phrases like “Illustrate for me” are unnecessary. Verbs like “make” and “render” are also redundant. Midjourney usually accepts descriptive words like nouns and adjectives. The prompt could be simpler: “warm soothing sunset over serene ocean, impressionistic oil paint.”

More specific synonyms often work better than general ones. For example, use precise words like “petite”, “compact”, “diminutive” and “tiny” instead of “small”. When creating your prompt, focus on specific details you want:

• Theme: People, animals, places, character, objects, events, etc.
• Use Case: Logos, web design, interior design, prototypes, product design, etc.
• Medium: Photos, drawings, illustrations, paintings, cartoons, watercolors, collages, sculptures, graffiti, mosaics, tapestries, pottery, etc.
• Environment: Indoor, outdoor, city, forest, island, desert, underwater, cave, future city, space, moon, space station, etc.
• Lighting: Soft, twilight, ambient, direct sunlight, overcast, moonlight, neon, candlelight, firelight, etc.
• Color: Vibrant, muted, bright, monochromatic, colorful, black and white, pastel, etc.
• Mood: Energetic, sedate, calm, raucous, restless, melancholy, dreamy, mysterious, etc.
• Composition: Portrait, close-up, headshot, bird’s eye view, symmetrical, leading lines, minimalist, silhouette, panorama, etc.
• Artists: Van Gogh, Picasso, Dali, Paul Cézanne, Leonardo da Vinci, Botticelli, Rembrandt, Hayao Miyazaki, etc.
• Art styles: Oriental landscape painting, Ukiyo-e, conceptual art, Bauhaus, Impressionism, Rococo, Surrealism, long exposure, etc.

You can also use a comma, plus sign, or “and” to separate different subjects. For instance, to depict a light and a house, you should separate them. Otherwise, if you type “light house,” Midjourney will show you a lighthouse.

Midjourney prompts can also include parameters to change how an image generates. Common parameters include:

1. –ar This parameter sets the screen’s aspect ratio. Default is 1:1; --ar 16:9 for computer screens and --ar 9:16 for cell phone screens.
2. –v This parameter specifies the Midjourney version. --v 5.2 represents version 5.2.
3. –s This parameter controls the artistic level of the image, with values from 0 to 1000. A higher value indicates a more artistic image, but the image will also deviate more from the actual effect described by the prompt. Lower values indicate that the effect of the image is closer to the description of the prompt, but less artistic. The suggested setting is --s 750.
4. –style This parameter --style raw is used for photographic images.
5. –q This parameter defines the image detail. More detail is created with larger values. Default value is --q 1. Recommended setting is --q 2.

With this knowledge, we’ll now explore 20 examples of prompts:

1. white cow inside pink flowers in style of national geographic photography
2. Giraffee, tropical, Paul Klee style, surreal masterpiece, octane rendering, focus, colorful background, detailed, intricate details, rich colors
3. a girl with long dark hair wearing a black hat, in the style of childlike innocence and charm, detailed facial features, enchanting, 32k uhd, rococo whimsy, ferrania p30, cartoon realism
4. she is posing in black and gold, in the style of eye-catching resin jewelry, urban decay, rim light, iconic, exotic, bold, cartoonish lines, meticulous design
5. Smiling sunflower, Paul Klee style, surreal masterpiece, artistic, focus, colorful background, detailed, intricate details, rich colors
6. falkor by Atey Ghailan and John Singer Sargent, Neverending Story
7. SSSniperwolf Nelliel Tu Odelschwanck from Bleach cosplay for women with cosplay sleeve, in the style of light emerald and light cyan, appropriation artist, anime influenced, exaggerated expressions, made of vines, heistcore, bombacore
8. a gold fish with human swimming in a bathtub filled with clouds, in the style of dusseldorf school of photography, 1970s, balloon
9. cinematic medieval interior photography in the style of Akos Major, Fine art
10. Young woman, long red hair, glasses, looking in camera with a small beautiful smile, red dress, futuristic background
11. 3D cartoon art, painting, Cats on the roof, masterpiece, octane rendering, focus, colorful background, detailed, intricate details, rich colors
12. abstract futuristic dystopian Japanese fashion craze, vibrant abstract surreal photography by Ed Emshwiller and Masamune Shirow and Beeple
13. A cyberpunk-inspired full-length illustration featuring a futuristic tight-fitting suit. The suit wraps the wearer’s body with a snug fit, exuding an air of sleekness and functionality. Embedded within the suit’s fabric are numerous tactile sensors, meticulously integrated to provide enhanced tactile feedback and data analysis. A holographic layer is projected onto the suit’s surface, producing shifting patterns and colors that mesmerize the viewer. Set against a black background, the suit stands out, evoking a sense of mystery and futuristic elegance. Illustration, digital art, brings this vision to life with meticulous attention to detail
14. oldtimer car in front of a vintage sunset, vintage tshirt design vector graphics, detail design contour, turqoise, pink, neon colors, white background
15. Sunset over the sea, dramatic clouds, masterpiece, octane rendering, focus, realistic photography, colorful background, detailed, intricate details, rich colors, realistic style
16. a large room with a center island and wooden flooring, in the style of dark bronze and dark beige, dark gray and dark black, rustic scenes, avocadopunk, cabincore, multiple styles, dark white and gray
17. knight of thorns woman in a spotlight, gradient, vector, fiberoptic, Hokusai, Banner Saga, glossy, cel shaded
18. traditional slavic folkloric monster, baba yaga, wurdulac, upyr in the style of alberto seveso
19. a teenage girl in black fur wears purple lipstick, in the style of monochromatic color schemes, dark brown and dark black, urban decay, bold yet graceful, emila medková, eye-catching, double tone effect
20. a girl with colorful eye makeup and a star shaped haircut, in the style of acidwave, dark brown and dark azure, sumi-e inspired, superheroes, babycore, bec winnel

You’ve learned how to use Midjourney prompt words by examples. Now, let them inspire you. I’ll share brief image ideas. Your task? Turn them into full, clear, creative prompt words. From now on, I’ll just give you my ideas. You’ll then provide two Midjourney prompts for each. Make sure you have added all the available parameters. Does this make sense?

After submitting this to ChatGPT, it understood it. Then, I gave it an image of a rocket-powered baby elephant. It provided two prompts. I picked the second and submitted it to Midjourney, where parameters can be customized.

Midjourney then produced 4 images to choose from.

Another example is a portrait of an ancient Chinese female warrior.

We can leverage ChatGPT’s potential to help us detail a short image idea before turning it into a Midjourney prompt. I began by writing a prompt. ChatGPT then supplied ten ideas.

I passed the first and fourth image descriptions to the trained Midjourney generator, resulting in two images of Persian cats.

The second image, although appealing, did not meet my request for a Persian cat playing an electric guitar. Instead, it held an acoustic one. Perhaps the country tavern, cozy firelight, calming melody, and warm tones in the prompt made Midjourney to choose an acoustic guitar to better fit the overall image ambiance.

ChatGPT offered several ideas to refine the image, but they didn’t all mesh. We had to select and tweak them to suit our needs. Still, ChatGPT saved us time and sparked many creative ideas.

To Sum Up

In summary, combining ChatGPT and Midjourney opens up a powerful avenue for image creation. Despite challenges, the rewards are compelling.

Trained ChatGPT can generate diverse prompts, enhancing Midjourney’s outputs. The process sparks creativity and often results in unexpected delights.

The Photorealistic plugin offers a quick solution for Plus users, but for others, training ChatGPT is a promising workaround. This merging of language and visualization skills heralds an exciting era in AI-assisted creativity.