How To Get Photorealistic Images With Midjourney [Steps & Prompts]

Over the past two years, AI has seen a significant expansion, and an impressive array of AI image generators are now available. 

The text-to-image model Midjourney is one of our favorite AI tools. With a simple text description or “prompt,” this tool can instantly produce a photorealistic image. It’s transforming the landscape for creative agencies by changing how creative projects are approached and executed.

While Superside firmly believes AI will never replace humans, we view artificial intelligence as a powerful tool to enhance creativity. Tools like Midjourney enable design innovation, helping us deliver bigger, better creative campaigns faster than ever before.

AI is now used in all our creative design processes, with Superside designers delivering everything from striking logo designs to pitch-perfect presentations with the help of bots. So far, we’ve saved our customers $1,400,000 using AI-enhanced creative and design processes—and we’re just getting started.

A case in point: We recently used Midjourney to create a campaign for a client who tapped into our design services, producing over 750 realistic images in less than 12 hours and cutting 90% of design time.

If you prefer a more hands-on approach to using AI in design—and seek valuable tips on achieving photorealism—keep reading. Emmanuel Cantón, AI-Enhanced Senior Graphic Designer and Illustrator at Superside, explains how to use Midjourney to produce stunning images and standout campaigns.

Understanding Photorealism in AI-Generated Images

Among AI aficionados, Midjourney stands out for its ability to blend human creativity with machine precision.

In November 2023, this tool’s user base exceeded 16 million, with daily active users ranging from 1.2 million to 2.5 million. Notably, 68% of users use Midjourney for entertainment, while 32% use it for practical purposes (e.g., to create realistic photos).

Key reasons AI image generators like Midjourney are having a massive impact on the design industry include:

• Lower barriers to entry: Design is now accessible to non-designers. AI enables businesses and individuals without design skills to create professional-looking images, logos and graphics with minimal effort or expertise.
• Faster design turnarounds: AI-powered tools like Midjourney or DALL-E help designers create images and visual concepts faster, reducing the time needed for brainstorming, prototyping and experimenting.
• A shift in roles: AI tools are simplifying image creation and shifting the role of designers. According to Playform, 65% of artists have used AI to find new ideas and develop assets. Designers are becoming more like creative directors, focusing on conceptual thinking and strategy and overseeing AI-driven creative processes.
• Opportunity for hybrid approaches:The fact that 54% of people recognize AI-generated art has created a unique opportunity for designers to blend AI with their own creative input. Knowing that AI has limitations, designers can use it to speed up processes or generate ideas while maintaining control over the final result.
• Developing targeted advertising campaigns: Using the right visuals is crucial for catching people’s attention and communicating marketing messages effectively. Marketers can use AI to create highly personalized campaigns by customizing visuals for specific audience segments. AI can analyze social media trends and create a photorealistic image or video likely to go viral.

Explore AI-Enhanced Creative Services

Unlock tomorrow's possibilities, today.
Learn more about our AI-Enhanced Creative
and AI Consulting Services.

Check them out

Midjourney Realistic Image Generation 101

To get started, you’ll need to access the Midjourney Bot on Discord via a web browser, mobile app or desktop app.

Create a verified Discord account and select a subscription tier based on the monthly processing minutes you’ll need (i.e., the time used to generate or enhance images).

Once in the server, type a brief description into the chat channel, and Midjourney’s Bot will generate several image variations. To achieve the best results per image, it’s essential to understand the key components—prompts, parameters, commands and tools—influencing Midjourney’s output.

1. Midjourney prompts

Crafting Midjourney prompts is the key to your interaction with this AI tool. These prompts are phrases or descriptions that tell Midjourney what to create. The bot breaks these phrases into tokens, compares them to its training data, and then uses them to generate realistic photos.

Prompts are the way we "talk" to Midjourney, providing detailed descriptions of what we want. To get accurate results, we need to be clear and specific. A prompt is a short phrase or sentence that describes the image you want.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

While a basic Midjourney prompt can be as simple as a single word, phrase or emoji, the structure and wording of your prompt directly impact the outcome.

When providing a prompt, be as detailed and descriptive as possible to align the results with your expectations. For photorealistic results, include styles, colors, moods, lighting, camera settings, lens descriptions and specific visual elements.

Aim to include a camera setting descriptor when writing a prompt. Typical photo prompt keywords include:

• Action shot
• Aerial view
• High resolution close-up shot
• Fashion photography
• Still life
• Landscape
• Product photography
• Award-winning photography
• Editorial photography
• Wide-angle shot

Use these prompts for realistic lighting results:

• Studio lighting: Lighting used for portraits, fashion photography and products.
• Backlighting: Dramatic lighting placed behind the subject to create a silhouette.
• Softbox lighting: Diffused lighting for soft, even illumination with minimal shadows.
• Bokeh lighting: Blurred light in the background for an artistic touch with a dreamy effect.

Advanced Midjourney prompts offer ultra-detailed results with image URLs, multiple text phrases and parameters. These include:

• Image prompts: You can add image URLs to give the Midjourney bot a reference image from which to work. Always put image URLs at the start of a prompt.
• Text prompt: This describes what you want Midjourney to create. Well-written prompts will help generate more realistic photos.
• Parameters: Always end a prompt with a parameter (more about this in the next section).

The key lies in supplying Midjourney with simple, short phrases. We advise using prompts that avoid long lists of requests and include reference images.

Here’s an example:

Instead of: Show me a hyper realistic portrait of many blooming poppies in sharp focus. Make them bright, vibrant orange and draw them in an illustrated style with colored pencils.

Try: Bright orange California poppies drawn with colored pencils or:

“/imagine bright orange California poppies drawn with colored pencils --style raw --chaos 0”

By carefully crafting prompts, you can guide Midjourney to produce more accurate and realistic images.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

2. Parameters

To produce good results, you’ll need to understand the Midjourney parameters—modifiers that help you fine-tune how the AI interprets your prompt. They control aspects like resolution, lighting or stylization.

In Midjourney, parameters allow you to fine-tune the outcome of your image prompts. For example, the --ar parameter controls the aspect ratio of the image. To use parameters, simply add them at the end of your prompt with a double dash, like --chaos 50. The chaos parameter controls variation; higher values (0-100) result in more differences between the generated images, while lower values make the images more similar. These parameters help customize the results based on your preferences.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

Some basic ones include:

• Aspect ratios: “--aspect” or “--ar” changes the aspect ratio of an image.
• Chaos: “--chaos <number 0–100>” changes how varied the results will be—a higher value produces a more unusual and unexpected photo, while a lower value produces a more realistic photo.
• Repeat: “--repeat <1–40>” or “--r <1–40>” creates multiple jobs from a single prompt. This is a useful parameter if you need to rerun a job multiple times and quickly.
• Style: “--style” controls how artistic or realistic a photo will look. A lower style setting produces cleaner and sharper images for a realistic look, while a higher setting makes the photos more imaginative. You can use the-- style raw option for a more neutral-looking image with minimal artistic influence. This reduces stylistic embellishments for a more accurate match.
• Style reference: You can use images as style references in your prompt to influence the style or aesthetic of the photography you want Midjourney to generate. To add a style reference to your prompt, include the --sref parameter with the image’s URL. For example: “/imagine A futuristic city skyline at dusk, with neon lights and tall skyscrapers --sref https://imageURL.jpg https://imageURL2.jpg”.
• Quality: “--quality <.25, .5, or 1>” or “--q <.25, .5, or 1>” determines how much rendering time you want to spend. The default value is always 1. Higher values use more “graphic processing unit” (GPU) minutes, while lower values use fewer GPU minutes. This means choosing a higher quality will not result in a high-resolution image.

Remember to add parameters at the end of your prompt.

Here are some examples of useful parameters in Midjourney:

• --chaos [value] Example: --chaos 50
Controls how much variety there is between the generated images. Higher values create more variation.
• --stylize [value]: Example: --s 500
Adjusts how beautiful the image looks. Lower values make the image less stylized.
• --ar [aspect ratio]: Example: --ar 16:9
Defines the aspect ratio of the image.
• --seed [number]: Example: --seed 12345
Recreates similar images by controlling the randomization factor.

3. Commands

Think of commands as instructions for Midjourney. You can use commands to create images, change settings and monitor user information. A commonly used command is “/imagine” to generate an image based on your prompt.

Commands in Midjourney are instructions that tell the AI what action to take. You access them by typing the “/” symbol in Discord, which brings up a list of available commands.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

Other commands offer additional functionality, such as:

• /ask gets you an answer to a question.
• /blend allows you to combine multiple images.
• /describe analyzes a photo and generates possible prompts based on it.
• /help shows you some basic information and tips about the Midjourney Bot.
• /shorten helps you simplify complex prompts.

Commands offer added flexibility for interacting with Midjourney and customizing its output to suit your needs.

4. Tools

Features within Midjourney allow you to adjust results further to create incredibly lifelike images. For example, tools like the Vary Region editor enable you to select a part of the image and modify it while keeping the rest unchanged.

In Midjourney, tools help you refine and enhance your images after they’ve been generated. They allow you to make adjustments like enlarging the image, zooming in, or modifying specific parts of it. For example, the Vary Region tool lets you make changes to a specific area of an image without altering the entire picture.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

You can use Upscale Tools to increase the resolution of your images or apply zoom effects for creative variations. 

Note: When you upscale images, your subscription’s GPU minutes are used faster. Creating a high-resolution image can take up to twice as many GPU minutes as generating an initial image grid.

A Creative Leader's Guide to Accelerating AI Adoption

Achieve AI Excellence

Take a quick quiz to assess your readiness and learn how to empower your team and strategically integrate AI into your workflows—in order to make the most of this transformative technology.

Get the guide

How to Get Photorealistic Images with Midjourney in 5 Steps

Now, let’s go over Emmanuel’s process for getting photorealistic images with Midjourney. It all starts here:

We need to speak to Midjourney in its own language to get the results that we want. There is not just one formula to get the expected results, but there are things that tend to work better than others. You won’t always need to apply all of these but if you are struggling to get realistic images these tips will help you to improve your results.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

Step 1: Define the level of Stylization (--s) you want

Midjourney tends to produce vibrant and polished images that can look almost too perfect or artistic. The --stylize parameter allows you to control how much of this artistic effect is applied. By adjusting this setting, you can guide Midjourney to create more realistic images or lean into a more stylized, artistic look.

For realistic images, lowering the --stylize setting helps reduce the "artificial" or "overly perfect" look that Midjourney often adds. You can experiment with different values until you find a balance that suits your preference, but a range of 500 to 750 generally works well for realism.

In the image below you can see how the images with --s 500 and --s 1000 have a good level of realism. Images with a higher --s number have more photographic beauty. With people usually --s 1000 tends to be too much.

Step 2: Indicate camera lenses and shot angles

We as humans are used to seeing a lot of photography, and our eyes are accustomed to perceiving certain styles derived from camera angles. Images generated while taking into account lens and angle settings make it easier for us to see them as real. It's not the same as looking at an image generated in 3D or artificially as it is looking like a photo taken with a camera.

So, tell Midjourney what camera angles you want in your image, such as whether it’s taken with a telephoto or a wide-angle lens, or from above or below. This helps Midjourney understand you want an image that looks like a photograph and gives you more realistic results. It also helps to specify if you want a studio photograph, an outdoor shot, or something taken in sunlight. These details will enhance the realism of your image and give you more customized results.

You can notice in the image on the right that the skin details are more realistic. Also, the use of the telephoto makes the depth of field more obvious and natural. If you know about photography you can use your own lenses in your prompt but if not you can use any of these in the list or just ask ChatGPT which is the best lens and camera for the type of photo you want to create.

Including details about illumination (luminous, sun-kissed, cold light, warm light), angle of the shot (frontal, lateral, from below, from above), and style of photography (outdoors, selfie, studio, fashion, cinematic) they help a lot to create realistic and also accurate results.

Step 3: Avoid using Niji models

Niji models are specialized versions of Midjourney that are designed to generate anime-style and illustrative images. They are excellent for creating hand-drawn, cartoonish, or vector-style artwork, with vivid colors and bold lines.

However, they are not ideal for producing realistic or photographic images. If you're aiming for something that looks lifelike or resembles a photograph, it's better to use Midjourney’s standard models rather than Niji, as Niji models focus more on creative, artistic rendering rather than realism.

If you want to create something that looks hand-drawn, vectorial, or similar, Niji will do a great job. But if you want to create images that look realistic and photographic, avoid using the Niji models. It would be really hard to get good results with them for this specific subject. To ensure this, go to the /settings of Midjourney and choose any of the models, preferably the most recent one, instead of the Niji model.

In the image on the left, the Niji model made the style illustration-like while the one on the right is more realistic. Also, more recent versions of Midjourney will tend to give more realistic results.

Step 4: Use style raw (--style raw) to get more accurate results adjusted to your prompt

In Midjourney, --style raw is a parameter you can add at the end of your prompt to reduce the artistic enhancements that Midjourney usually applies to its images. This means the AI will stick more closely to your original description, creating images that look more natural and less stylized or beautified.

It's especially useful when you're aiming for a more realistic, less polished result, as it removes some of the artistic flair that can make images look overly perfect or exaggerated.

Using --style raw at the end of your prompts can also help you get more accurate results. This parameter helps you to get images that are not that enhanced by Midjourney, so it will give you a less beautified result, which can help your images look more realistic. For this one it is harder to detect the difference though.

In this case, the parameter --style raw helps a lot with complex poses and scenarios where Midjourney doesn’t understand well what we want to get as a result.

Step 5: Avoid using low-resolution references when using --sref and--cref

If you need to apply a certain style to the images you are creating or get specific face features, ensure the images have great resolution and are not pixelated. Pixelated images could bring you low-quality results and less realistic photography.

The image on the right has better realism on the skin and shades due to the higher quality of the reference.

Bonus: Upscaling an image in Midjourney

Now, it is time to upscale the image. But what does upscaling mean? Well, usually AI images are created in relatively low resolution to save processing resources. Once you get the result you want, you'll need to enlarge the image to a higher resolution.

In Midjourney, there are two options: Upscale Subtle and Upscale Creative. In my experience, Upscale Subtle works best for faces that take up a large portion of the image, as it preserves the details without altering them much.

However, when the face or body is smaller in the image, Upscale Creative tends to perform better because it reconstructs details more effectively. Upscale Subtle can sometimes create a plastic effect on small faces since it tries to stay too close to the original, whereas Upscale Creative adds interpretation to improve resolution in those areas.

You can also get help from external upscallers like Krea.ai, Upscayl, Topaz, Magnific and many others. These bring a higher level of realism to your images, making the texture and shades of the skin more rich.

Example Prompts for Photorealistic Images in Midjourney

Let’s take a look at some photorealistic prompt examples by our expert, Emmanuel Canton:

1. Full-body shot of a man in Korean fashion

2. Woman in boho aesthetic with minimalist style

3. Close-up of white t-shirt on beige texture

4. Studio portrait of man in white Gildan t-shirt

5. Angled shot of a ceramic mug on concrete

6. Close-up of woman in unisex t-shirt with modern look

Common Pitfalls and How To Avoid Them

To create hyper-realistic photos with Midjourney, avoid these common pitfalls:

1. Writing very long prompts

Pitfall: Midjourney struggles to understand long and complex prompts.

There are two types of errors with prompts: too-short prompts which can be translated to missing information in camera lenses and illumination styles or too-long prompts can lead to Midjourney getting focused on irrelevant information and ignoring the most important parts of the prompt.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

Solution: Keep prompts for realistic images short and straightforward and provide clear context and essential details. When crafting your Midjourney prompts, consider:

• The subject: Person, animal, character, location or object.
• The medium: Photography, ultra-detailed painting, illustration, sculpture, doodle or tapestry.
• The environment: Indoors, outdoors, on the moon, underwater, in the city or inside a home with contemporary interior design.
• The lighting: Natural light (low contrast), dramatic lighting (high contrast), soft, ambient, overcast, neon, studio lights or warm glow.
• The colors: Vibrant, muted, bright, monochromatic, colorful, black and white or pastel.
• The camera setting: Shallow depth of field, macro lens, wide-angle lens, telephoto lens, fisheye lens or portrait lens.
• The mood: Calm, raucous, energetic or vintage style.
• The type of composition: Portrait, headshot, close-up portrait or birds-eye view.

2. Being afraid of using parameters

Pitfall: Initially, it’s understandable to feel overwhelmed and make common Midjourney mistakes.

To get any good results in Midjourney you need to enter open to the idea that you may not get initially the results that you expect and it is ok. After the first iteration, when you see the image, you start exploring how to correct certain things.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

Solution: Start by focusing on a few key parameters, such as aspect, chaos and quality. Your results will improve significantly once you understand what they do and how they work.

The --ar (or --aspect) changes the aspect ratio of an image and is typically written as two numbers separated by a colon. For example:

• Square (1:1) “--ar 1:1” is best for social media posts, avatars or profile photos because it displays well on most platforms without cropping.
• Landscape (16:9 or 3:2) “--ar 16:9” or “--ar 3:2” is best for wide scenes, wallpapers and YouTube thumbnails. 
• Portrait (9:16 or 4:5) “--ar 9:16” or “--ar 4:5” is best for social media stories, posters and full-body portraits. A portrait aspect ratio is ideal for vertical content, like Instagram Reels, YouTube Shorts or TikTok posts.

The "--chaos" or "--c" parameter affects the grid’s variety of images. A higher value creates more unusual results, while a lower value yields more reliable ones.

The “--quality” or “--q” parameter controls the time it takes to generate images. A lower value means faster generation but less detail. This parameter is helpful for quick results with new Midjourney prompts.

3. Unnatural poses or object placements

Pitfall: Midjourney may create photos with awkward or unnatural poses or objects positioned in a way that doesn’t make sense, breaking the illusion of realistic photography.

Solution: When prompting poses, ask for natural positions like “relaxed posture” or “walking naturally.” Provide clear instructions for objects such as “centered” or “on the table.” And don’t forget to give the person’s hands something to do!

Explore AI-Enhanced Creative Services

Unlock tomorrow's possibilities, today.
Learn more about our AI-Enhanced Creative
and AI Consulting Services.

Learn more

4. Lack of consistency in lighting and shadows

Pitfall: One of the main issues with AI-generated photography is inconsistent or unrealistic lighting and shadows, such as conflicting light sources or illogical shadows.

Solution: When creating Midjourney prompts, specify the lighting setup you want, such as “natural light,” “soft shadows” or “dramatic lighting,” and provide directional cues like “light from the left” or “sunset lighting” to help generate a more realistic portrait.

5. Inconsistent textures and details

Pitfall: A highly detailed photo needs the right textures, especially for skin, fabric or foliage. If the prompt is too vague, it might create inauthentic-looking images, especially in close-up portraits.

Solution: Make your Midjourney prompts very specific, using phrases like “high-resolution,” “fine detail” or “close-up texture.” Also, specify elements like “realistic fabric,” “skin with pores” or “detailed wood grain.”

6. Over-stylization leads to unnatural results.

Pitfall: A common issue with Midjourney’s algorithms is that they favor artistic styles, resulting in images looking overly stylized rather than realistic. This becomes problematic when the output includes exaggerated colors, unrealistic textures or unnatural lighting effects.

The most common mistakes are usually the extremes. Very high stylization --s 1000 or very low stylization --s 100. Using not clear enough references such as pixelated images, low photographic quality images, and too much variety of style in the references.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

Solution: When writing Midjourney prompts, use specific keywords related to realism, such as “hyper-realistic,” “high detail,” “true-to-life,” “natural light” or “photo-like.” You can also reduce the stylize parameter (“--stylize 50 or lower”) to minimize the artistic effects and maintain more realistic portraits.

7. Odd or distorted features

Pitfall: AI images, like those from Midjourney, may contain errors such as incorrect facial features, extra fingers or unnatural background objects.

Solution: Use clear, straightforward prompts to reduce any confusion. If you encounter consistent issues with specific elements, like hands or faces, break the prompt down into smaller, more specific commands focusing on them. You can also improve the output by making minor changes to your prompt each time until the distortion is fixed.

8. Being impatient

Pitfall: Don’t expect to capture realistic photos immediately, especially when using a new prompt. It may take several attempts for Midjourney to understand your intentions.

Solution: Provide feedback to Midjourney and adjust your prompts to help it improve. Over time, you’ll improve at creating realistic prompts, and Midjourney will understand your needs better.

How Superside Can Help With High-Quality AI Visual Content

AI tools like Midjourney make it possible to create high-quality images that resemble actual photographs at great speed and reduced cost. As AI’s capabilities expand, more businesses will tap into AI tools to stay competitive. However, we should always take this with a grain of salt:

Midjourney is a great tool capable of bringing excellent results but it gets complicated to obtain specific results such as realistic images if there is not a deep knowledge of all the factors involved in creating realistic images. And is not only about creating a realistic image but about thinking strategically and conceptually about what type of realistic image would convey the message better.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

But staying ahead also means knowing when to call in the pros. Does this sound like you? Partnering with Superside can help you leverage AI-enhanced design with the full support of our best-in-class human designers.

Midjourney doesn’t decide what pose, what environment, what target audience would feel represented in an image, and here is where Superside’s AI Certified creatives play a crucial role.

Emmanuel CantonAI-Enhanced Senior Graphic Designer, Superside

Book a call to bring your brand or campaign vision to life.

Empower AI on Your Team

Whether you're experimenting or scaling, Superside brings you a full range of AI-Enhanced Creative Services and Consulting to help you seize tomorrow's possibilities, today. Learn more about how we can help.

Book a call