Creating realistic AI images can be pretty tricky for beginners on Midjourney since the AI only picks up certain kinds of language. Prompts are how we “chat” with Midjourney, describing what we’re after. To get the results you want, you've got to be clear and specific. Basically, a prompt is just a short phrase or sentence that outlines the image you’re looking for.

If you’re just diving into Midjourney, creating prompts can be as simple as tossing in a single word, a phrase, or even an emoji. But how you craft and phrase your prompt can totally change the outcome.

The difference between a simple and a detailed prompt is huge. Here’s how it goes:

• Simple Prompt: "A photograph of a sunset over the gushing river."
• Detailed Prompt: "A close-up photograph of a vintage car in a desert landscape during golden hour."
• Using camera details: "A wide-angle shot of a crowded open-air market at sunset with a Canon 24mm lens."
• Lighting and setting: "A portrait in a studio setting with soft, natural light."

Being specific about the lighting and the scene can seriously help you get more spot-on results.

Then, there are parameters. It helps you define the size of the image, style characteristics, and the version of the AI model. But we are not here to discuss parameters. If you want to know more about it, please visit this tutorial.

A more interesting feature is the commands. They are like giving instructions to Midjourney. You can use them to whip up images, tweak settings, and keep tabs on user info. A popular command is “/imagine,” which lets you create an image based on your prompt. Type the “/” symbol in Discord, and a list of available commands will appear.

Some common examples of commands are:

• /imagine: Generates an image based on the prompt you provide.
• /blend: Combines two or more images into a single one.
• /shorten: Analyzes your prompt and suggest simpler versions.
• /describe: Creates potential prompts based on an image you upload.

These commands help you guide Midjourney in creating and refining your images.

After going through this tutorial, you’ll be able to:

1. Define the level of stylization
2. Use camera lenses and shot angles
3. Avoid using niji models
4. Get accurate results with your prompt by using the style raw parameter
5. Avoid low-resolution references when using –sref and –cref

Step 1 - Define the level of stylization

Midjourney often creates bright, polished images that sometimes seem almost too perfect or artistic. The --stylize parameter lets you control how much of this artistic effect is applied. Adjusting this setting allows you to guide Midjourney to produce more realistic images or go for a more stylized, artistic look.

For more realistic images, lowering the --stylize setting helps reduce the "artificial" or "overly perfect" appearance that Midjourney typically adds. You can experiment with different values until you find a balance that you like, but a range of 500 to 750 usually works well for realism.

The images with --s 500 and --s 1000 maintain a good level of realism. A picture with -–s 500 looks like the following:

‍

And the image with --s 1000 looks like this:

Images with a higher --s number have more photographic beauty, but when it comes to people, --s 1000 often tends to be too much.

Let’s say you want to generate an image of a car and a motorcycle on a busy street in Paris in the 1950s. Use the following prompt to get a photorealistic image that isn’t too shiny.

Prompt:

A vintage car and a vintage motorcycle roaming the crowded streets of Paris with a city background –s 700 –chaos 50

‍

That’s beautiful. Write your prompt with as many details as possible, and let the midjourney generate some beautiful images. 

However, to make it less photorealistic, lower the stylized number to below 500. This will generate a less realistic image. 

Prompt:

Vintage motorcycles with men riding roaming together the crowded streets of Paris with a clear city background –stylize 400

‍

‍

This came out pretty amazing. The level of detail is good. 

Step 2 - Use camera lenses and shot angles

We’re used to seeing lots of photos, and our eyes pick up on different styles based on camera angles. When you include lens and angle details in your images, they look more real. It’s different from a 3D or artificial image because it feels like an actual photo.

Equip Midjourney with the camera angles you want, like using a telephoto or wide-angle lens or shooting from above or below. This permits Midjourney to create images that look like real photographs. Also, specify if you want a studio shot, an outdoor scene, or something in sunlight. These details make your images more realistic and customized to what you’re looking for.

Use one or more of the top 10 lens and shutter speed combinations that photographers use to capture stunning images across various genres:

‍

‍

Let’s create a photo with the camera lens and shutter speed combinations. Let’s see what midjourney does with these additional details. 

Prompt:

A young brunette woman sitting on a cozy sofa shows a gentle smile as she reads a letter next to a wooden window. Shot with 85mm f/1.4 telephoto lens with a shutter speed of 75 seconds. The warm light from a candle lamp creates a cozy, intimate moment in lifestyle photography –style raw – v6.1 –s 800

‍

‍

In the image above, you can see the skin looks much more realistic. Using a telephoto lens also makes the depth of field appear more natural. If you’re into photography, feel free to specify your lenses in your prompt. If you have no idea about photography, you can choose from the available options or ask ChatGPT for the best lens and camera for the photo you want.

Adding details about lighting (like bright, sun-kissed, cool, or warm), camera angles (front, side, below, above), and photography styles (outdoor, selfie, studio, fashion, cinematic) really helps create images that look realistic and match what you’re aiming for.

Step 3 - Avoid using niji models

Niji models are special versions of Midjourney made to create anime-style and illustrated images. They’re perfect for producing hand-drawn, cartoon-like, or vector artwork with bright colors and strong lines.

But if you’re looking to make realistic or photo-like images, Niji isn’t the best choice. It’s better to use Midjourney’s standard models for lifelike pictures that resemble actual photos. Niji focuses more on creative and artistic styles instead of realism.

If you want to enable the niji model, click the settings icon at the right of the prompt box and go to the model area of the settings window. Select the niji model in the version dropdown.

‍

‍

Here’s what a niji model does when you use it with a prompt.

Niji Prompt:

A young 18 year old with pink dyed hair exhibits a mischievous grin while playing in a park. Shot with a 35mm prime lens f/4 on a tripod 2 seconds. The bubble catches the soft moonlight in a magical way creating an intimate whimsical mood in a street photography setting. --chaos 5 --niji 6 --style raw --stylize 800

Realistic Prompt:

A young 18-year-old with pink dyed hair exhibits a mischievous grin while playing in a park. Shot with a 35mm prime lens f/4 on a tripod for 2 seconds. The bubble catches the soft moonlight magically, creating an intimate, whimsical mood in a street photography setting. --chaos 5 --style raw --stylize 800 --v 6.1

‍

‍

In the left image, the Niji model gives it an illustration-style vibe, while the right one looks more real. Plus, the newer versions of Midjourney generally create more lifelike images.

Step 4 - Get accurate results with your prompt by using the style raw parameter

In Midjourney, you can use the --style raw option at the end of your prompt to tone down the artistic touches the AI usually adds to images. This makes the AI stick closer to what you described, resulting in more natural pictures that look less fancy or exaggerated.

This is especially handy if you want your images to appear more realistic and not too polished. It cuts out some extra flair that can make pictures seem overly perfect.

Adding --style raw to your prompts also helps you get more accurate images. Since the AI isn’t enhancing the picture as much, the results are simpler and look more real. However, the difference might be subtle and not always easy to notice.

Prompt without the style raw:

An action shot of a young woman with a slightly chubby build running in a park captured with a 200mm f/2.8 wide angle lens in 100 sec. The color tone features vibrant green and warm natural hues. The setting is a tree-lined park near a sea with a visible jogging path. The texture of her running gear shows smooth fabric with live motion. The mood is one of determination and fitness, with the woman appearing focused. The composition is a wide shot with the subject mid-stride and slightly off-center, illuminated by soft morning light casting long shadows. Shot in a candid style to emphasize natural movement and realism –v6.1 –s750

Prompt with the style raw:

An action shot of a young woman with a slightly chubby build running in a park captured with a 200mm f/2.8 wide angle lens in 100 sec. The color tone features vibrant green and warm natural hues. The setting is a tree-lined park near a sea with a visible jogging path. The texture of her running gear shows smooth fabric with live motion. The mood is one of determination and fitness, with the woman appearing focused. The composition is a wide shot with the subject mid-stride and slightly off-center, illuminated by soft morning light casting long shadows. Shot in a candid style to emphasize natural movement and realism –v6.1 –s750 –style raw

‍

‍

As you can see that the parameter --style raw helps create complex poses and scenarios that would otherwise be difficult for Midjourney to understand well. 

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

If you want to add a specific style or achieve certain facial features in your images, make sure they have high resolution and aren’t pixelated. Pixelated pictures can lead to low-quality results and make your photos look less realistic.

Prompt:

An elderly American man has a thoughtful gaze as he walks through a field holding a cane with his elderly wife, captured with a 50mm portrait lens, shutter speed f/2.4 in 20 sec. The scene is bathed in soft morning sunlight, creating a serene atmosphere in this lifestyle photography shot. --style raw --v 6.0 --s 750 --chaos 5 --sref https://thumbs.dreamstime.com/b/thoughtful-elderly-couple-stylish-cane-walks-park-fresh-air-grandfather-gray-hair-stands-against-backdrop-houses-189238312.jpg

‍

‍

Now, with a clearer photo:

Prompt:

An elderly man has a thoughtful gaze as he walks through a field holding a cane, captured with a 50mm portrait lens. The scene is bathed in soft morning sunlight, creating a serene atmosphere in this lifestyle photography shot. --style raw --v 6.0 --s 750 --chaos 5 --sref https://thumbs.dreamstime.com/b/senior-african-couple-loving-looking-one-another-over-white-background-33290513.jpg

‍

‍

Now this photo has better realism on the skin and deep shades due to the high quality of the reference photo. 

That’s all, folks! We hope you loved this tutorial. Go to Midjourney and create some photorealistic images. Use it in webpage headers, ad banners, or social media posts. Be proud of yourself, knowing you can generate your own images without spending much money.