How to Use the MidJourney Omni Reference Feature in Under 20 Minutes

MidJourney’s Version 7 “Omni Reference” feature is transforming how creators maintain visual consistency in AI-generated images. This powerful tool lets you include specific characters, objects, vehicles, or creatures across multiple images with unprecedented control and consistency.

This powerful tool lets you maintain perfect visual consistency across all your marketing materials by incorporating your products, branded characters, or even yourself into different scenes and contexts. Imagine showcasing your handcrafted jewelry in multiple seasonal settings, placing your restaurant’s signature dish in various styled photography layouts, or creating consistent character-based branding across your social media—all with remarkable accuracy and without hiring a photographer or designer for each variation.

If you’ve been frustrated by inconsistent results when trying to include the same character or object in different scenes, you’re in the right place. This step-by-step guide will have you mastering Omni Reference in minutes—no coding or technical expertise required.

What we’ll be using in this how-to guide:

• MidJourney Version 7 – The latest version with Omni Reference capability
• One reference image – A clear photo of the character, object, or creature you want to include
• Basic prompting knowledge – Familiarity with MidJourney’s prompt structure

What is Omni Reference?

At its core, Omni Reference is MidJourney’s way of letting you point to an image and say “Put this in my picture.” It allows you to incorporate specific visual elements from a reference image into your generated output with remarkable consistency.

You can use Omni Reference for:

• Characters (including yourself)
• Objects (from simple items to complex products)
• Vehicles (cars, bikes, spacecraft, etc.)
• Non-human creatures (pets, fantasy beings, etc.)

This feature replaces the previous Character Reference (–cref) parameter with a more versatile and powerful system, offering significantly higher consistency rates and better integration with different art styles.

Step 1: Prepare Your Reference Image

What You’ll Do: Select a high-quality reference image that clearly shows what you want to include in your generated images.

How To:

1. Choose a clear image with good lighting and minimal background distraction
2. For faces/characters:
   • Front-facing or three-quarter view works best
   • Clear visibility of facial features
   • Close-up shots typically perform better than distant ones
3. For objects:
   • Clean background (white or transparent ideal)
   • Multiple angles can help (though you’ll only use one at a time)

Tip: The quality of your reference image significantly impacts results. For characters, portrait-style photos with neutral expressions work exceptionally well.

Step 2: Access Omni Reference in MidJourney

What You’ll Do: Learn how to access and activate the Omni Reference feature in both the Web UI and Discord.

In the Web UI:

1. Navigate to the MidJourney web interface
2. Click the image icon in the prompt box
3. Either drag and drop your reference image or click to browse your computer
4. Add your prompt text describing the scene or context you want
5. Adjust the Omni Weight parameter (more on this later)
6. Generate your image

In Discord:

1. Use the /imagine command
2. Include your prompt describing the desired scene
3. Add the --oref parameter followed by the URL of your reference image
4. Optionally include the --ow parameter to adjust the Omni Weight
5. Send your command

Step 3: Configure Your Omni Weight Parameter

What You’ll Do: Learn how to adjust the crucial Omni Weight parameter to control how strongly the reference image influences your generated images.

How To:

1. In the Web UI:
   • Find the Omni Weight slider after uploading your reference image
   • Adjust to your desired setting (see recommendations below)
2. In Discord:
   • Add the --ow parameter followed by a number
   • Example: --ow 100 for default strength

The Omni Weight parameter controls how strongly the details from your reference image are transferred to your generated output:

• Parameter name: --ow in Discord or a slider in the Web UI
• Range: 1 to 1,000
• Default: 100

Finding the Right Weight:

Weight Range	Best For	Notes
50-130	General use, facial consistency	The “sweet spot” for most applications
200-300	Better detail retention	Good middle ground
400+	Clothing replication, intricate details	Higher fidelity, especially for specific elements
500+	Maximum detail transfer	May make results less predictable

Tip: Start with the default (100) for your first attempts, then adjust based on results. The optimal weight varies depending on your specific reference and desired outcome.

Step 4: Craft Your Prompt Effectively

What You’ll Do: Learn how to write prompts that work well with Omni Reference for different subject types.

How To: Create a prompt that includes:

1. A clear description of the scene or context you want
2. Any specific details about how the reference should appear
3. Style descriptors (realistic, anime, watercolor, etc.)
4. The Omni Reference parameter and weight

For People and Faces:

Example prompt:

Portrait of the person in a cyberpunk setting, neon lights, dramatic lighting, detailed, 8k --oref [URL] --ow 100

Tips:

• Include key attributes that match your reference (hair color, clothing, etc.)
• Use descriptive terms like “portrait” or “character” to signal importance
• For photorealistic results, include terms like “photorealistic” or “detailed photograph”

For Objects:

Example prompt:

The object displayed in a minimalist product photography setup, white background, professional lighting, detailed --oref [URL] --ow 100

Tips:

• The default weight (100) often works well for objects
• For complex or unusual objects, mention specific features
• Use terms that match your object’s purpose (gadget, tool, artifact, etc.)

For Pets and Non-Human Creatures:

Example prompt:

The [animal type] in a forest setting, natural light, detailed fur texture, majestic pose --oref [URL] --ow 400

Tips:

• Higher Omni Weights (around 400) work better for non-human subjects
• Include the breed or species name in your prompt
• Use descriptive terms about fur/feather/scale texture

Step 5: Generate and Refine Your Results

What You’ll Do: Learn how to generate your image, evaluate the results, and make adjustments to get closer to your vision.

How To:

1. Generate your initial grid of four images using your prompt and reference
2. Evaluate the results – look for:
   • Facial/object similarity to your reference
   • Appropriate integration into the scene
   • Overall aesthetic quality
3. Select the best option from your grid (or regenerate if needed)
4. Use variations to fine-tune:
   • Click V1-V4 under a promising result for subtle variations
   • Try “Strong Variation” for more diverse options
5. Adjust your approach based on results:
   • If likeness is poor: try increasing Omni Weight
   • If integration is poor: refine your prompt language
   • If style clashes: add more style descriptors

Remember: With current success rates estimated between 25-45%, you may need multiple attempts to get the perfect result. Don’t get discouraged!

Step 6: Explore Advanced Techniques

What You’ll Do: Take your Omni Reference skills to the next level with these advanced techniques.

Combining with Other Features:

Omni Reference works well with:

• Style references – combine your character with specific artistic styles
• Mood boards – incorporate your reference into complex visual themes
• Art style variations – try prompting for “watercolor,” “anime style,” etc.

Example Advanced Prompt:

Character in the style of Alphonse Mucha, art nouveau, decorative elements, ornate, flowing lines, pastel colors --oref [URL] --ow 150

Troubleshooting Common Issues

What You’ll Do: Learn how to identify and solve common problems when using Omni Reference.

Problem: Reference Not Appearing Clearly

Possible Solutions:

• Increase the Omni Weight (try 200-400)
• Make sure your prompt doesn’t contradict your reference (e.g., specifying different hair color)
• Use a clearer, higher-quality reference image
• Try a closer-up shot if using a face/character reference

Problem: Unexpected Elements Appearing

Possible Solutions:

• Check your reference image for any hidden or background elements
• Use a reference with a clean, solid, or transparent background
• Be more specific in your prompt about what you want to include/exclude
• If background elements from a “transparent” reference appear, try a different image format

Problem: Technical Errors

Possible Solutions:

• Verify you’re using Version 7 (Omni Reference doesn’t work with older versions)
• For loading errors, try re-uploading your reference or using a different image format
• If nothing works, try a completely different reference image

Important Limitations to Know

Before diving deeper, be aware of these current constraints:

Limitation	Details
Single reference only	Only one reference image per prompt
Version restriction	Only works with MidJourney Version 7
Incompatible features	Does not work with inpainting, outpainting, vary region, pan, or zoom out
Mode restrictions	Not available in Fast Mode, Draft Mode, or Conversational Mode
Resource usage	Consumes double the GPU time compared to standard renders
Success rate	Currently estimated between 25-45%, requiring multiple attempts

Tip: Being aware of these limitations will save you time and frustration. Plan your projects accordingly, and always be prepared to try multiple approaches.

Creative Projects to Try

What You’ll Do: Get inspired with these practical projects that showcase the power of Omni Reference.

Project 1: Character Consistency Series

What You’ll Need:

• A clear portrait photo
• 3-4 different scene ideas

Steps:

1. Select a high-quality portrait as your reference
2. Create a series of prompts placing your character in different scenarios: [Character] exploring a jungle temple, adventure, cinematic lighting --oref [URL] --ow 100 [Character] in a futuristic cityscape, cyberpunk, neon lights --oref [URL] --ow 100 [Character] as a fantasy wizard casting a spell --oref [URL] --ow 100
3. Create a collection of images with consistent character appearance

Pro Tip: For best results, maintain similar framing (portrait, full body, etc.) across your series to encourage consistency.

Project 2: Product Visualization

What You’ll Need:

• A clean photo of your product against a white background
• Marketing context ideas

Steps:

1. Photograph your product against a clean white background
2. Create prompts that showcase your product in different contexts: [Product] on a minimalist desk in a modern office --oref [URL] --ow 150 [Product] in an outdoor setting with natural lighting --oref [URL] --ow 150 [Product] as part of a lifestyle scene with people using it --oref [URL] --ow 150
3. Use the results for marketing materials, social media, or concept exploration

Pro Tip: For product images, adding specific lighting descriptions in your prompt can dramatically improve results.

Additional Resources

• Complete MidJourney V7 Documentation
• Omni Reference Community Examples
• AppAthena’s Guide to AI Agents

Final Tips

• Start with simple scenes before attempting complex compositions
• Be patient – current success rates are between 25-45%, so multiple attempts are normal
• Document what works – when you find effective settings, save them for future reference
• Join the community – share your findings and learn from others exploring this feature

As with all AI tools, Omni Reference will continue to evolve. The techniques that work best today may be different tomorrow, so stay curious and keep experimenting!