The --sref parameter in Midjourney lets you lock a visual style and reuse it across every image you generate. You create one image you like — the colors, the texture, the overall feel — and then pass its URL as a style reference to every subsequent generation. The result is a set of images that look like they belong together, even when the subjects are completely different. For course creators, this solves the most common visual problem: a slide deck or course platform that looks like it was assembled from five different stock photo libraries.
What you’ll walk away with:
- A reliable style reference workflow for consistent imagery
- Fine-tuned style weight balancing consistency with variety
- A reusable reference library for your visual brand
Why style references matter for courses
Visual consistency is one of those things students notice without realizing they notice it. When every module header, lesson thumbnail, and worksheet illustration shares the same palette and texture, the course feels intentional. Professional. Like one person designed the whole thing — which, using style references, is essentially what happened.
The alternative is what most course creators end up with: a hero image from Unsplash, a few AI-generated illustrations with clashing styles, and some hastily cropped screenshots. Each image is fine on its own, but together they create visual noise. Students may not articulate it, but research on the aesthetic-usability effect from Nielsen Norman Group consistently shows that visual coherence influences how people perceive quality and trustworthiness.
Style references eliminate this problem at the source. Instead of trying to find images that happen to match, you generate images that are guaranteed to match because they share the same stylistic DNA.
Step by step: Building a visual system with --sref
Generate your reference image
Start with a prompt that captures the visual mood you want for your course. This is not about the subject — it is about the style. If you teach yoga, you might prompt something like "soft watercolor illustration of a peaceful garden, warm earth tones, gentle light." If you teach business strategy, maybe "clean minimal illustration of geometric shapes, cool blues and grays, crisp edges." Generate several variations and pick the one whose overall feel best matches your course's tone.
Spend time here. This single image becomes the visual foundation for everything else. Upscale the one you choose, then copy its Midjourney URL — you will use this URL in every subsequent prompt.
Apply --sref to your first content image
Now generate an image you actually need for your course. Write your prompt as usual — describe the subject, the composition, whatever details matter — and append --sref [your reference image URL] at the end. Midjourney will generate the new subject in the visual style of your reference. The colors will rhyme. The textures will feel related. The overall mood will carry through.
You can also adjust the strength of the style influence with --sw (style weight), which ranges from 0 to 1000. The default is 100. Higher values make the output adhere more tightly to the reference style; lower values give more room for the prompt's own direction. Start at the default and adjust if the results feel too constrained or too loose.
Generate module-specific imagery
Work through your course module by module. For each section, write a prompt that describes what that module is about — the subject changes, but the --sref stays the same. A five-module course might produce five header images with five different subjects that all look like pages from the same book. That visual continuity signals to students that your course is a coherent journey, not a collection of disconnected lessons.
Extend to supporting materials
Use the same style reference for worksheet headers, slide backgrounds, social media promotional graphics, and your course thumbnail. The --sref parameter works regardless of the output format or subject matter, so you can maintain the same aesthetic across every touchpoint — from the sales page to the final lesson. This is where the real leverage appears. A single style reference creates an entire visual brand without a style guide, a designer, or hours of manual adjustment.
Save your reference for future use
Keep a document with your reference image URL, the original prompt that generated it, and notes on any --sw adjustments you found worked well. If you create courses regularly, you might build a small library of reference styles — one for each course or course series. Midjourney URLs persist as long as your account is active, but having the original prompt means you can regenerate a similar reference if needed.
Prompts to try
These starting prompts are designed to produce strong, distinctive styles that transfer well across different subjects. Use whichever fits your course's personality, then pass the result as your --sref for all subsequent images.
- Warm and approachable:
soft watercolor botanical illustration, warm golden light, cream paper texture, hand-drawn quality, muted earth tones --ar 16:9 --v 6.1 - Clean and professional:
minimal flat illustration, cool slate blue and white palette, subtle grid structure, geometric shapes, modern corporate aesthetic --ar 16:9 --v 6.1 - Bold and energetic:
vibrant collage-style illustration, rich jewel tones, layered paper cutout textures, dynamic composition, playful but polished --ar 16:9 --v 6.1
The human layer
Midjourney with style references gives you visual consistency. It does not give you visual meaning. The tool cannot decide that Module 3 should feel more intense than Module 1, or that your worksheet on self-reflection deserves a softer palette than your worksheet on goal setting. Those decisions require your judgment about what your students are experiencing at each point in the course.
The strongest course imagery tells a visual story that parallels the learning journey. That means making deliberate choices about which subjects to depict, which metaphors to lean on, and when to break the pattern for emphasis. Style references handle the surface layer — the palette, the texture, the mood. You handle the substance: what the images actually communicate about the content they accompany.
Course creator tips
Generate more than you need, then curate
For each prompt, generate at least four variations and pick the strongest one. AI image generation is probabilistic — some outputs will capture the style reference better than others. Building a course's visual library is faster when you think of it as a selection process rather than a one-shot process. Generate a batch, choose the best, move on.
Match your reference style to your audience
A watercolor style signals warmth and approachability — good for wellness, coaching, or creative courses. A clean geometric style signals precision and professionalism — better for business, technology, or certification programs. Your students will form impressions of your course's rigor and personality from the imagery before they read a single word. Choose a reference style that sets accurate expectations.
Use consistent aspect ratios
Add --ar 16:9 (or whatever ratio fits your course platform's layout) to every prompt alongside your --sref. Consistent dimensions are just as important as consistent style for making a course look polished. Mismatched image sizes create visual clutter even when the colors and textures match perfectly.
What it gets wrong
Style references are not style clones. Midjourney extracts general characteristics — color tendencies, texture patterns, compositional habits — but it does not replicate every detail. Two images with the same --sref will feel related, but they will not look like they were cut from the same canvas. If you need pixel-level consistency (identical backgrounds, matching characters), you will need to do post-production work in a tool like Canva or Figma.
Text in images remains unreliable. Midjourney has improved significantly with v6, but generating clean, accurate text — lesson titles, labels, captions — is still hit or miss. Generate your images without text and add typography separately. This also gives you more flexibility to reuse images across contexts.
The learning curve is real but short. Understanding how --sref interacts with different prompt structures takes experimentation. Your first few attempts may not capture the reference style as strongly as you expect. Adjusting --sw values and simplifying your prompts usually resolves this within a session or two.
Frequently asked questions
Does --sref work on all Midjourney plans?
Yes, the --sref parameter is available on all paid Midjourney plans, including the Basic plan at $10/month. It does not consume additional credits beyond the normal image generation cost. You need to be on Midjourney v6 or later for style references to work reliably. The feature is not available on free trial generations.
Can I use --sref with images I did not generate in Midjourney?
Yes, you can pass a URL to any image as a style reference, including photographs, illustrations, or designs created outside Midjourney. The tool extracts the visual style — color palette, texture, composition tendencies — and applies it to your new generation. Results vary depending on how distinctive the source image's style is. A watercolor painting gives clearer style signals than a generic stock photo.
How many images can I generate with one style reference before the results start to drift?
There is no hard limit. The --sref parameter applies the same style extraction every time you use it, so the hundredth image should be as stylistically consistent as the second. Where drift happens is when you change your text prompts significantly — a prompt for "serene mountain landscape" and "chaotic city intersection" will produce different compositions even with the same style reference. Keep your prompts structurally similar for the tightest consistency.
Your visual system is ready — now build the course
A consistent visual identity makes your course feel like a single, intentional experience rather than a collection of lessons. With your style reference locked in, every module header and lesson image shares the same aesthetic. The next step is putting those images where students see them. On Ruzuku, you upload visuals directly into course steps — alongside video, text, downloads, and discussion. Your branded imagery becomes part of the lesson, not just a thumbnail on a landing page.
The result is a course that looks and feels designed from end to end, built with tools you already know, on a platform that keeps everything in one place.
Related guides
- How to Repurpose Course Content Using ChatGPT — turn existing material into new formats while maintaining your voice
- How to Write Social Media Posts for Your Course Using ChatGPT — pair consistent visuals with consistent messaging
- How to Create Course Images Using Midjourney — start here if you are new to Midjourney before diving into style references
- The Content Repurposing Stack — use your consistent imagery across repurposed content with Descript, Canva, and Buffer
- How to Create Your First Online Course — go from visuals to a live course your students can enroll in