Midjourney's --sref parameter lets you extract the visual style from one image and apply it to every new image you generate. You create a single reference image you love, then use it as the style seed for all your module headers, lesson thumbnails, and sales page graphics. The result is a cohesive visual identity across your entire course — without hiring a designer or learning Photoshop.
What you’ll walk away with:
- Consistent visual style using --sref parameter
- A reference library and settings for future content
- Professional visuals that feel designed by one person
Why visual consistency matters for course creators
I've seen this problem hundreds of times in our Help Scout support queue: course creators struggling to make their materials look like they belong together. We've fielded hundreds of support conversations about brand color matching alone — it's one of the most common frustrations creators face. A lesson thumbnail that looks hand-drawn next to a module header that looks photographic next to a sales page banner that looks like clip art makes your course feel cobbled together, even if the content is excellent.
Visual consistency signals professionalism and care. It tells your students that this course was designed intentionally. And until recently, achieving that consistency required either design skills, expensive tools, or hiring a freelancer. Midjourney's style references change that equation.
How style references work
Midjourney V7 introduced the --sref parameter, which extracts the visual style from a reference image and applies it to your new prompt. Think of it as copying the "feel" of an image — the color palette, lighting, texture, level of detail, and artistic treatment — while generating completely different subjects.
The --sw (style weight) parameter controls how strongly the reference influences the output. It ranges from 0 to 1000, with 100 as the default. Lower values give you a subtle nod toward the reference style. Higher values make the output closely mirror it. For course imagery, I'd start around 200-400 — enough for clear visual cohesion without making every image look identical.
The grid-then-split workflow
E-learning designer Christy Tucker recommends a workflow that's particularly smart for course creators who need a set of related images: generate all your images in a single grid first, then split. Here's why this works.
When you generate a grid of four images in one Midjourney prompt, those four images share internal consistency — similar lighting, similar color temperature, similar level of detail. Pick the best one from the grid, upscale it, and use that image as your --sref reference for all remaining images. This gives you a visual anchor that's been "tested" in context alongside other variations.
For a course with 8 modules, you might generate one grid prompt for "module header illustrations," select your favorite, then create 7 more images using that one as the style reference. The result is a set of module headers that look like they were designed together — because in a sense, they were.
Prompts to try
Module header set
watercolor illustration of [your module topic], soft warm palette, educational, clean white background --ar 16:9 --sref [your reference URL] --sw 300
Lesson thumbnails
minimal flat illustration representing [lesson concept], centered composition, course thumbnail style --ar 1:1 --sref [your reference URL] --sw 400
Sales page hero
aspirational lifestyle scene showing [your student's desired outcome], natural lighting, warm and inviting --ar 3:2 --sref [your reference URL] --sw 250
The human layer
AI-generated images require human judgment at every step. Your first decision — what style to reference — sets the tone for your entire course. That's not a decision you want to automate. You know your students. You know whether your audience responds better to warm watercolors, clean vector illustrations, or moody photography.
Even with style references, Midjourney won't nail every image on the first try. You'll generate some images where the style drifts, where the composition doesn't work for a header, or where the subject doesn't quite match your lesson content. Plan to generate 3-4 variations per image and pick the best. The style reference gets you in the right neighborhood — your eye gets you to the right address.
What it gets wrong
- Text in images — Midjourney still garbles text. If your module headers need readable titles, generate the background in Midjourney and add text in Canva or Figma. Or consider Ideogram for text-heavy visuals.
- Exact brand colors —
--srefcaptures the general palette, not your exact hex codes. If you need precise brand color matching, you'll need to adjust in post-production. - Consistent human faces — style references maintain artistic style, not character identity. If you need the same person appearing across images, you'll see variations in features.
- No free tier — Midjourney starts at $10/month (Basic plan). Standard is $30, Pro is $60. All plans include commercial rights and unlimited relaxed-mode generations. If you're on a tight budget, consider free alternatives like Ideogram's free tier first.
Frequently asked questions
Does Midjourney have a free tier?
No. All plans require a paid subscription, starting at $10/month for Basic. Each plan includes commercial usage rights, so you can use generated images in your course materials and sales pages without additional licensing.
What is the --sref parameter in Midjourney?
It stands for "style reference." You provide a reference image URL, and Midjourney extracts the visual style — color palette, texture, lighting, artistic treatment — and applies it to your new prompt. Different subjects, same visual feel.
Can I use multiple style references in one prompt?
Yes. Midjourney V7 supports multiple --sref URLs, each with optional weighting. This lets you blend visual elements from different references — though for course consistency, sticking to one reference usually produces cleaner results.
Your course visuals need a home
Once you've built a consistent visual library, you need a platform that displays them well. Ruzuku's course builder lets you upload images directly into lessons and module headers, with your visuals appearing alongside discussions, exercises, and materials — everything in one place for your students.
Related guides
- How to Create Course Visuals with Readable Text Using Ideogram — when you need text that's actually legible in your images
- How to Create Custom Course Icons Using AI — icon sets for module navigation and visual systems
- How to Create a Course Brand Kit in Canva — establishing colors, fonts, and templates for your course
- How to Build a Visual Brand for Your Course Using AI — the broader strategy for AI-assisted course branding