Style Reference (--sref): Match Any Aesthetic
- Apply --sref to match the color palette and mood of a reference image
- Control style influence strength with --sw
- Distinguish between --sref (style) and --oref (specific subject/character)
What Style Reference Does
Style Reference (--sref) lets you point Midjourney at an existing image and say: generate something in this aesthetic. It captures the visual qualities of the reference — the color palette, the lighting character, the texture feel, the overall mood — and applies them to your new image.
Critically, Style Reference copies style, not content. If you reference a photograph of a red barn at sunset, your output won't contain a red barn — it will inherit the warm orange-red tones, the moody directional lighting, and the rustic texture quality of that image. The subject is still entirely controlled by your text prompt.
This distinction matters. Use --sref when you want to match an aesthetic or visual language. Use Omni Reference (--oref, Lesson 7) when you want to keep a specific subject or character consistent across images.
Basic Usage
In Discord: add --sref [image URL] to the end of your prompt.
a portrait of a woman reading outdoors --sref https://example.com/reference.jpg
In the web app: click the style reference icon in the prompt bar and upload an image, or paste a URL. You can also drag and drop directly.
You can use multiple style references at once by listing multiple URLs:
--sref https://example.com/image1.jpg https://example.com/image2.jpg
When you stack references, Midjourney blends their aesthetics. Use this to create hybrid styles — the color palette of one image with the compositional feel of another.
Style Weight (--sw)
--sw [0–1000] — Controls how strongly the style reference influences the output. Default is 100.
- Low (25–75): Light influence. Your text prompt still drives most of the image; the reference adds a subtle tonal quality.
- Default (100): Balanced. Usually the right starting point.
- High (500–1000): Strong influence. The output will closely match the visual character of the reference. Your text prompt becomes secondary.
Start at the default and adjust based on results. If the output looks too similar to the reference image in ways you don't want, lower the weight. If the aesthetic isn't coming through strongly enough, raise it.
Practical Use Cases
Brand consistency: You have a brand visual language — specific color tones, a certain feel. Upload an existing brand image as --sref and new generations will stay within that visual system.
Matching a photographer's style: You love the moody desaturated look of a specific photographer. Upload one of their images and your generations will take on that character.
Replicating art styles without naming them: Some style names are hit-or-miss in prompts. Showing Midjourney the style via an image reference is often more reliable than trying to describe it in words.
Seasonal and mood consistency: Creating a set of images that all feel like they belong in the same autumn editorial? Set the same --sref across all generations.
Style Reference Codes
When Midjourney generates an image, it creates a style code for that specific output. You can reference this code directly in future prompts using --sref [code] instead of a URL. This means you can say "generate something in the same aesthetic as this image I made earlier" without needing to upload anything.
The style code appears in the image details panel when you click on any generated image. Saving style codes for results you like is a way to build a personal aesthetic library over time.
- --sref copies aesthetic qualities (colors, mood, lighting) from a reference image, not its subject
- --sw (0-1000, default 100) controls how strongly the style reference overrides your text description
- Midjourney generates a style code for every image — save these to build a reusable aesthetic library
Progress is saved in this browser only — clear your cache or switch device and it's gone. Create a free account and it follows you everywhere.