Sponsored by
Hey Creator,
Every prompt you run is a clean slate with zero memory of what came before. Great for brainstorming; terrible for a cohesive launch or social feed.
If your generations keep wandering, you need a visual anchor.
Let's look at how to lock in your exact aesthetic so your images actually look like they belong in the same universe.
You generate a batch of images for a campaign. The first one nails the vibe. The second is close. By the fifth, you're somewhere else entirely — different colors, different mood, different everything.
This isn't a skill problem. It's a defaults problem. AI image tools are built to be creative and varied — every prompt is a fresh generation with no memory of what came before. Great for exploration. A nightmare for brand consistency.
The fix for this is called Style Locking.
WHAT IS STYLE LOCKING?
Style Locking means anchoring every generation to a fixed visual reference — so your images consistently share the same palette, mood, and aesthetic, regardless of what the prompt says. Instead of describing your style in words, you hand the AI an image and say: this is what everything should look like.
Today we're covering how to do it in the two tools most creators are already using: Midjourney and Adobe Firefly.
HOW TO DO IT IN MIDJOURNEY
Midjourney gives you three reference tools. They do different things, and knowing which to reach for is what makes Style Locking actually work.
--sref — Style Reference: Controls the look. Add a reference image URL and your generations will match its aesthetic — colors, textures, lighting, mood. This is your visual brand anchor. Use it across every prompt in a campaign to keep everything in the same universe.
--cref — Character Reference: Controls the who. Pass it a character image and Midjourney carries that character's facial features, hair, and clothing into new scenes. Use it for brand mascots, recurring personas, or storytelling series.
Omni Reference (--oref) — V7 and above: The newest and most powerful of the three. Unlike --cref, which is limited to characters, Omni Reference works with anything — characters, objects, vehicles, creatures. The --ow parameter (1–1000) controls how strongly the reference influences the output. Worth knowing: it uses 2x GPU time compared to a standard V7 generation.
Using them together: --sref + --cref (or --oref) in a single prompt gives you the full Style Lock — same visual world, same character, every time.
One caveat worth knowing: Midjourney treats references as strong inspiration, not exact clones. Very specific details like complex logos may not transfer perfectly and could need a manual touch.
HOW TO DO IT IN ADOBE FIREFLY
Firefly approaches Style Locking at three levels — and the right one depends on how seriously you need consistency.
Style Reference: The quickest entry point. In Text to Image, go to the Styles section and upload any image as your reference. A Strength slider controls how closely outputs follow it — dial up for tight consistency, ease off for variation within the same visual language.
Style Kits: A step up. Bundle reference images, effects, prompts, and objects into a single shareable kit. Your whole team now generates on-brand content without manually setting up the same references every time.
Custom Models: The deepest level. Train Firefly directly on your own image library and it generates content that reflects your specific visual identity — not a general approximation. Currently available on enterprise plans, not individual Creative Cloud subscriptions.
The commercial safety advantage
Firefly is trained exclusively on licensed content. Using your own brand assets as style references carries no IP risk — which matters for client work and commercial campaigns.
WHICH ONE WHEN
| WHICH ONE, WHEN? | ||
|---|---|---|
| Midjourney | Adobe Firefly | |
| Best for | Artistic, character-driven content | Brand-consistent commercial work |
| Style tools | --sref, --cref, --oref | Style Reference, Style Kits, Custom Models |
| IP safety | No indemnification | Commercially safe |
| Starts at | $10/month | Included with Creative Cloud |
Many creators use both — Midjourney to develop the visual language, Firefly to produce it safely at volume.
TRY THIS WEEK
Pick one image that represents your visual style — a post that performed well, a design you're proud of, a mood board image that captures your brand.
On Midjourney: Upload it to Discord, copy the URL, and add --sref [URL] to your next three prompts. Notice how the aesthetic carries through even as the subject changes.
On Firefly: Open Text to Image, go to Styles, upload your reference, set Strength to around 70, and run five prompts. Compare those outputs to your last five unanchored generations.
That's your Style Lock. Once you find one that works, you'll use it on everything.
The World's Biggest Dev Event Hits Silicon Valley
WeAreDevelopers World Congress comes to San José, CA — September 23–25, 2026. 10,000+ developers, 500+ speakers, and the full software development lifecycle under one roof, in the heart of Silicon Valley.
Kelsey Hightower. Thomas Dohmke (fmr. CEO, GitHub). Christine Yen (CEO, Honeycomb). Mathias Biilmann (CEO, Netlify). Olivier Pomel (CEO, Datadog). The people actually building the tools you use every day — all on one stage.
AI, cloud, DevOps, security, architecture, and everything real builders ship with. Workshops, masterclasses, and the official congress party.
