Verified against Midjourney · 2026-08-05
Brief a multi-look fashion editorial story for Midjourney
A structured multi-look editorial fashion brief that holds one model, one location, and one visual identity constant across several distinct outfit changes — the actual structure a fashion lookbook or lookbook page needs, rather than four unrelated fashion images that happen to share a subject description.
The prompt
Ready to copy — highlighted parts are example details you can swap.
MODEL AND CONSISTENT FEATURES a tall model with sharp cheekbones, close-cropped dark hair, and an angular, editorial bone structure — restate this identically across every look in the story, since without a character reference locking the face, each look is a separate generation and will otherwise show a different-looking model per shot. LOCATION FOR THIS STORY a stark, minimalist concrete gallery space with tall industrial windows and diffuse grey daylight — one location, described identically across every look, since an editorial story reads as one continuous shoot specifically because the setting doesn't change between frames, only the pose and outfit do. LOOK FOR THIS SPECIFIC FRAME an oversized charcoal wool coat over a cream silk slip dress, black leather ankle boots — the one thing that should actually differ between frames in this story; be as specific about garment, texture, and color here as the model description is about the model, since a vague outfit description drifts more between generations than a precise one. POSE AND ATTITUDE FOR THIS LOOK standing at a three-quarter angle, chin slightly lowered, an unreadable, aloof expression PHOTOGRAPHIC TREATMENT shot on a Hasselblad medium-format camera, high-contrast black-and-white grade, sharp focus throughout — camera, lens, and grade, restated identically across every look in the story, for the same reason the location and model description are restated: a lookbook is judged on whether it reads as one photographer's consistent eye across the whole set, not four different photographic treatments loosely related by subject. CHARACTER CONSISTENCY --cref https://cdn.midjourney.com/jkl012-model-anchor.png --cw 40 — lower the character weight from the default so the reference locks the model's face and build without also forcing this frame's specific outfit description to fight against the reference image's original clothing. WHAT TO AVOID --no text, watermark, a second model, visible brand logos PARAMETERS --cref https://cdn.midjourney.com/jkl012-model-anchor.png --cw 40 --ar 4:5 --v 7 OUTPUT a tall model with sharp cheekbones, close-cropped dark hair, and an angular, editorial bone structure, an oversized charcoal wool coat over a cream silk slip dress, black leather ankle boots, standing at a three-quarter angle, chin slightly lowered, an unreadable, aloof expression, in a stark, minimalist concrete gallery space with tall industrial windows and diffuse grey daylight, shot on a Hasselblad medium-format camera, high-contrast black-and-white grade, sharp focus throughout --cref https://cdn.midjourney.com/jkl012-model-anchor.png --cw 40 --no text, watermark, a second model, visible brand logos --ar 4:5 --v 7 REPEAT FOR EVERY LOOK Generate this same structure once per outfit change, keeping the model description, location description, photographic treatment, character reference URL, and character weight identical across every look, and changing only the outfit and pose lines — this is what makes the finished set read as one lookbook rather than a loose collection of separately-styled shots.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
A fashion lookbook is judged on whether it reads as one continuous shoot with one model, one photographer's eye, and one location across several outfit changes — which is a structurally different problem from generating four independently strong fashion images that each happen to describe "the same model" in text, since text descriptions alone give each generation only a loose, non-specific resemblance to the others rather than a genuinely consistent face and build. Locking the model with --cref, and restating the location and photographic treatment identically across every look, is what turns four separate generations into one coherent story instead of four unrelated fashion images that share a vague family resemblance at best.
Lowering --cw below its default specifically for this workflow addresses the direct tension a lookbook creates for character reference: the default --cw 100 pulls clothing from the reference image along with the face, which actively fights the entire point of a multi-look story — different clothing in every frame. A lower weight biases the match toward the model's facial features and build while leaving room for each frame's specific outfit description to actually take effect, which is the opposite tuning from the earlier general-purpose character-consistency use case, where holding the reference's original outfit steady across scenes is often exactly what's wanted.
Restating the photographic treatment identically in every look's prompt, rather than assuming it carries over from an earlier generation, matters because each look in this workflow is its own independent generation with no memory of the others besides the shared --cref value — a subtly different camera or grade description on look three, even unintentionally, produces a photograph that looks like it came from a different shoot when the four looks are laid out side by side in a finished lookbook spread, which is exactly the failure a lookbook cannot afford.
Being specific about the outfit description, at the same level of detail the model description gets, addresses an asymmetry that's easy to fall into: it's tempting to over-invest description effort in getting the model right since that's the harder consistency problem, and under-describe the outfit since it changes every frame anyway — but a vague outfit description ("a nice coat") drifts unpredictably between generations in exactly the same way an underspecified subject drifts in any other Midjourney prompt, and a lookbook with inconsistent garment rendering quality across its looks reads as unpolished even when the model's face is perfectly consistent throughout.
Verified against
Midjourney v7 · 2026-08-05
Changelog
- 2026-08-05 — Initial publish, verified against Midjourney v7 --cref-anchored multi-look fashion editorial workflow.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
