Verified against Midjourney · 2026-08-01
Blend two distinct concepts with explicit multi-prompt weights
A multi-prompt (::) recipe for deliberately fusing two named concepts — a creature hybrid, a two-culture architectural style, a genre mashup — with explicit numeric weights, instead of writing one blended sentence and hoping Midjourney splits its attention the way you intended.
The prompt
Ready to copy — highlighted parts are example details you can swap.
FIRST CONCEPT a red fox SECOND CONCEPT a stained-glass cathedral window WHAT KIND OF BLEND THIS IS a physical hybrid — a fox whose fur and body markings are made of stained glass panels, translucent and lit from within, not a fox standing near a window — state plainly whether this is a physical hybrid (one creature/object with traits from both), a stylistic fusion (one subject rendered through the visual language of two different traditions), or a compositional pairing (both concepts present in one frame, interacting) — multi-prompt weighting behaves differently depending on which of these you actually mean, and being explicit about which one prevents a blend that reads as neither. WEIGHT BALANCE 2::1, fox anatomy leading, stained-glass as the surface material rather than a co-equal subject — a roughly equal split (1::1) treats both concepts as co-dominant; an uneven split (3::1) treats the first as the dominant subject with the second folded in as a strong secondary influence rather than an equal partner. Pick the ratio based on which concept should visually lead. SHARED SETTING AND STYLE standing in a dark forest clearing at dusk, light glowing outward from within its own body — this part is not split by :: and should not repeat either concept's name, since anything stated outside the weighted segments applies evenly to the whole image regardless of the weight split. MULTI-PROMPT SYNTAX a red fox::2 a stained-glass cathedral window::1 standing in a dark forest clearing at dusk, light glowing outward from within its own body WHAT TO AVOID --no text, watermark, a literal window frame, actual glass shards on the ground PARAMETERS --ar 1:1 --v 7 OUTPUT a red fox::2 a stained-glass cathedral window::1 standing in a dark forest clearing at dusk, light glowing outward from within its own body --no text, watermark, a literal window frame, actual glass shards on the ground --ar 1:1 --v 7 IF THE BLEND COLLAPSES TO ONE CONCEPT If the result looks like concept A with only a token nod to concept B (a color, a texture) rather than a genuine fusion, the weight ratio was too uneven for what you actually wanted — move it toward 1::1 and re-run rather than adding more adjectives describing concept B, since the imbalance is a weighting problem, not a description problem. If the result looks like two separate things awkwardly touching in one frame rather than a true hybrid, the blend relationship you actually wanted was probably a physical hybrid, not a stylistic fusion — restate a physical hybrid — a fox whose fur and body markings are made of stained glass panels, translucent and lit from within, not a fox standing near a window explicitly and re-run.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Midjourney's :: syntax splits a prompt into separately-weighted segments the model optimizes toward independently before combining them, which is mechanically different from writing one unified sentence describing a hybrid — a single sentence like "a fox made of stained glass" asks the model to satisfy one combined description as best it can, while concept_a::2 concept_b::1 explicitly tells it how much relative attention each named concept deserves in the final composition. This distinction matters most exactly when a plain-sentence blend keeps collapsing toward one concept dominating with only superficial nods to the other — a common failure because a single sentence gives the model no explicit signal about which noun should structurally dominate versus which should just flavor the surface, so it defaults to whichever concept is more common in its training data for that kind of scene.
Separating "what kind of blend this is" from the weighting itself targets a distinct, often-overlooked failure: even a correctly weighted multi-prompt produces a different-looking result depending on whether the intended blend is a physical hybrid, a stylistic fusion, or a compositional pairing, and the :: syntax alone does not disambiguate which one was meant — it only controls relative emphasis, not the nature of the relationship between the two weighted concepts. Two users running the identical fox::2 stained-glass window::1 prompt with different unstated assumptions about which blend type they wanted will judge the same output completely differently — one calling it a success, the other calling it a failure — because the ambiguity was never in the weights, it was in an assumption neither of them wrote down.
The diagnostic guidance for when a blend collapses toward one side, versus when it reads as two things merely touching, gives a concrete, checkable next step instead of the usual response to a disappointing multi-prompt result — rewriting the whole prompt from scratch. A collapse toward one concept is specifically a weight-ratio problem, correctable by moving the ratio toward 1::1 without touching a single word of description; a juxtaposition-instead-of-fusion problem is specifically a blend-relationship problem, correctable by restating what kind of merge was intended. Treating these as two different, separately diagnosable failure modes — rather than one generic "the blend didn't work" bucket — is what turns multi-prompt experimentation from trial-and-error into a workflow with an actual next step after a disappointing grid.
What you get back
An image where fox anatomy structurally dominates the silhouette and pose, with stained-glass panel texture, leading, and internal glow replacing fur across the body — a genuine hybrid rather than a fox posed near a window, with the 2::1 weighting visibly favoring the fox as the leading form.
Verified against
Midjourney v7 · 2026-08-01
Changelog
- 2026-08-01 — Initial publish, verified against Midjourney v7 multi-prompt (::) weighted concept blending.
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
