How to Use Midjourney's Multi-Prompt Syntax to Control Concept Weight
Most Midjourney prompts are written as a continuous phrase, and the model interprets the whole thing as one blended concept. Multi-prompt syntax breaks that into separate, independently weighted components. It gives you a level of compositional control that continuous prompts cannot provide, particularly for images where two or more distinct elements need to coexist without blending into each other.
The syntax is simple. The results it enables are significantly more precise than standard prompting for the specific cases where it matters.
The Basic Syntax
In a standard Midjourney prompt, everything is parsed as a single concept. The model blends and associates all the words together. Adding a double colon :: between sections tells Midjourney to treat each section as a separate, independent concept:
concept one :: concept two
With no numbers after the colons, each concept has equal weight. Adding a number immediately after the double colon assigns a relative weight to that concept:
concept one ::2 concept two ::1
Here, concept one has twice the influence on the output as concept two. Weights are relative, not absolute. ::2 ::1 produces the same ratio as ::4 ::2 or ::6 ::3.
Negative weights
Assigning a negative weight tells the model to actively suppress that concept in the output:
forest landscape ::1 snow ::-0.5
This generates a forest landscape while actively suppressing snow characteristics. Negative weights work differently from the --no parameter: --no excludes specific objects or elements, while negative weights suppress a concept's visual influence across the whole image. Both can be used together.
When Multi-Prompt Syntax Is More Useful Than Standard Prompting
When two concepts would blend into a hybrid
This is the most important use case. Midjourney naturally blends concepts that appear in close proximity in a prompt. "Cyberpunk forest" tends to produce either a cyberpunk scene with some tree elements or a forest with some neon elements, but rarely a clear visual separation of both ideas. Separating them:
cyberpunk city ::1.5 ancient forest ::1
Treats each as a distinct visual component that must coexist in the image rather than merging into one hybrid concept. The result is more likely to show both elements clearly present rather than a blended middle ground.
When you want to emphasize one part of the scene over another
Without weight control, every word in a prompt receives roughly proportional influence. If you want the background to be a minor supporting element rather than a competing visual equal to the subject:
lone figure on mountain summit ::3 dramatic storm clouds ::1 vast landscape ::1
The figure gets three times the compositional weight of either the clouds or the landscape. It will be more prominent, more detailed, and more clearly the primary subject.
When prompt word order is producing wrong results
Midjourney generally gives more weight to words earlier in a prompt. If you want a later concept to compete equally with or dominate an earlier one, multi-prompt weighting overrides the position-based influence:
warm autumn forest ::1 single crow in foreground ::2
The crow gets more compositional weight than the forest despite being described second.
Practical Weight Ranges
Weights do not need to be large numbers. A 2:1 ratio is already a strong compositional difference. Practical ranges:
- Equal weight (::1 ::1 or no numbers): Both concepts compete equally. Good for images where you want genuine visual balance between two distinct elements.
- Moderate emphasis (::2 ::1): One concept is noticeably dominant without completely suppressing the other. Most common ratio for main subject vs supporting environment.
- Strong emphasis (::3 ::1 or ::4 ::1): The dominant concept overwhelms the secondary. The secondary may be visible as a background or contextual element only.
- Suppression (negative weights): Actively removes influence of the suppressed concept. Use sparingly; very strong negative weights can destabilize the composition.
