Multi-Prompts, Permutations & Image Weights
When simple descriptive sentences fail to provide the surgical control you need, advanced syntax operators unlock exact mathematical weighting between competing concepts, batch exploration, and hybrid image-text blending.
1. Multi-Prompting with Concept Weights (::)
By default, Midjourney processes words sequentially. Adding double colons ::splits the prompt into isolated conceptual tokens, preventing words from bleeding into each other:
The model interprets "hot dog" as a single compound food noun: generating a frankfurter sausage inside a bread bun with mustard.
The model treats "hot" and "dog" as two separate entities: generating a canine animal panting under blazing sunshine or radiating thermal heat!
Assigning Numerical Weights:
You can append integers or decimals immediately after the double colon to set relative concept ratios:
In this prompt, the cityscape receives 3x the attention of the vintage cars, ensuring cars complement the architecture rather than dominating the frame.
2. Negative Concept Weighting (::-0.5)
While --no [element] is equivalent to setting a weight of -0.5, negative weights allow fine-tuning the exact degree of removal:
This forces the botanical garden to generate blue, purple, and yellow orchids while aggressively suppressing red hues without removing floral elements.
3. Permutation Prompts: Batch Combinations
Available on Pro and Mega subscription plans (and up to 10 jobs on Standard),Permutation Prompts use curly brackets {a, b, c} to launch multiple jobs simultaneously from a single command:
How Midjourney Multiplies Permutations:
The command above automatically unpacks 3 locations × 2 weather conditions = 6 separate generation jobs! A confirmation dialogue will display in Discord asking you to confirm launching the batch queue.
4. Image Weight: --iw [0–3]
When passing an image URL at the start of your prompt, the --iw (Image Weight) parameter determines how strongly that image influences the final render relative to your text description:
| --iw Value | Influence Balance | Ideal Use Case |
|---|---|---|
--iw 0.5 | Text prompt heavily dominates (90% text, 10% image). | Borrowing a loose color scheme while completely changing the subject. |
--iw 1.0 | Default balanced influence (50% image, 50% text). | Standard stylized variations of a reference sketch. |
--iw 2.0 – 3.0 | Reference image strongly locks composition and geometry. | Logo redraws, product photography preserving physical shape. |
5. Zero-Prompt Blending with /blend
If you don't want to write any text prompts at all, the /blend command allows you to upload 2 to 5 images directly. Midjourney analyzes the aesthetics, subjects, and color palettes of all uploads, synthesizing a cohesive hybrid image.
Advanced Syntax Key Takeaways
- Use
::to prevent words from blending together into compound nouns (e.g.hot:: dog). - Assign numerical weights (e.g.
subject::2 background::1) to mathematically balance composition dominance. - Use permutation brackets
{option1, option2}to test multiple styles or lighting variations in a single batch. - Adjust
--iwbetween 0.5 and 3.0 to dictate how strictly an uploaded reference image anchors the final render.
Explore Related Tools
Hand-picked utilities and calculators related to this tool.
