If you need the syntax and values right now, here’s what you need to know: In Midjourney, you control how strongly a reference image influences your result using the --iw parameter. The valid range for V7, V6, and Niji 6 is 0 to 3, with a default value of 1.
A higher --iw value makes the output closer to your reference image, while a lower value lets your text prompt dominate the generation. Think of it as a slider between image-driven and text-driven results.
Here’s a basic example to get started:
/imagine <image_url> a moody cyberpunk portrait --iw 2
This prompt gives your reference image twice the default influence, creating stronger adherence to the composition and colors while still respecting your text description.
⚡ Pro Tip
Combine
--iwwith prompt chunk weights (like::1.5or::-0.5) to target specific elements while keeping your overall composition anchored to the reference image. This combination unlocks precise control over your generations.
Start by testing values like 0.5, 1, 2, and 3 to develop an intuitive feel for how the parameter shifts your results. For more applied workflows, check out our Midjourney Text-to-Image guide. Keep your prompts clean and experiment systematically to understand the balance between image and text influence. If you’re planning regular experiments, explore our Midjourney subscription tips to save on your plan.
What is « Image Weight »?
Understanding what image weight actually does is crucial for mastering Midjourney’s reference system. This section breaks down the --iw parameter, clarifies how it differs from related controls, and explains why it matters for achieving consistency and creative control in your generations.
The --iw Parameter in Plain English
The --iw parameter sets the image-to-text influence ratio in your generation. At --iw 0, Midjourney essentially ignores your reference image and relies almost entirely on your text prompt. At --iw 3, the model follows your reference image as closely as possible within its creative constraints.
The default value is 1, which provides a balanced relationship between your reference image and text description. This middle ground works well for most use cases, but adjusting up or down gives you fine-grained control over the final output.
Think of it as telling Midjourney: « How important is this reference image compared to what I’m writing? » The answer determines everything from composition and lighting to color palette and mood.
--iw vs. Prompt Weights vs. Style Weight
Midjourney offers several « weight » parameters, and understanding the differences prevents confusion:
--iw (Image Weight) controls the influence of image prompts versus text prompts. It’s the main tool for balancing reference images against your written description.
Prompt weights (::) control the relative importance between different text chunks in your prompt, and they can also be applied to multiple images. For example, vibrant sunset::2 calm ocean::0.5 tells Midjourney that the sunset matters more than the ocean.
Style reference (--sref + --sw) is a completely separate system that borrows stylistic elements from reference images without affecting composition. The --sw parameter ranges from 0 to 1000 with a default of 100, and it specifically controls how strongly the style reference affects your output.
These three systems work independently but can be combined strategically for sophisticated control. You might use --iw for composition, prompt weights for element emphasis, and --sw for artistic style—all in a single generation.
Version Nuances (V7, V6, Niji 6)
Midjourney’s different models handle image weight with slight variations. Knowing these supported ranges and behavioral differences ensures your prompts work correctly across versions and helps you troubleshoot unexpected results.
Supported Ranges & Defaults
The official documentation confirms that V7, V6, and Niji 6 all support an --iw range of 0 to 3 with a default value of 1. This consistency across recent models makes it easier to develop reliable workflows.
If you encounter tutorials mentioning different ranges (some older content references 0-2 or other values), prioritize the official Midjourney documentation. The platform has evolved, and outdated information can lead to frustrating results or error messages.
Related Parameters That Affect Results
Image weight doesn’t work in isolation. Several other parameters interact with --iw to shape your final output:
--stylize (or --s) controls how much Midjourney applies its trained aesthetic preferences. Higher stylization can sometimes override reference adherence, even with elevated --iw values. If your generations drift too far from your reference, consider lowering --s.
--quality (or --q) affects rendering time and detail level. While it doesn’t directly change image weight behavior, higher quality settings can reveal subtle differences in how the model interprets your reference.
Raw mode reduces Midjourney’s automatic aesthetic enhancements, giving you more literal interpretations. Pairing Raw mode with carefully tuned --iw values creates predictable, reference-faithful results for technical or commercial work.
| Parameter | Range | Default | Effect on Image Weight |
|---|---|---|---|
--stylize | 0-1000 | 100 | High values may reduce reference adherence |
--quality | 0.25, 0.5, 1 | 1 | Affects detail rendering, not weight behavior |
| Raw mode | on/off | off | Increases literal interpretation of references |
Step-by-Step: Adjusting Image Weight (With Copy-Paste Prompts)
This practical workflow covers the most common scenarios: single images, multiple references, and advanced combinations. Each example includes copy-ready prompts you can adapt immediately.
One Image + Text (Baseline)
Start with the simplest case: one reference image plus a text description. If your upload fails, see our Midjourney image upload fix guide. This workflow helps you understand the --iw spectrum before adding complexity.
Step 1: Upload your reference image to Discord or use a direct image URL.
Step 2: Begin with --iw 1 (the default) to establish a baseline, then test 0.5, 2, and 3 to see the full range of effects.
Here’s a practical example:
/imagine https://your-image-url.jpg cinematic portrait, rim light, teal and orange color grading --iw 2
This prompt creates stronger adherence to your reference’s pose and color palette while incorporating the cinematic lighting and color grading from your text description. The --iw 2 value doubles the default influence, making the reference more dominant.
For comparison, the same prompt at --iw 0.5 would prioritize the text description, using the reference only as loose inspiration. At --iw 3, you’d get maximum fidelity to the reference image’s composition and visual elements.
Multiple Reference Images (Blend + Steer)
When working with multiple reference images, you need to control not just the overall image-to-text balance but also the relative importance of each image. This is where prompt weights and --iw work together.
Chain your image URLs first, then add your text description. Use prompt weights (::) after each image URL to emphasize which references matter most:
/imagine https://pose-reference.jpg ::2 https://texture-reference.jpg ::0.8 vintage leather jacket, shallow depth of field --iw 1.5
In this example, the pose reference drives the composition with a weight of 2, while the texture reference provides supporting detail at 0.8. The --iw 1.5 parameter then balances all the images collectively against the text prompt.
This technique excels for character consistency projects, where you might combine a face reference, a pose reference, and a clothing reference with different emphasis levels.
Style Reference + Image Weight (Advanced Control)
The most sophisticated approach combines a content reference image with a style reference using --sref. This separation gives you independent control over « what » and « how. »
/imagine https://character-photo.jpg thoughtful close-up --sref https://painterly-style.jpg --sw 250 --iw 1.2
Here’s what’s happening: The character photo (with --iw 1.2) defines the composition and subject, while the style reference (with --sw 250) applies painterly aesthetics. The elevated --sw value makes the style prominent without destroying the character’s structural integrity.
This workflow is powerful for maintaining brand consistency while exploring different artistic treatments, or for creating character variations across different visual styles.
Troubleshooting & Best Practices
Even with correct syntax, image weight can produce unexpected results. Here’s how to diagnose common issues and fix them quickly.
« My Image Barely Influences the Output »
If your reference image seems ignored or barely visible in the results, try these solutions in order:
First, increase your --iw value to 2 or 3. The default of 1 may be too subtle for complex references or when competing with detailed text prompts.
Second, simplify your text description. Busy, detailed prompts can overwhelm image influence. Let the reference do more of the visual heavy lifting by reducing text complexity.
Third, lower your --stylize value or enable Raw mode. Midjourney’s aesthetic preferences sometimes conflict with reference adherence, especially at higher stylization settings.
« Too Close to the Reference »
When generations look like near-copies of your reference image without enough creative interpretation:
Reduce --iw to 0.5 or 1. This shifts the balance back toward your text prompt, giving Midjourney more creative freedom while still respecting the reference.
Add negative prompt weights to suppress unwanted elements: rustic kitchen::-0.6 tells Midjourney to downplay certain aspects from the reference, creating selective emphasis.
Consider whether you actually need a reference image at all. Sometimes a well-crafted text prompt alone produces better results than fighting against an overly influential reference.
« Multiple Images Fight Each Other »
When using multiple references that seem to cancel each other out or create chaotic results:
Apply prompt weights directly to the image URLs before your text prompt. Structure it like https://primary.jpg ::2 https://secondary.jpg ::0.8 to establish clear hierarchy.
Keep your --iw moderate (around 1 to 1.5) when using multiple images. Extreme values amplify conflicts between competing references.
Test images individually first to understand their characteristics, then combine them with appropriate weights. This systematic approach prevents wasted generations.
Live Examples (Swipe & Try)
These concrete, copy-ready prompts demonstrate low, medium, and high image weight values, plus a multi-image scenario. Replace the image URLs with your own references.
Low Image Influence (Interpretive)
/imagine https://city-street-photo.jpg neon noir street at night, rain reflections --iw 0.5
Expectation: A text-led composition with subtle borrowing from the photo. Midjourney has maximum creative freedom while using the reference as loose inspiration. Perfect for when you want the mood of a reference without copying its composition.
Balanced (Safe Starting Point)
/imagine https://city-street-photo.jpg neon noir street at night, rain reflections --iw 1
Expectation: Even balance between image and text. This default setting works for most scenarios and provides a reliable starting point before fine-tuning. The reference clearly influences results without dominating them.
High Adherence (Match the Reference)
/imagine https://city-street-photo.jpg neon noir street at night, rain reflections --iw 2.5
Expectation: Much closer framing and lighting from the reference. Composition, perspective, and major visual elements strongly mirror the source image. Use this when you need consistent results across variations or want minimal creative interpretation.
Multi-Image Blend + Emphasis
/imagine https://pose-reference.jpg ::2 https://wardrobe-reference.jpg ::0.7 editorial portrait, soft window light --iw 1.3
Expectation: The pose dominates the composition while wardrobe elements provide supporting detail. The combined image influence (at --iw 1.3) remains strong but allows room for the editorial portrait style and lighting description.
FAQs
What’s the valid range for --iw today?
The current valid range is 0 to 3 for V7, V6, and Niji 6, with a default value of 1. Values outside this range will either be clamped or produce errors.
Does --iw replace prompt weights?
No. These are complementary systems: --iw balances images versus text globally, while prompt weights balance parts of your prompt against each other (including multiple images). Use both together for maximum precision.
How is style weight different?
--sw (0 to 1000, default 100) controls how strongly a style reference (--sref) affects your output. It’s completely separate from image weight and controls aesthetic treatment rather than compositional influence.
Can I use negative --iw values?
No. The valid range starts at 0 (minimal image influence). For selective negative emphasis, use negative prompt weights within your text instead (like unwanted element::-0.5).
Do all Midjourney versions support --iw?
V7, V6, and Niji 6 all support the 0 to 3 range. Older versions (V4, V5) had different systems and ranges. Always check the official documentation for your specific version.
Key Takeaway: Image weight (--iw) is your primary tool for balancing reference images against text prompts in Midjourney. Start with the default value of 1, test the full 0 to 3 range to understand its effects, and combine it with prompt weights and style references for sophisticated control. Keep your workflow systematic—test individual elements before combining them, and always refer to official documentation for the latest parameter ranges and behaviors.