Stylization and quality: Using --stylize, --raw, and --quality

Midjourney offers a range of parameters to help you fine-tune how your images are generated. Among the most impactful—but often misunderstood—are Stylize, Raw Mode, and Quality. Each of these parameters changes how your prompt is interpreted, how the model behaves, or how much processing is applied. Used together, they allow you to strike the right balance between artistic expression, realism, and technical control.

In this section, we’ll break down what these parameters actually do, how they behave in Web and Discord UIs, and how they can be applied in combination for best results.

stylize: Guiding the level of artistic freedom

The --stylize parameter (also written as --s) tells Midjourney how much creative freedom it has to transform your prompt into visual form. Think of it as a dial between “literal interpretation” and “artistic imagination.”

When stylize is set to a low number like --s 0, the AI aims to follow your prompt very closely. The images tend to be more realistic, straightforward, and precise. But as you increase the value—up to a maximum of --s 1000—Midjourney begins to experiment more boldly, adding flair, drama, and stylization that might go beyond what your prompt explicitly described.

You can adjust the Stylize value using the slider in the Web UI under the Aesthetics panel. Or, if you're using Discord, you can also type --s followed by a value directly into the prompt bar.

Below is a visual comparison showing how different stylize values influence the final output using the same prompt “anime style dog”:

Stylize value: 0

With --s 0, the output stays very close to the literal meaning of the prompt. The image above has clean, sharp lines and a flat, poster-style aesthetic. The AI focuses on clarity and structure, avoiding any added flair or abstraction.

Stylize value: 1000

With --s 1000, the model adds significant artistic interpretation. The result becomes more expressive and cinematic—the lighting is dynamic, the colors are more saturated, and the overall tone feels narrative-driven. It evokes emotion rather than precision.

This contrast demonstrates how Stylize influences Midjourney’s creative freedom: the higher the value, the more the AI “paints outside the lines” to deliver visually rich, mood-driven scenes.

raw mode: Reducing the model’s aesthetic bias

The --style raw parameter tells Midjourney to suppress its built-in artistic bias and focus purely on your prompt. Think of it as flipping a switch from “Midjourney’s cinematic flair” to “plain and direct interpretation.”

In Standard mode, Midjourney tends to add its signature style—moody lighting, dramatic depth, and painterly finishes—regardless of how straightforward your prompt is. Raw Mode reduces that layer of stylization, producing images that feel cleaner, flatter, and closer to what you described.

You can activate Raw Mode in the Web UI by selecting the Raw button under the Model section. Or, if you're using Discord, you can type --raw directly into the prompt bar.

Below is an example showing the difference Raw Mode makes when interpreting the same prompt “anime style dog”:

With Raw Mode turned on, the output becomes more neutral and restrained. The image above is a simple, flat-shaded character portrait with limited texture, controlled lighting, and clean edges. There are no dramatic flourishes or atmospheric effects. It’s ideal when you want visual clarity, focus on form, or base art for further editing.

This setting gives you full control over the visual tone—making sure the AI doesn’t add its own artistic filter unless you want it to.

Raw Mode reduces Midjourney’s default aesthetic styling, but it doesn’t limit your ability to create a visually rich image. If your prompt includes specific artistic elements—like style, lighting, setting, or mood—Raw Mode will still honor those instructions.

For example,

Even with Raw Mode enabled, this prompt generates a highly stylized and atmospheric image—because the stylistic intent is built directly into the words.

quality: Adjusting detail and processing time

The --quality parameter (also written as --q) tells Midjourney how much rendering power to devote to generating the initial image grid. Think of it as a setting that controls how much time and GPU effort is spent polishing the image—where lower values prioritize speed, and higher values prioritize detail.

This setting doesn’t affect what the image contains—it doesn’t change composition or stylization. Instead, it affects how refined and detailed the image appears when it’s first generated.

When you lower the value to --q 0.25, Midjourney produces results more quickly, but with simplified textures and structure. At the higher end, like --q 2 or --q 4, the model spends significantly more time rendering intricate textures, lighting effects, and surface detail.

You can use the Quality parameter by typing --q followed by a value directly into the prompt bar. This is the only way to apply it in the Web UI, as there’s currently no slider or toggle for --q.

Below is a visual comparison showing how different quality values influence the final output using the same prompt: “anime style dog”

Quality value: 0.5

With --q 0.5, the image renders quickly but lacks fine detail. Lines may appear softer, background elements are simplified, and subtle textures (like fur or fabric) are either muted or missing. It’s ideal for quick drafts or when you're generating many options rapidly.

Quality value: 2

With --q 2, the image is rendered with noticeably more richness. The fur texture looks layered, lighting is more nuanced, and even background foliage or clothing reflects subtle shading and color depth. While it takes longer to render and uses more credits, it’s useful when you want final-quality outputs.

This comparison highlights how the Quality parameter influences visual fidelity: lower values deliver speed and simplicity, while higher values invest in clarity, precision, and finish—without changing the core composition.

Practical case study

The best way to grasp how parameters like --stylize or --raw influence Midjourney outputs is by reviewing side-by-side image comparisons. Websites like AI Compare Hub offer a helpful environment for this, allowing users to observe how subtle changes in values can significantly affect the outcome.

Below are several real examples based on the same prompt, where only the parameter settings were adjusted:

1. stylize 0

Twilight Flight Pair – Trial 1: Chaos 5 Stylize 0 (Midjourney V7)

With --s 0, the image remains closely tied to the prompt. Even with randomness introduced through --chaos 5, the result stays clean, structured, and easy to interpret.

2. stylize 500

Twilight Flight Pair – Trial 2: Chaos 5 Stylize 500 (Midjourney V7)

Increasing --s to 500 adds a stronger stylistic layer. The image starts to adopt more atmosphere, color blending, and a sense of visual storytelling.

3. stylize 1000

Twilight Flight Pair – Trial 3: Chaos 5 Stylize 1000 (Midjourney V7)

At the maximum stylize setting, the result becomes highly interpretive. Visual elements may drift further from the original prompt, and the image can take on an abstract or dreamlike quality.

4. stylize 1000 + raw

Twilight Flight Pair – Trial 4: Chaos 5 Stylize 1000 + Style Raw (Midjourney V7)

Adding --raw to a high stylize setting helps restore structure. The image still carries expressive qualities, but elements like form, layout, and clarity are better preserved.

---

Stylize, Raw Mode, and Quality are foundational controls for any Midjourney user seeking consistent and personalized output. By learning how they interact and experimenting with different values, you’ll be better equipped to turn your creative intent into precise results.