Purpose of Part 2: Style Tuner

The aim of Part 2: Style Tuner is to elucidate the process of creating a Style Tuner and the relevant factors when utilizing the /tune command.

Quick Summary

• Style Tuner is an essential tool for discovering and developing personalized styles for your creations. It produces Style Codes which can be used to infuse an artistic style into your images.
• The /tune command generates a Tuner page. When crafting a Tuner, consider these four aspects: Have you tested the prompt? Is creating a Tuner necessary? What are the costs and number of style directions? Which mode: raw or default?
• Regardless of the prompts, there is only one Style Tuner tool. The visual directions maintain consistency and can be described in clear terms.
• Conversely, the interaction between a Style Code and the prompt is a complex “alchemical reaction” that is unique to each prompt.
• The Style Tuner, with its sample images showcasing the outcomes of these creative interactions, aids users in selecting and refining their customized styles.

What is the official term for a Style Tuner page?

The correct terminology is “Tuner” or “Style Tuner,” referring specifically to the Tuner page or URL. Unfortunately, this term has not been consistently applied in the Prompt FAQs. The Style Tuner is also described as a tool for style exploration. When specifying the type of Tuner, for example, 128 Style Directions, it is referred to as a 128-row Tuner (128-Tuner) with 128 Visual Directions.

Differentiating Style Directions from Visual Directions

In most instances, these terms can be used interchangeably. “Style Direction” is used in the /tune menu, while “Visual Direction” refers to the rows or grid pairs displayed on the Tuner page. The Style Tuner allows for a maximum of 128 visual directions corresponding to the 128 style dimensions within the aesthetic space. It remains unclear why the developers chose to employ two different terms for the same concept.

Anatomy of a Tuner

The image above illustrates a Tuner featuring 128 Style Directions, providing users with a selection of images per row. The images on the Tuner page are labeled as “sample images” and serve solely as canvases to exhibit the applied styles of visual directions. Each row in the “Pair” view presents two visual directions. Users can opt for either the Left (L) or Right (R) style to blend into the final aesthetic or select the neutral middle position (“neither”).

The first visual direction typically showcases the most popular styles among users. The default Midjourney house style often aligns closely with this visual direction.

Users have the option to configure Visual Directions 2, 3, and others by selecting left or right images or choosing to skip the selection.

In “Grid” view mode, each selected image enhances the aesthetic in that direction while omitting the alternative aesthetic direction. Opting for neither direction maintains a neutral stance.

Personally, I prefer the Grid view, especially for a 128-Tuner with over 100 images, as scrolling through numerous rows can be quite taxing.

At the bottom of the Tuner page, you will find the Style Code and the generated prompt.

What is the function of a Style Tuner?

1. To explore a variety of styles.
2. To craft a customized style.

Exploring Different Styles

The Style Tuner allows for the exploration of countless styles represented as Style Codes, eliminating the need for users to have a deep understanding of aesthetic terminology.

Creating a Customized Style

You can generate a personalized style by configuring the visual directions on the Tuner page. A Tuner with 128 Style Directions offers 128 dimensions in pairs (left and right), allowing users to select their preferred visual directions. The configuration you make will instantly influence the generated Style Code.

The Style Code produced is your tailored style, which can be applied to other prompts to create a cohesive collection of images that share the same aesthetic.

Distinction Between /tune and Style Tuner

The /tune command generates a link to a Style Tuner page (or Tuner). Utilizing this command consumes Fast Hours upon confirmation, irrespective of whether you're in /relax or /fast mode.

Depending on the bot used to create the Style Tuner, there are two variations of the /tune command. The default /tune command from the Midjourney bot will create a Tuner in either Default or Raw mode, while the Niji·Journey bot can produce a Style Tuner tailored to anime styles.

The developers have introduced a /relax mode that does not consume Fast Hours for the /tune command, making it a cost-effective option—though its free status might not last indefinitely.

The Style Tuner acts as a web page that allows you to (1) explore Midjourney’s vast array of styles and (2) customize a reusable Style Code. You can utilize another user’s Tuner without employing the /tune command, thus preserving your Fast Hours.

Considerations for Using the /tune Command

1. Have you tested your prompt?
2. Is a new Tuner page necessary?
3. The cost in relation to the number of Style Directions.
4. Raw Mode versus Default Mode.

Testing Your Prompt

If your prompt is lengthy or complex, consider running a /imagine job to evaluate it before generating a large volume of images in the Tuner. Ensure the prompt effectively captures all desired elements and is free from errors. Only proceed with creating a Tuner when satisfied with the prompt's output to avoid generating a multitude of unsatisfactory images.

Necessity of a New Tuner Page

Should someone else have already created a Tuner for that prompt, the bot will alert you to refrain from creating another. It’s unnecessary to duplicate a Tuner as the generated styles would be akin.

You can also generate Style Codes from Tuner pages created by others. Access the collection of Tuner pages developed by the Midjourney community [here](https://tuner.midjourney.com).

Consider the utility of the Tuner page. Are you primarily aiming to use the Style Codes for that specific prompt, or do you want to apply them across various subjects? If your goal is to experiment with different styles, “Part 3: Explore the styles” will detail how to explore options without creating a Style Tuner.

Cost vs. Number of Style Directions

The /tune command allows you to specify the number of Style Directions (or Visual Directions) you wish to explore, with four options available: 16, 32, 64, and 128. Each Style Direction generates a pair of image grids to refine the style on the Tuner page. For instance, choosing “16 Style Directions” results in 32 image grids (16 x 2 = 32).

Creating a Style Tuner page incurs costs: “16 Style Directions” requires 0.15 Fast Hours, while “128 Style Directions” demands 1.2 Fast Hours! The higher the number of Style Directions, the greater the cost.

Raw Mode vs. Default Mode

The styles produced by the Tuner can either adjust the (1) default Midjourney house style or (2) raw output. Selecting “Default Mode” implies utilizing the default Midjourney house style, with the generated styles modifying it. The resulting Style Code is likely to function as intended when applied to another prompt using the default house style.

Opting for “Raw Mode” means generating output without the default base style, with the styles altering the raw output. The Tuner page will automatically append the --style raw parameter to the generated prompt. For further insights on default versus raw output, refer to Part 1: Create a style.

Should I Create Multiple Tuners for a Single Prompt?

Short Answer: It's advisable to create one 128-Tuner for future reference.

Longer Answer: There is only one Style Tuner tool. This tool encompasses 16 to 128 visual directions that can be adjusted to generate the final Style Code. If you’ve already established a 128-Tuner, it includes all visual directions from the 16-, 32-, and 64-Tuners, making it unnecessary to recreate the 128-Tuner for the same prompt.

Most Extended and Advanced Answer: The visual directions remain constant regardless of the prompt. For example, if you create a 128-Tuner for two prompts (A and B) with identical selections, the Style Codes produced will yield the same results.

Example 1: Comparing Tuners of the Same Prompt

Consider the first to fourth visual directions for a prompt in both a 32-Tuner and a 128-Tuner.

Prompt: /tune prompt: a kitten and a puppy in a kitchen — ar 16:9

• [32-Tuner](https://tuner.midjourney.com/NLzPvR0)
• [128-Tuner](https://tuner.midjourney.com/JVjnTCc)

While the sample images differ, the styles of the visual directions are fundamentally the same. The 128-Tuner encompasses a broader array of visual directions, including those from the 16-Tuner.

Example 2: Comparing Tuners of Different Prompts

Examining the first to fourth visual directions for two distinct prompts using 128-Tuners reveals similar styles.

Despite differing prompts, the styles remain consistent across visual directions.

Example 3: Style Sliders for Midjourney

The visual directions can be represented through words and sliders. Think of the visual directions as “themes.” This analogy can be explored further using the online tool.

Compatibility of /tune with Other Features in Midjourney

Midjourney recommends using the /tune prompt command without additional parameters. However, it is compatible with prompts that incorporate:

• Style phrases or keywords (e.g., watercolor, photography, etc.)
• Aspect ratio parameter --ar
• Chaos parameter --c
• Tile parameter --tile
• Multiprompts

It’s important to note that /tune and Style Codes do not work with image prompts devoid of a text prompt. Weights and image references in the /tune prompt will only influence the sample images, not the generated styles.

Role of the Prompt in Creating a Style Tuner Page

The prompt itself is not part of the generated styles in the Tuner. It serves to create the canvases upon which the styles will be applied. These canvases are referred to as “sample images” on the Style Tuner page. The prompt aids in generating the sample images with applied styles, providing insight into the function of the visual directions.

For instance, if a Tuner is created with only a dot as a prompt, it produces 32 visual directions, even though the dot lacks any substantive information.

Prompt: /imagine prompt: .

While the original prompt may seem insignificant, it is essential for crafting reproducible styles when applying a Style Code to other prompts. The “alchemy” of the style pertains to the interaction of the style and the prompt, and the results of this interaction are unique to each prompt. Hence, utilizing a specific prompt to create a Tuner is advantageous compared to using a non-informative prompt like a dot.

What Happens When You Create a Tuner with 128 Style Directions?

Creating a 128-Tuner establishes forces acting on the prompt across 128 aesthetic dimensions. The grid view illustrates “left” (L) or “right” (R) visual directions along these dimensions. The sample images on the Tuner page represent these visual directions.

When selecting the left or right sample images, the prompt is influenced accordingly, resulting in the final style. The visual directions correspond to coordinates within a 128-dimensional aesthetic space. Choosing left, right, or neither indicates varying levels of engagement with that area of the aesthetic space.

How Many Sample Images Should Be Selected to Fine-Tune the Visual Direction?

The chosen visual directions are blended together. Opting for a larger number of sample images means combining several visual directions, making the resulting style less predictable. Selecting fewer sample images enhances the likelihood of achieving a coherent style that aligns with your selections. Midjourney typically advises choosing 5–10 sample images.

I often select fewer than 5 sample images. What’s your preference?