Imagine Weekly – Issue 4 [June 27th]

Welcome to Issue 4 of Imagine Weekly 👋

Hello everyone,

hope you’re doing great! Yesterday's Midjourney Office Hours provided some great insights. I'm especially excited about the progress in 3D and video development, with 3D moving along faster. I can't wait to try it out.

In this issue you’ll find:

• 📖 TL;DR; Weekly Recap of the Office Hours (or the extended recap)

• 🎓 A tutorial on how to use the Style Reference Feature in the web UI Jump to the article or read below

• 🎨 Some new srefs to tinker with

Office Hours TL;DR; Recap 2024-06-26

If you’re interested in a more extended recap of the latest Midjourney office hours, check out 🍿My Summary (external link).

🛠️ Model Updates

• Midjourney 6.5: Training stage 2 of 3-4, final run pending issues

• Midjourney 7: Delayed, aims for better knowledge, prompts, speed

📊 Demographic Surveys and Aesthetics

• Analyzing user surveys for demographic aesthetics

• Observations:

  • 75% male users, male-dominated aesthetic

  • Differences in male vs. female, LGBTQ+, regional aesthetics

  • Future: Balance default aesthetics, push for personalization

👤 Personalization

• Encourage more user personalization

• New onboarding with fast hour reward

• Better, faster personalization tests

• Considering reset option, working on profiles

🌐 Website Updates

• Modify room navigation, lightbox button

• Improve help, onboarding tutorials

• Test limited access without Discord

• Explore personalized recommendations for explore feed

🎨 Style Exploration

• Considering random style explorer

• Potential user-driven "hot" and "top" style pages

🎥 3D and Video Development

• Both in progress, 3D advancing faster

• Focus on quality, recent 3D progress promising

📝 Miscellaneous

• Niji team may add more styles

• Exploring better style discovery tools

Ongoing undisclosed projects

📚 Tutorials & How-To’s

How to Use the Style Reference Feature in the Web UI

The Style Reference is an excellent tool in Midjourney that accomplishes several tasks:

• 🎨 Transfer Specific Style onto an Image: Style Reference clones the aesthetic qualities of one image onto another, useful for artists wanting a consistent look or to replicate classic styles.

• 🖼️ Generate Multiple Images with the Same Aesthetics: Style Reference ensures stylistic consistency across a series of images, beneficial for cohesive visual identities in art collections or branding.

• 🔍 Explore New Looks (with sref codes): Users can experiment with new visual styles using specific sref codes, enhancing creativity and broadening artistic horizons.

• ✨ Enhancing Creativity: Style Reference allows for the mixing of various styles, fostering unique artistic combinations and new directions.

• 📌 Consistency in Projects: Ensures all project outputs maintain a uniform style, crucial for effective branding and thematic series.

• 🧪 Experimentation: Facilitates testing how different styles affect various subjects, aiding artists in developing their unique style.

• ⏱️ Efficiency in Workflow: Streamlines artistic workflow by allowing quick application of preferred styles, saving time and effort.

• 📚 Learning Tool: Offers insights into visual design elements like color theory and composition through style analysis, serving as an educational resource.

Alright, let's go over this.

If you're unfamiliar with how to use it in Discord, I've written a post about it which you can find here.

Step Step guide

Let's explore how to utilize the Style Reference Feature in Midjourney's Web UI.

First you have to provide an image to your prompt:

• To upload an image, you can either paste the URL of the image directly into the prompt input or click on the image icon.

• You can also choose an image from your Midjourney image library by selecting it from the available options.

It doesn't matter which method you choose, as they all achieve the same result. After the image appears below your prompt input, hover over it to see three icons. For using the Style Reference, click on the paperclip icon. If it's already active, click it again to deselect. When you stop hovering over the image, the paperclip icon should remain visible.

Next, proceed to enter your text prompt as usual. If you wish to adjust the Style Weight, you can use the --sw parameter in your text prompt to fine-tune the settings.

If you want to use style reference codes, simply append them to your text prompt in the same way you would on Discord.

What about multiple srefs and individual weights?

Currently, the Midjourney web UI does not support using multiple style weights simultaneously, except when using sref seeds. When using sref codes, you can append :: followed by a weight to each code. For example:

--sref sref_code1::weight1 sref_code2::weight2

However, when using image URLs or images, the style reference feature in Midjourney only allows for a single style reference at a time, with one style weight parameter. To achieve a similar effect, you can try a few workarounds:

1. Blend Multiple Styles Manually: Create a composite image that combines elements from the different styles you want to reference. Use this composite as your style reference image.

2. Iterative Approach: Apply one style reference, generate an image, then use that result as a new base image and apply another style reference. This method allows you to layer different styles sequentially.

3. Prompt Engineering: Incorporate descriptive text in your prompt that reflects the additional styles you want to include. This can help guide the AI to blend multiple styles, even if you're only using one style reference image.

4. Adjust Style Weight: Use the --sw (style weight) parameter to control how strongly the style reference influences the output. Experiment with different weights to find a balance between the reference style and your prompt description.

While these methods can help approximate the use of multiple style weights, they do not provide the precise control of applying different weights to multiple style references simultaneously. The Midjourney team may introduce more advanced features in future updates, but for now, these workarounds can help you achieve more complex style combinations in your generated images.

As a last resort there’s still Discord.

🙌 Alright, that's it. I hope this helps you transition further away from Discord and make full use of the Web UI. Stay tuned for upcoming tutorials about the new Web UI!

🌈 Related

🎨 Inspiration, Prompts and Srefs

For today's issue, I've gathered some nice references for you to save.

See you next week 👋🤖

I hope you enjoyed this issue of Imagine Weekly. If you found it helpful, we'd be thrilled if you could share it with your colleagues and friends. Please feel free to reach out with any suggestions or feedback.

To support me:

😘 Recommend the newsletter to your friends: it really helps!

📬 If you’re not subscribed yet to my newsletter Imagine Weekly, I’d be thrilled to welcome you on board!