Better Midjourney V7 Images: Simple Tips for Impressive Art

Midjourney V7 is out in alpha, bringing new ways to create stunning images. It's still improving, but you can start using its features right now. These tips work great for V7, and many are helpful for V6 too. Let's look at how to get the most from Midjourney's latest.

Crafting Effective Prompts for V7

Put Important Subject First

Words matter in your prompt, especially the order. Midjourney V7 handles longer prompts well, but the item you put first usually gets the most focus. If you want a knight to be the main subject, put 'knight' at the beginning of your prompt.

Use Clear, Natural Language

Midjourney loves clear sentences. Good grammar and spelling help it understand exactly what you want. Think of your prompt like describing a scene to someone. Say what you see and what you want.

Here’s a simple structure that works well:

• Start with the main subject and its key action or placement.

• Add details about the background or surrounding elements, being specific about location.

• Finish with style descriptions or overall atmosphere details.

For example, instead of just mentioning a chandelier, say where it is, like "a grand chandelier hanging from the ceiling in the background." Structuring your prompt helps Midjourney see each idea clearly. While word lists can be fun for exploring, clear sentences work best when you need specific results.

Be Specific with Details

Don't just ask for a "cartoon style." Specify the kind of cartoon, like "1970s animated cartoon style." Different cartoon styles look very different in Midjourney. Pinpoint the specific style you want. You can name artists or shows, but also think about the style's features: simple shapes, bold lines, flat colors, exaggerated faces, etc.

If Midjourney seems to miss a detail, try describing it in different ways. For instance, to get a robot parrot, you might say "a mechanical parrot" and also "with mechanical parts." Repeating important ideas helps Midjourney focus on them.

Exploring V7's Improved Understanding

V7 is better at understanding many things. It's improved with styles, artists, facial looks, and some terms for how a picture is set up.

See Better Subject Recognition

V7 understands specific requests like "CCTV footage" better. It usually shows the view from a camera, not just pictures of cameras themselves. It also nails specific art styles more accurately, like Diesel Punk.

Easier Full Body Shots

Simply adding "full body" to your prompt often works now, even with wide images. If that doesn't work, add details about clothing, shoes, or saying the person is "standing on the ground." For pictures from the waist up, use "2/3 body" and add details like hair color or what they are wearing.

Improved Facial Expressions

Getting faces that are winking or have specific looks is easier now. V7 understands the expression without zooming in too much on just that part of the face.

Setting Up Scenes Clearly

Getting specific camera views, like an over-the-shoulder shot, can be tricky. Instead of just saying "over the shoulder," describe the parts of the scene. Say who is in the foreground (with their back showing) and who is in the background (with their face showing). Specify their positions (left/right) and what they look like. This method works for different image sizes and gives you more control.

You can also create layered images, like double exposures, by describing what is on the "bottom layer" and what is on the "top layer."

Want to maximize your creative output without manual work? The TitanXT Midjourney Automation Suite helps you automate prompting and image generation, freeing up your time.

Using Style Raw

Midjourney balances making nice pictures with following your prompt exactly. Sometimes its default look gets in the way of your specific request. Adding [--style raw] tells Midjourney to reduce its own art style and stick closer to what you wrote. If a prompt isn't working, try it with [--style raw]. Keep in mind that high stylized values, personalization codes, or style references can sometimes lessen how closely Midjourney follows your prompt, so you might need to adjust those too.

Get Culturally Accurate Images

V7 is much better at understanding other languages. You can use this to make images that feel more authentic to different cultures. Translate your prompt into the language associated with the culture you want to show. Midjourney might tailor the look based on that language and culture.

Understand Your Personalization Code

If you use a personalization code, see how it affects your images with a permutation prompt. Use curly brackets { } and commas to run multiple prompts at once. For example, you can test the same prompt with and without your personalization code to see the difference. Permutation prompts need Fast or Turbo mode enabled.

Tips for Style Reference (SREF) and Remix

Trying SREF in V7

The style reference feature is still being worked on in V7, so results can vary. Here are a few things that help:

• Increase the [--stylize] value. This adds more of Midjourney's general art style, which can sometimes look better with SREF results, though it might make the prompt followed less closely.

• Include your V7 personalization code. If you do this, the [--stylize] value controls how strong your personalization is, not the default Midjourney style.

• Blend multiple SREF codes. Using two or more SREF codes that look similar or go well together can produce interesting results. You can add weights to make one style stronger than others.

Remix Your V6 Images in V7

Remix mode lets you change the prompt and parameters while creating variations of an existing image. You can use this to take your favorite V6 images and see new versions made with the V7 model. Choose an image, click Remix Subtle, change the model to V7, and update the prompt or add parameters like your V7 personalization code.

Automating tasks like using different styles or testing variations can save significant time. Check out the TitanXT Midjourney Automation Suite to see how it can streamline your workflow.

Making the Most of Draft Mode

Draft mode is a faster, lower-quality mode great for trying out lots of ideas quickly. Click the lightning bolt button to enter it.

Conversational Mode Commands

In draft mode, you can use conversational mode. This means you give Midjourney commands, and it creates the prompt based on what you say. You can even set a word or phrase that tells Midjourney when to run a prompt. For example, you could set it so Midjourney only creates images when you say "run that prompt."

Run Multiple Ideas at Once

In conversational mode, you can command Midjourney to run several prompts based on one idea. Ask it to show "a bear greeting a fox in five different cartoon styles," and it will create five variations for you, choosing the styles itself.

Create and Use Named Characters (Experimentally)

Conversational mode remembers past jobs in a session. You can create a character by naming them and giving details (e.g., "a female warrior named Sai with brown hair..."). Then create another character ("a white-haired magician named Robert"). Afterward, you can give a command using their names ("place Sai and Robert in a forest scene"). Midjourney will often pull their descriptions from the previous jobs to create a picture with them together. This is a fun way to quickly test character ideas, though it's not character reference.

The Enhanced Button

Once you get an image you like from draft mode, click the Enhance button under it. This will give you a new grid of higher-quality images very similar to the one you chose. The variation is even smaller than using 'Very Subtle.'

These tips can help you push your Midjourney creations further in V7. Experiment with prompt structures, explore the new model's understanding, and use features like Draft Mode and Remix to refine your ideas.

Remember, tools like the TitanXT Midjourney Automation Suite exist to help you scale your creative process and make complex tasks easier.

Final Thoughts

Midjourney V7 is a powerful tool that keeps getting better. By focusing on clear prompting, understanding the new features, and using modes like Draft and Remix, you can get fantastic results. Don't be afraid to try things out! If you have a prompt that's challenging, try rephrasing it or adjusting parameters.