7 Midjourney v5 Prompts for Marketers and Creators

New algorithm, new tricks!

New algorithm, new tricks!

big “AI Ethics: Humans Strike Back” episode (#4) of No Brainer: An AI Podcast for Marketers. Greg Verdino and I discuss Midjourney and ChatGPT entanglements and other AI ethics issues. You can watch or listen to it here: https://nobrainerpodcast.com/ai-ethics/

Midjourney 5 arrived a few weeks ago with a dynamic increase in capability, delivering photorealistic images good enough to use in marketing materials (and provoke international deep fake controversies). In addition to the much-improved image quality, the new algorithm offers a few new tricks for marketers and creators looking to build content.

Rather than rehashing the same 10 tips from my last Midjourney article (circa version 4), these are new prompt codes with the exception of one, which has new functionality.

It is helpful to set up your Midjourney configuration in advance to version 5. Access them by entering the prompt “/settings” and choosing MJ version 5. My settings include Midjourney version 5 and I produce images publicly. You can access my profile via the handle “livingwf8”.

And with that, let’s get to it.

Tip 1: Narrative Prompting

In prior versions of Midjourney, you needed to add descriptors separated by commas and prompt codes. In version 5, you can add a narrative description and the engine will accept it as a prompt code.

This is much more useful to normal human beings who have not taken on the challenge of becoming prompt engineers. Narrative prompting also allows you to use text-generative AIs to create an image prompt. Let’s go through a couple of examples.

First, let’s use my narrated prompt. It took me four turns to create this image. Why so many? I kept trying to put elephants in the background, and Midjourney 5 could not successfully add the next element. This is no different than version 4s hiccups. I find this is common with narrative prompts. So I simplified.

Prompt: a male lion walking on the Serengeti during the golden hour, surrounded by long blades of grass. Place the lion on the right third of the photo. The image is superphotorealistic shot with a Canon R5 in 8k. — ar 3:2

And of course, I used ChatGPT 4 to generate my next image prompt. Here’s what ChatGPT’s prompt produced.

Prompt: As the golden hour casts a warm, ethereal glow over the vast expanse of the Serengeti, a majestic lion emerges from the shadows. His tawny coat, kissed by the sun’s fading light, shimmers with an almost otherworldly radiance. With each graceful stride, his powerful muscles ripple beneath his fur, embodying the essence of regal strength. The lion’s amber eyes, intense and keen, scan the horizon as he navigates the savannah with quiet confidence. The golden grasses sway gently in the breeze, parting before him like a sea of whispers, as if paying homage to the king of the wild. In this moment, the lion’s presence is a testament to the untamed beauty and raw power of nature, forever etched in the memory of the Serengeti at the golden hour. — ar 3:2 — v 5 — q 2 — s 250

As usual for ChatGPT, the prompt sounds academic and windy, but you can’t argue with the end result here. Very nice!

Tip 2: Image Text for Search

Prompt: A photorealistic ultra-detailed 8k long exposure with light trails at night in Tokyo, lots of motion from car traffic, show speed, lots of bright light, night view with neon and glowing lights, landscape photography, with highly detailed buildings and shops and cars, — ar 3:2 — v 5 — s 250 — q 2

When building an image for your article or content and you should associate keywords with their images to help with search and accessibility. In fact, some platforms like WordPress and LinkedIn will prompt. Enter Midjourney’s new image recognition description feature.

Simply prompt the engine with the phrase “/describe.” A window will pop up prompting you to upload the image to describe. Four suggested descriptions are offered after analysis.

Here are the suggestions Midjourney offered for the above image, also generated on the platform.

I like the first one as it is closest to the original prompt. However, the others are interesting. If you are writing an article with a series of similar images you could use them to develop similar images. Or you could simply use them to possibly produce a better image.

Here is one example that came out of the variants.

Prompt: kyoto yokohama city by alex kaodong, in the style of motion blur, neon lights, nikon d850, outrun, uhd image, digital art techniques, colorized — ar 3:2 — v 5 — q 2 — s 250

Tip 3: Unlimited Aspect Ratios

The new and improved Midjourney 5 offers unlimited aspect ratios. As noted in my original Midjourney prompt article, the — ar x:x prompt is absolutely necessary for anyone using these images to produce commercial images. Otherwise, you get stuck with the standard Midjourney squares.

However, aspect ratios were capped at 2:1 or 1:2. Now the ratios are unlimited. Creators can generate more panoramic wide images, which are very useful for header images on social networks and more.

Prompt: A photorealistic oceanscape with white-capped waves and a colorful sunset shot with a Hasselblad in 8k — ar 16:3 — v 5 — q 2 — s 250

The above ocean sunset at 16:3 was the perfect scale for my new LinkedIn header image. You can see it here.

Tip 4: Cinematic Portraits for Content

How good is Midjourney 5? This good.

Prompt: Cinematic close-up shot, one woman, one man having a business conversation, office scene, shallow depth of field, bright LED light — ar 16:9 — seed 3000 — v 5 — q 2 — s 250

You can create cinematic images for your marketing content that is at least competitive with more costly stock imagery. Be mindful to order the prompts, asking for “Cinematic” and then the type of shot, then subjects, followed by scene descriptors. If you prioritize the scene first, you’ll get a lesser product.

Here is another example, this time with a wide-angle scene.

Prompt: cinematic shot, two women having a conversation at an outdoor lunch table, office park scene on a sunny day — ar 16:9 — seed 2000

I would have to zoom in on these images to identify that they are not photographs. Outstanding! From over the shoulder to a cinematic scene, content creators have more to gain from Midjourney 5 than ever before. If you want to learn more about cinematic prompting, check out this primer.

Tip 5: Batch Jobs

One new feature allows you to process batch image concepts simultaneously. With the Permutation and Repeat feature, you enter a prompt with multiple variables.

This is ideal if you have a concept but would like to see it executed with different elements. One barrier: You must have a standard or pro account to use this feature. Here is how it works:

Create your prompt with your varying elements in curly brackets. Here is an example: “Prompt: An ethical quandary of AI versus human, with a cyborg in the middle shown in a {cityscape, office, field, celestial} scene — ar 3:2”.

Let’s see how it worked out…

Cityscape scene.

Office scene.

Field version.

Celestial version.

Pretty cool, right? The batch approach is a great feature for those creative moments where you have a general idea of what you want, but are unsure about the actual execution. You know what you like when you see it. Sometimes you need to see several versions before you land on the ideal.

Tip 6: Mashup Images

The ability to blend images was technically available in version 4. But I was not impressed with the older version‘s execution of the prompt and did not share the codes. Midjourney 5 blends existing photos and images in a much better fashion.

To blend images, simply get 2–5 images, and type in the command “/blend”. This lets you upload the images you want to mash up and choose an aspect ratio. Unfortunately, mashup aspect ratios are pretty limited to portrait, square, and landscape, but it's still a cool feature.

Check out this blend I created using a photo I took of my friend and model Holly Ahrens and two light trails created by street cars in Houston.

Here is the mashup. It’s pretty cool, but perhaps a little too animated for some. Still, it is usable for basic social media posts and blog articles.

Future versions will make it even better, but the feature still lags behind Photoshop’s image-blending capabilities. I suspect professionals needing to blend images will look to Adobe Illustrator or Photoshop for more control and the aesthetic license to combine photos and graphics.

Tip 7: Image Weighting

Another key feature with Midjourney 5 is the ability to use an image as a prompt, but now with a custom prompt number. So instead of 1, 2, 3, etc. as weights, you could use pi as a weight, though I am unsure why you want to get into long decimals ;).

I decided to illustrate this using the above Holly photo that served as one of the images prompted, but this time placing an interpreted version of my friend in a blossoming orchard of cherry trees. Here is one with the model photo weighted at 0.5.

Prompt: https://s.mj.run/pFpalwdit7c a woman walking in an orchard of fully blossoming cherry trees, there are blossoms everywhere, with falling blossoms looking like snow, and a carpet of blossom petals on the grass, photorealistic, ultra hd, 8k, detailed — iw 0.5 — ar 3:2 — v 5 — q 2 — s 250

And with the image weight at 1.

Prompt: https://s.mj.run/pFpalwdit7c a woman walking in an orchard of fully blossoming cherry trees, there are blossoms everywhere, with falling blossoms looking like snow, and a carpet of blossom petals on the grass, photorealistic, ultra hd, 8k, detailed — iw 1 — ar 3:2 — v 5 — q 2 — s 250 -

And finally, with the image weighted at 2.

Prompt: https://s.mj.run/pFpalwdit7c a woman walking in an orchard of fully blossoming cherry trees, there are blossoms everywhere, with falling blossoms looking like snow, and a carpet of blossom petals on the grass, photorealistic, ultra hd, 8k, detailed — iw 2 — ar 3:2 — v 5 — q 2 — s 250

Conclusion

These prompts are cool, but the obvious and real progress made by Midjourney v5 is its incredible photorealistic imagery. Several of the images above illustrate the platform’s incredible progress.

Still, the algorithm has progressed in other, more subtle ways, making it the best AI image generator available to marketers and creators hands down. The real question is: can Midjourney build on its technological lead, or will competitors like Stable Diffusion, Dall-E, and Firefly catch up?

Stay up to date with the latest news and updates in the creative AI space — follow the Generative AI publication.