How to Generate Full-Body Images with Midjourney: 5 Expert Tips

Are you struggling to create full-body images using Midjourney’s generative AI?

I am. I think we all are, in one way or another.

Midjourney continues to pose certain challenges when trying to achieve full-body portraits, as it often leans toward half-body or portrait compositions. While I do love my close crop composition more often than not, certain scenes call for a full-body image. So I’ve experimented–a lot. Here are five effective strategies that I’ve found to significantly increase your chances of generating dynamic full-body images with Midjourney.

1. Add Specific Details About Shoes or Feet

One powerful technique to encourage Midjourney to produce full-body images is to include specific details about shoes or feet in your prompts. It seems so simple, but it wasn’t until several argumentative sessions with Midjourney that I figured this one out. By mentioning descriptors like “black boots,” “ballet slippers,” or “barefoot,” you can enhance the likelihood of obtaining full-body portraits. While the AI may occasionally deviate slightly from your request, this approach remains highly effective.

*photography of ballerina on stage wearing white ballet slippers on a dimly lit stage, ethereal, spooky, hassleblad --ar 3:2*

Consider that mentioning specific types of clothing, such as “evening gown” instead of “dress,” can also yield impressive results. Midjourney appears to favor full-body representations when prompted with a more precise description like “evening gown” because it implies a longer garment that extends below the knees.

*photography of beautiful black woman in gold shoes and evening gown at a gala event, festive atmosphere, soft lighting, hassleblad --ar 3:2*

2. Incorporate Dynamic Action Verbs

If you want to capture dynamic and action-packed scenes, incorporate action verbs like “walking,” “twirling,” “running” or “dancing” into your prompts. These verbs encourage Midjourney to create images that feature a subject engaged in motion, rather than a static portrait, increasing the likelihood of obtaining full-body portrayals. However, keep in mind when using “walking” it may sometimes lead to back-view perspectives, which I do rather like but isn’t applicable to many scenes.

*photography of dancer in red shoes and evening gown dancing in the street at sunset, vivid, energetic, hassleblad --ar 3:2*

*barefoot woman in long ethereal white dress running through a dark forest, misty, hazy lighting, autumn foliage, hassleblad --ar 3:2*

3. Specify the Environment

Another effective strategy is to specify the environment in which you want your subject to appear. Including details like “futuristic room,” “misty forest,” “desert” or “abandoned building” prompts Midjourney to allocate more space for environmental elements, leaving ample room for a full-body character within that setting. It isn’t always a guarantee, but it certainly helps.

Also keep in mind the nature of the setting– a gothic library or castle will sometimes result in frustratingly tight spaces that don’t lend themselves to full-body images. But this is where it is our job as promptengineers to keep experimenting and teach Midjourney from our results.

*photography of futuristic soldier exploring a vast abandoned building, dystopian, gritty, dynamic lighting, hassleblad --ar 3:2*

*photograph of man in top hat and long coat walking through the streets of London in 1888, gas lamps, dim lighting, atmospheric --ar 3:2*

4. Opt for Vertical Image Formats

Creating full-body images in a horizontal format can be challenging. Believe me, it’s 90% of what I do. I’ve more recently given in to generating vertical images if I know I’ll be posting to Instagram, but generally, I do prefer a landscape format. It’s a habit I can’t fully shake. But since Midjourney tends to optimize for half-body compositions in horizontal layouts, consider using vertical image formats like “--ar 4:5” or “--ar 2:3” when aiming for full-body character portrayals. This format will provide more space to display the subject’s entire body, resulting in more satisfying outcomes. So when Midjourney is being stubborn, give the vertical orientation a try.

*photography of woman in silver shoes and red formal gown in a gothic cathedral, diffused lighting, hazy, moody, hassleblad --ar 4:5 --s 500*

*photography of archeologist in brown boots holding a torch in an ancient tomb, dusty, diffused lighting, hazy, mysterious, hassleblad --ar 4:5*

5. Avoid Excessive Facial Details

This is hard if you’re creating a specific look and feel with your prompt, but for effective full-body portraits, steer clear of overwhelming your prompts with intricate facial details or expressions. Think of Midjourney as a bot with very limited attention span, while general descriptions like “smiling” or “sad” are acceptable, specifying specific facial features can limit the AI’s focus to the face, reducing the chances of generating full-body images. Removing unnecessary details, such as eye color, from your prompts can significantly improve your results.

*photography of a couple in black hiking boots standing on top of a mountain, smiling, victorious, facing the camera, dynamic lighting, hassleblad --ar 4:5*

*photography of alternative girl wearing red boots standing in the rain, sad expression on her face, moody, side view, hassleblad --ar 4:5*

Alternatively, you could also experiment with weighing your prompts. For instance, the full-body portion of the prompt is most important, so I’ve given it a weight of 2, but since I’d also like my subject to have green eyes, I tried giving it an equal weight of 2. That’s where I cross my fingers and hope for the best.

*full body photography of barefoot woman standing on a beach::2 with green eyes::2 tropical vibes, sunshine, hassleblad --ar 3:2 --s 500*

It took some doing, and even then I had to use the zoom and pan features in order to get a proper full-body image. Midjourney gave me everything from gorgeous eyes and close up portraits, to beautifully rendered feet on a beach. But it had trouble interpreting the full prompt. I’ll have to play with this some more.

Looking Beyond Midjourney

While we tend to focus heavily on Midjourney here at Pink Horn, I’ve more recently started to dabble a bit with Leonardo AI and Blue Willow. In prior side-by-side testing, Leonardo AI held its own rather nicely against Midjourney. Blue Willow runs on Discord, just like Midjourney, right down to the /imagine prompt, making it easy to input the same prompts across the two Bots (or servers). But while it’s got its strengths, I’m not quite sold on its photorealistic capabilities just yet but I think there’s a lot of promise and room for improvement.

I used one of my older prompts from this article to compare how the three generators understood the simple “full body” prompt.

full body photograph of alternative woman in flowy black dystopian clothing walking through the desert, cinematic, dynamic lighting, high detail, leica m11

The results are interesting. While Midjourney’s photorealistic quality remains top notch, both Blue Willow and Leonardo understood the “full body” prompt without the need for extra prompting or tricks. Leonardo’s PhotoReal (paid feature) generator continues to impress me. While the final image is not quite as striking as Midjourney’s, it’s a very close second and convinces me to keep experimenting with Leonardo.

I hope these were helpful in enabling you to create more dynamic full-body portraits in Midjourney. Happy prompting!