Five Women Wearing The Same Dress

How to Actually Get Five Women Wearing The Same Dress Right in Midjourney

I've spent more time than I care to admit wrestling with this particular prompt. Everyone tries it once and gets something that looks like five photoshopped heads pasted onto generic mannequin bodies. Here's what actually works. Start with your base prompt structure. You want something like this at minimum: "five women wearing the same dress, standing in a line, full body shot, photorealistic, natural lighting." But if you stop there, you're going to get garbage. The AI keeps morphing the figures together or making them look like clones. It happens every time. The trick that took me forever to figure out is adding spatial separation cues. Words like "spaced apart," "each woman standing individually," "clear separation between figures." Without those, Midjourney sees "five women" and merges them into one blobby mass because the model doesn't inherently understand discrete human spacing. I learned this after accidentally generating what looked like a Greek chorus made of dough for about three hours straight.

Five Women Wearing The Same Dress

Here's the prompt that actually worked for me on the third or fourth revision: "five women wearing the same floral midi dress, each woman standing separately spaced apart, full body portrait, natural outdoor lighting, candid photography style, individual poses, distinct body types, realistic skin textures, shallow depth of field, shot on 35mm lens, not AI generated look --ar 16:9 --stylize 150" The key additions here are "distinct body types" and "individual poses." Without those, the AI makes all five women look identically proportioned. They'll have the same face shape, same posture, same everything. Adding "distinct body types" forces some variation that makes the image read as five separate people rather than one person cloned five times.

I also recommend running it at V6 mode if you're on Midjourney. V5 tends to make the figures look plasticky and uncanny. V6 handles multiple subjects significantly better, though it's not perfect. The --stylize parameter matters too. I keep it around 150 because higher values make the AI go creative and start changing the dress or the composition. Lower values keep it closer to your prompt intent. One edge case I ran into: if you specify a dress pattern that's busy, like florals or stripes, the AI sometimes starts blending the patterns across different women's dresses as if they're sharing fabric. The fix was adding "identical dress print consistent on each woman" to the prompt. It sounds redundant but the AI needs that reinforcement. I don't know why. It just works. Another thing nobody mentions: generating five figures is computationally heavier than generating one. Your process time will be longer, maybe 45 seconds to a minute per generation instead of the usual 20 seconds. And you'll burn through your fast hours faster. Budget accordingly if you're on a limited plan.

If you're getting repeated failures after three or four attempts, stop and switch tactics. Sometimes the solution is generating the women individually and compositing them, which takes more work but gives you control over spacing and poses. I do this when the automated generation keeps producing ghostly overlapping figures that look like a double exposure gone wrong. Bottom line: this prompt works but it requires patience and specific vocabulary additions that the default suggestions won't give you. Start with the base structure above, tweak the spacing language until the figures separate cleanly, and don't expect it to nail it on the first try.