Midjourney v7 Review: The good, the bad, and the road ahead

The latest version of the beloved AI image generator is here

Since its 2022 launch, Midjourney has consistently set the bar high in the AI image generation space. After almost a year, the much-awaited Midjourney v7 is here, and like every major update, it comes loaded with exciting possibilities and a few head-scratchers.

Is it good? bad? Does it have video? 3D? Should you get it?

Take it easy. Let’s dive in.

The Good

1. Draft Mode

Draft Mode stands out as one of v7’s strongest new features:

While in Draft More, instead of typing your full prompt again, just type the changes you want, and the prompt will be fully updated. In the above example, after generating the images in the first row, I just typed “Change uniform for stillsuit” and MJ updated the prompt.

It churns out images up to 10 times faster and consumes less credits, making it ideal for rapid ideation, experimenting with variations, or just chasing inspiration. It’s fast, fun, and refreshingly affordable.

2. Voice Prompting

Perhaps the coolest innovation yet, v7 introduces voice prompts. No more typing intricate descriptions. Just speak your idea aloud, and Midjourney translates your words into image prompts. It’s even better with Draft Mode, the iteration is just fast. To use it, just activate the microphone icon in the top bar.

Get ready for vibe-prompting.

3. Stunning Visuals
The rendering of detail and lighting is nothing short of exceptional. From intricate scenes to sophisticated textures, v7 flexes some impressive muscle, delivering results that are more photorealistic than ever.

However, this was already Midjourney’s strength with v 6.1. These are a few thing it does better now:

Unique styles: Midjourney was already the King of Styles. The new version keeps the very popular sref styles.

Photography, Dynamic shot, unusual angle, woman with biomechanical heavy armor in combat charging forward holding a giant broadsword, gray eyes, motion, blur, action, bokeh, fire, sparks, nebula, --chaos 10 --ar 3:4 --sref 1512812739 3304908547 --p --sv 4 --sw 300 --stylize 1000

---

More details: It can handle many more subjects and/or details than previous versions. See this example:

Photography of a white-haired man standing at an Icelandic black beach. The man is holding a fat black cat in his arms and a black umbrella in his shoulder. The man is wearing a dark green sweater and a black cape long cape flowing in the wind. The weather is stormy with lightning bolts. There is an old ship sailing in the sea. A seagull stands nearby.

---

The Bad

1. Turbo and Draft Mode Only - Fast Burnout
The lack of a clear standard mode at launch has already led to some painful surprises. Early adopters have burned through their "fast hours" unknowingly due to unclear UI prompts.

Remember: Turbo jobs cost 2x more than a normal V6 job and Draft jobs half as much.

---

2. Hands, Text, and Anatomy are still awkward after all these years
Some bad things just keep coming back.

Hands and human anatomy, perennial problems in generative AI, remain problematic. Midjourney v7 claims improvements here, yet initial tests feel disappointingly familiar, hands still look less like graceful limbs and more like experimental art pieces.

And don’t get me started with text. I used the following prompt:

A street sign with the text "Let's test Midjourney Text Generation", busy city life in the background

It wasn’t a big deal in v6. But since then, we've seen Ideogram and GPT-4o handle text like champions. I was really hoping for at least a slight improvement here.

---

3. Alpha Stage Limits - Not Fully Baked
As of now, v7 is firmly in alpha territory. Key features like upscaling, inpainting, and advanced editing still rely on the v6 model. It’s a bit like opening a shiny new toy box only to find last year's batteries inside.

These are the features currently available:

And no. No video integration. No 3D. Yet.

---

Final Thoughts

Midjourney v7, despite its quirks and alpha-stage growing pains, sets the foundation for impressive future potential.

Keep in mind it appears to be designed for explorers rather than perfectionists, those who enjoy discovering beauty in creative chaos rather than demanding pinpoint accuracy.

Is it revolutionary? Not quite yet. But at its best, it’s a blast, balancing state-of-the-art quality with features that genuinely streamline creative processes.

What’s your take?