Midjourney’s Niji 7 is here. Let's see what it can do
Niji 7 is Here: Why Midjourney still owns the anime aesthetic in 2026
Niji is back! And it’s more literal than ever.
Niji 7 has officially launched, marking the first major update to Midjourney’s anime-specific model in over a year. Developed in collaboration with Spellbrush, “Niji” (Japanese for “rainbow”) isn’t just a filter; it’s a total retuning of Midjourney’s engine to master the unique “line language” of Eastern animation.
While the rest of the industry is chasing “fast and easy” with models like Nano Banana Pro, the Niji team is obsessing over the physics of linework and the beauty of empty space. If you’ve ever felt the frustration of an AI “over-rendering” your anime art until it looks like a shiny 3D figurine, Niji 7 is the fix you’ve been waiting for.
How to activate it
Getting started is simple. Just append --niji 7 to the end of any prompt.
On the web interface, you can select it directly from the “Version” dropdown in your settings.
Is Niji 7 an improvement?
Short answer. Yes.
For the past year and a half, we’ve been waiting for a leap in anime quality and detail, and Niji 7 delivers it.
Prompt: A girl with short green hair, hair in a single bun, blue eyes, fang, smile, standing under the sakura trees, visible sakura flowers --ar 16:9 --niji 7
We’re talking about high-definition upgrades where you can see individual flower petals and the reflected light in a character’s eyes.
But.
Is it better at following prompts?
Sometimes.
A samurai girl with long black hair, hair in a single bun, red eyes, fang and smile, with four arms, each of them holding a gun, fire in the background, wind and sparks --ar 16:9 --niji 7
No problem so far. But let’s increase the —Stylize parameter.
Uh, oh. We are missing two arms. It appears to prioritize cool stuff over the prompt.
Is it better at text?
Let’s test it with this prompt:
A tactical overlay on a video call. Futuristic Heads-Up Display (HUD) interface overlaying a video conference call. The background is a video feed of a woman’s face. Overlaid on the face is a glowing cyan tactical targeting square framing the eyes. Text readouts in a technical font hover next to the square: “STATUS: VERIFIED HUMAN” The overall aesthetic is military-grade technology, clean lines, transparent glowing cyan data on dark background. --chaos 10 --ar 16:9 --stylize 1000 --niji 7
Ugh. It’s not legible. The style is great, though.
And finally.
Does it work with sref codes?
Let’s compare:
Close up, towering robotic Dragon with dark, armor-plated, metal wings, glowing blue eyes shooting laser beams, flying on the void of deep space The robot’s body is adorned with blue bolts and neon lights, contrasting against the metal frame. Futuristic, sleek design, long, powerful tail extends behind the robot, swishing through the air with sharp, smooth segments, The mech’s eyes glow intensely. Realistic, high-detail, cinematic style. --ar 3:4 --sref 1549001673 --stylize 500
Both are great. But the anime art is way more dynamic.
A dynamic woman adventurer, illustrated in color, in the art style of Jim Lee: hyper-realistic figure with exaggerated but believable anatomy; dramatic, cinematic poses with foreshortening and perspective; highly detailed clothing with folds, seams, and texture; expressive face with intense emotion; clean, crisp, highly polished linework with bold blacks and high-contrast shadows emphasizing volume and depth; intricate environment with depth, rendered in sharp lines and crosshatching; heroic, epic, and dramatic composition, reminiscent of 1990s comics, cinematic lighting, high detail, ultra-sharp focus, dynamic energy. --ar 3:4 --sref 1232499358 --stylize 1000
No debate here. I can just dive an entire weekend exploring sref combinations with Niji 7. It’s a pleasure. The styles are great. And no other tool comes near that.
What do you think?
My take: The Midjourney moat in 2026
Midjourney is taking a fundamentally different path than the “Google/OpenAI” giants. While tools like Nano Banana Pro are winning the “workflow” war, making it incredibly easy to deliver standard commercial assets, Midjourney is doubling down on Deep Taste.
They are obsessing over things that seem small, like linework and spatial coherency, but these are exactly the things that matter to serious artists and character designers. If they keep pushing in this direction, they won’t be swallowed by the “big tech” models; they will remain the premium choice for creators who care about craft over convenience.
There is a catch.
Midjourney is still “Hard Mode.” You can’t just talk to it like a friend yet (at least not with the same ease as Nano Banana). If you’re producing professional game assets or a manga, you need to put in the work to master the parameters like sref codes and stylize.
Curious, btw. Are you still using Midjourney?












Has Midjourney made it's creations private yet? Last I knew, you prompted, your art showed up in the public feed, and you downloaded it from there.