Midjourney V6.1 New Feature Breakdown: Upscaler, Text Accuracy, and How to Use --q 2

After Midjourney updates to V6.1, the most obvious change isn’t “more realistic,” but more stable: character limbs are more coherent, details are cleaner, and image generation is faster. This article breaks down the key new features in Midjourney V6.1 and gives you ready-to-copy usage so you can get started immediately in Discord.

Improved visual coherence: characters no longer feel “assembled”

When using Midjourney to generate characters, what many people fear most is hands and feet going wrong. V6.1 focuses on improving the continuity between arms, legs, hands, and the body. You’ll find it easier to get characters with natural poses and unified structure; animals and complex limbs are also more stable.

If you create e-commerce posters or character designs, this kind of “structural stability” improvement is highly valuable: when you generate repeatedly with the same prompt set, the rate of unusable outputs drops noticeably.

Texture and artifact handling: cleaner and more tactile

Midjourney V6.1 further reduces pixel artifacts while strengthening skin, material textures, and layering, making the image feel more “like a complete finished work.” Some retro looks (such as 8-bit style) are also rendered more accurately, making it well-suited for stylized visuals.

In addition, “details that easily get blurry,” such as eyes, small faces, and hands in the distance, are more precise in Midjourney V6.1, so they’re less likely to fall apart when viewed enlarged.

New upscaler and speed boost: get usable large images faster

V6.1 introduces a new upscaler (Upscaler) that focuses on improving image and texture quality; combined with the usual U-button upscaling, final images look cleaner and sharper. At the same time, standard image generation is about 25% faster, which is very noticeable when you’re on a deadline.

For people who iterate repeatedly, this round of “faster + clearer” from Midjourney is a real efficiency gain: you can spend time choosing directions instead of waiting in the queue.

More reliable text generation: use quotes to lock in the words you want

Midjourney V6.1 is more accurate when generating images that contain text—especially when you put the text you want to display in quotation marks within the prompt, which boosts the hit rate. For example, writing “SALE” or “咖啡馆” in the prompt is more likely to produce a close match than leaving out the quotes.

It’s recommended to keep the text length to short words or short phrases, and pair it with clear scene descriptions like “poster,” “sign,” or “front of packaging,” so Midjourney better understands where the text should appear.

How to use --q 2: trade more time for finer textures

V6.1 adds the --q 2 mode, which spends more generation time in exchange for richer texture detail; the official note says it takes about 25% more time and may sometimes slightly sacrifice coherence. You can treat it as a “texture-first” switch—great for product materials, clothing fabrics, and close-up shots.

It’s simple to use: after entering /imagine in Discord, just add --q 2 at the end of your prompt. If you care more about structural stability (such as group photos or complex actions), prioritize the default settings first, then try Midjourney’s --q 2 on specific parts as needed.