Midjourney v6.1 Update: Major Enhancements in Image Coherence and Detail Quality

Midjourney v6.1 is now officially live, finally addressing user complaints about distorted limbs and pixel artifacts. The new version delivers noticeable improvements in image coherence, detailed textures, and generation speed, while also bringing back the classic --quality parameter adjustment. If you're using Midjourney for design or content creation, these upgrades in v6.1 are worth a closer look.

Significantly Enhanced Image Coherence: More Natural Limbs and Faces

v6.1 focuses heavily on improving the coherence of visual elements. Generated arms, legs, and body proportions are now much more natural, without the strange distortions or misalignments seen before. Even when faces occupy a small portion of the frame, facial features remain clear. Hand and eye details at a distance are also far more accurate compared to v6.0. This means generating full-body shots or complex scenes with Midjourney v6.1 will require much less post-editing work.

25% Faster Generation with Richer Detail Textures

The new model generates standard images about 25% faster than v6.0. If a single image used to take 8 seconds, it now takes around 6 seconds. More importantly, v6.1 doesn't sacrifice image quality for speed—it actually enhances skin texture, layers of detail, and color performance. For designers who need to batch-produce images, this improvement can significantly boost workflow efficiency.

--quality Parameter Returns: Easy Control Over Detail and Speed

v6.1 re-expands the --quality parameter range, now supporting three settings: 0.5, 1, and 2. The default value of 1 offers balanced performance. Setting --q 0.5 allows faster generation for sketches or prompt testing, while --q 2 uses 25% more GPU time to deliver richer textures and finer details. This adjustment lets users flexibly allocate computing power based on their needs—whether for rapid iteration or high-fidelity output.

Improved Text Generation Accuracy: Quotation Marks Matter More

v6.1 achieves significantly higher accuracy when generating images containing English text. If you need specific text to appear in an image, remember to enclose the target text in quotation marks in your prompt. That way, Midjourney v6.1 will prioritize rendering the text correctly. Although support for Chinese text still has room for improvement, this enhancement is already valuable for users creating posters, logos, or brand materials.