Midjourney has consistently set the bar for advancements in video creation and other forms of creatives. The recent release of the Midjourney 6.1 model has once again taken the internet by storm.
Shortly after its launch, the news spread on X, highlighting it as the most advanced AI art model to date.
Midjourney 6.1 brings new updates, enhanced image coherence, new 2x upscalers boost image and texture quality, with processing speeds about 25% faster. Moreover, the text accuracy in prompts is improved, and a new personalisation model adds more nuance, surprise, and accuracy to the generated content.
Here are seven examples of Midjourney 6.1 generated creatives:
Midjourney 6.1 + Runway 3 Combo
A user tried Midjourney V6.1 and Gen-3 i2v with the prompt “a girl laughs as a fire swirls far away behind her”. Runway handled the fire well, but 'far away' was added to avoid flagging. Despite a bit of hair being caught, the image is expressive, aligns with their style preferences, and the personalisation parameter (--p) offers more customisation than V6."
Updated Personalisation
Version 6.1 is now more personalised, offering enhanced nuance, surprise, and accuracy. This system allows for quicker and more precise adjustments. Users can continue using their current personalisation style with the–p code, and previous personalisation codes remain compatible.
As users rank more images, the system will automatically evolve to reflect these updates. For example, a user on X provided the code "night”, and the system generated an image of a nighttime scene.
Another example is the code "dream”, which produced this image.
Creation of cinematic scenes
Several cinematic scenes were created using Runway ML Gen-3's image-to-video feature. The images were generated with Midjourney 6.1, based on specific prompts. For example, one of the images was prompted to depict the subject with a “seductive smile”.
Midjourney + Luma Dream Machine
A user on X, created a video with the image generated with Midjourney v6.1 and Dream Machine. Sharing the experience, they said, “Taking this as the base image and popping it into Dream Machine is so next-level. 2024 truly is the year for AI video. I'm exploring a few Greek myths currently, so cool seeing all this come to life.”
Creating Abstract images
A user on X highlighted that Midjourney excels at generating abstract images with outstanding aesthetics. They shared two prompts: one for a high-chaos, stylised optical fibre image (optical fibre --chaos 10 --ar 16:9 --sref 1558173677 4244977961 1214440399 --p --sw 800 --stylize 800 --weird 3 --v 6.1)
another for optical fibre integrated into a landscape (optical fiber along a landscape --sw 25 --sref 531168650 2959860992 --ar 7:3 --v 6.1). Both prompts are crafted to produce visually striking images.
Blending SREF Codes with Midjourney
A user successfully integrated SREF Codes (parameters that guide the AI in producing specific visual outputs) with Midjourney v6.1 to create synthetic life forms using specific parameters. The process involved:
- Step 1: Generate a movie still with --sref random, --sw 1000, and --s 0.
- Step 2: Evaluate satisfactory images and run additional tests (product and street photography) with --sref [CODE], --sw 100, and --s 300.
- Step 3: Select the best codes from Step 2 and test with complex, multi-prompts for optimal results.
2x Upscaling
In Midjourney 6.1, users have different preferences for upscalers. The Creative upscaler is widely appreciated for its improved artifact removal and overall rendering quality.
However, some users, particularly those working with images that have heavy textures like film grain and paper overlays, prefer the subtle upscaler, as it preserves these textures better and avoids the more artificial look that creative can introduce.
Additionally, version 6.1 is faster than 6.0, has fewer artifacts, and offers a notable improvement in text integration within generated images.
- Jev Isn’t An Overnight Fix For Your Lack of AI Fundamentals
- Internet on Alert as AI Uncovers Critical Flaws in PHP Used by 70% of Websites
- From Claude Code Mods to Super Mario: What People Are Building on Jev
- 8 Reasons Why Anthropic’s New Report is an AI Doomer’s Worst Nightmare
- How VECV Built a Real-Time Data Backbone for 2,50,000 Connected Vehicles
- Anthropic Stands Alone Against Open Models