Midjourney, a pioneer in AI-based image generation, has launched its V7 AI image model, marking its first major update in nearly a year. The highly anticipated model began rolling out in alpha on Thursday night (EST), just a week after OpenAI introduced its own image generation feature within ChatGPT, which went viral for its ability to produce Ghibli-style images.
However, while Midjourney’s V7 is not specifically optimized for Ghibli-style art, it offers an impressive range of stunning visuals. Early reviews suggest that the model generates aesthetically pleasing and detailed images that cater to a variety of creative needs.
What’s New with Midjourney V7?
Midjourney CEO David Holz described the new V7 AI model as a complete overhaul, offering smarter and more refined image generation capabilities. The model is built on a “totally different architecture” that provides superior performance when it comes to handling text prompts. According to Holz, V7 improves image quality significantly, with better textures, more realistic bodies and hands, and enhanced coherence across all objects.
The new model comes in two versions: Turbo and Relax. Turbo is more expensive to run but delivers faster rendering, while Relax offers a more budget-friendly option for slower, high-quality image creation. Draft Mode, a new feature within V7, is also available, which allows users to generate images at 10 times the speed and half the cost of standard mode. Although these images are of lower quality, they can be enhanced with just a click.
Personalization by Default: A Key Feature of V7
One of the most exciting features of Midjourney V7 is that personalization is enabled by default. Users are required to rate about 200 images to create a personalized profile, which tailors the model to their specific visual preferences. Once the profile is created, users can easily toggle between V7 and previous models on the website or via the Midjourney Discord chatbot. Selecting the model is as simple as choosing it from a drop-down menu next to the “version” label on the web app.
What’s Still Missing in Midjourney V7?
While V7 comes with several advanced features, some well-known tools are still in development. For example, the ability to upscale images or retexture them will not be available immediately, but Holz assured users that these features will likely arrive in the coming months. This delay is expected as the team continues to refine the new model.
Holz further mentioned that V7 may require users to adapt their prompting styles to get the best results. Midjourney is actively seeking user feedback to understand the model’s strengths and weaknesses. As Holz put it, “This is an entirely new model with unique strengths and probably a few weaknesses.”
Midjourney’s Future and Ongoing Innovations
Midjourney continues to push the boundaries of AI-generated imagery. Although it hasn’t sought external funding, the company has been self-sustaining since its launch in 2022. Reports indicate that Midjourney generated approximately $200 million in revenue in 2023, and the company is now expanding into hardware projects to explore new creative possibilities, including video and 3D object generation.
However, despite its success, Midjourney faces legal challenges. The company is currently embroiled in lawsuits that allege it infringed upon artists’ rights by scraping images from the web without consent to train its AI models. This legal battle could impact the company’s future operations, but it remains a key player in the rapidly evolving field of AI-powered creative tools.
