Midjourney, a company known for its creative imagery, introduced its first video generation model V1, and is directly competing with industry Titans such as Openai, Adobe, Google and Runway.
Released Wednesday, V1 can create videos generated from a single image into four 5-second AI, whether it was sent by a user or created using MidJourney's image model. This feature is currently only accessible online and is originally offered through Discord, which supports Midjourney's traditional platform strategy.
In contrast to models that prioritize commercial-grade materials, such as Google's VEO and Openai's SORA, Midjourney's V1 supports a creative-first approach for startups.
The user demos are not photorealistic, but emphasize creative goals rather than industrial goals, as evident from the initial output of their stylized aliens.
CEO David Holz characterized the V1 as a component of Midjourney's larger initiative, providing the tool for “real-time open world simulation” in a blog post that coincided with its debut. He added that AI models for 3D rendering and interactive content creation are part of future plans and could indicate a shift towards gaming or metaverse applications.
But the launches happen at a controversial moment. Two important Hollywood studios, Disney and Universal, sued Midjourney last week after allegedly using copyrighted characters in AI-generated visuals.
The lawsuit is due to industry concerns regarding training these AI systems and the growing potential impact on future creative work.
The lawsuit highlights the legal dangers that Midjourney faces as the AI business grows, even if it tries to separate itself from its competitors with a focus on creativity and individual expression.
New pricing and usage guidelines are also included in the V1. The video generation is about eight times more expensive than the image generation, so consumers run out of subscription credits more quickly. High-rise users who pay $60 or $120 a month have unlimited video access in the slower “relaxed” mode, but the company's $10 basic plan only allows limited access to the V1.
Custom factors are included in V1 to help developers customize the results. Users have the option to use a text prompt to select from automatic random animations to explicitly describe the desired movement. Additionally, the movie can be extended by 4 seconds at a time, with each clip going up to 21 seconds, and motion levels can be changed to “low” or “high.”
The release of the model will strengthen competition in areas that are already busy and rapidly changing. Joining Midjourney is Openai's Sora, Adobe's Firefly, Google's Veo, Runway's Gen-4 are all fighting to create all-loyal, controllable AI video solutions.
The V1 generally receives excellent early reception despite its abstract looks and limited control options. According to analysts, its creative focus may appeal to a professional yet passionate group of digital storytellers and artists.
