Midjourney, a prominent AI image generation startup, made headlines recently with the launch of its highly anticipated AI video generation model, V1. This innovative model allows users to upload an image or utilize one generated by Midjourney’s existing models, and in return, V1 produces a series of four five-second videos based on the image. The unique feature of V1 is that it is exclusively accessible through Discord and is currently only available on the web.
With the introduction of V1, Midjourney enters a competitive market alongside other companies like OpenAI, Runway, Adobe, and Google, all offering AI video generation models. While many competitors focus on developing controllable AI video models for commercial applications, Midjourney has distinguished itself by catering to creative individuals with its distinctive AI image models.
CEO David Holz revealed in a blog post that V1 is just the beginning of Midjourney’s ambitious plans for AI video models. The company envisions creating AI models capable of real-time open-world simulations, with future developments in 3D renderings and real-time AI models on the horizon.
However, Midjourney recently faced legal challenges from Hollywood giants Disney and Universal, who sued the startup for allegedly depicting copyrighted characters in images generated by its AI models. The entertainment industry has expressed concerns about the impact of AI tools on creativity and intellectual property rights, leading to accusations that these products are trained on copyrighted works.
Despite positioning itself as more focused on creativity than commercial applications, Midjourney has not escaped scrutiny. The startup plans to charge significantly more for video generation than image generation, with pricing options ranging from a $10 Basic plan to higher-tier Pro and Mega plans offering unlimited video generations.
V1 includes customizable settings that allow users to control the video model’s outputs, such as automatic animation, manual animation descriptions, and toggling camera and subject movement levels. While V1’s videos are currently limited to five seconds, users can extend them by four-second increments up to 21 seconds.
Early demos of V1’s videos have been well-received, showcasing a unique and somewhat otherworldly aesthetic. As V1 competes with established AI video models in the market, its true potential and capabilities will become clearer in the coming months.
