- Midjourney has launched its first AI video model, V1.
- The model lets users animate images into five-second motion clips.
- The tool is relatively affordable and a possible rival for Google Veo or OpenAI’s Sora.
Midjourney has long been a popular AI image wizard, but now the company is making moves and movies with its first-ever video model, simply named V1.
This image-to-video tool is now available to Midjourney’s 20 million-strong community, who want to see five-second clips based on their images, and up to 20 seconds of them extended in five-second increments.
Despite being a brand new venture for Midjourney, the V1 model has enough going on to at least draw comparisons to rival models like OpenAI’s Sora and Google’s Veo 3, especially when you consider the price.
For now, Midjourney V1 is in web beta, where you can spend credits to animate any image you create on the platform or upload yourself.
To make a video, you simply generate an image in Midjourney like usual, hit “Animate,” choose your motion settings, and let the AI go to work.
The same goes with uploading an image; you just have to mark it as the start frame and type in a custom motion prompt.
You can let the AI decide how to move it, or you can take the reins and describe how you want the motion to play out. You can pick between low motion or high motion depending on whether you want a calm movement or a more frenetic scene, respectively.
The results I’ve seen certainly fit into the current moment in AI video production, both good and bad. The uncanny valley is always waiting to ensnare users, but there are some surprisingly good examples from both Midjourney and initial users.
AI video battles
Midjourney isn’t trying to compete head-on with Sora or Veo in terms of technical horsepower. Those models are rendering cinematic-quality 4K footage with photorealistic lighting and long-form narratives based solely on text. They’re trained on terabytes of data and emphasize frame consistency and temporal stability that Midjourney is not claiming to offer.
Midjourney’s video tool isn’t pretending to be Hollywood’s next CGI pipeline. The pitch is more about being easy and fun to use for independent artists or tinkerers in AI media.
And it really does come out as pretty cheap. According to Midjourney, one video job costs about the same as upscaling, or one image’s worth of cost per second of video.
That’s 25 times cheaper than most AI video services on the market, according to Midjourney and a cursory examination of other alternatives.
That’s probably for the best since a lot of Hollywood is going after Midjourney in court. The company is currently facing a high-stakes lawsuit from several Disney, Universal, and other studios over claims it trained its models on copyrighted content.
For now, Midjourney’s AI generators for images and video remain active, and the company has plans to expand its video production capabilities. Midjourney is teasing long-term plans for full 3D rendering, scene control, and even immersive world exploration. This first version is just a stepping stone.
Advocates for Sora and Veo probably don’t have to panic just yet, but maybe they should be keeping an eye on Midjourney’s plans, because while they’re busy building the AI version of a studio camera crew, Midjourney just handed a magic flipbook to anyone with a little cash for its credits.
Leave a comment