Midjourney adds end frames and seamless loops

Generating AI video used to feel like pure gambling, especially when you needed a clip to end on a specific visual or loop seamlessly without glitching.

Midjourney just shipped a major update to its video capabilities, adding custom start and end frames, automatic video looping, and direct access inside the Discord bot.

Controlled morphs and endless loops

Creating consistent transitions between two distinct scenes has always been a pain point in generative video. Midjourney is addressing this by letting users define both the start and end images for a generation.

By adding the end frame argument along with an image link to your prompt, the model smoothly morphs the initial picture into your target ending. This makes narrative transitions and shape changes far more predictable.

For creators building website backgrounds, digital signage, or social media graphics, there is also a new loop parameter. Adding the loop argument tells the system to align the final frame with the opening shot, generating a seamless visual loop right out of the box.

You are also no longer restricted to Midjourney creations. Users can paste an external image link at the beginning of their prompt alongside the video argument to animate third-party artwork or real-world photos.

Video creation lands on Discord

The update also integrates video generation straight into Discord, where the community originally started, though Midjourney notes that the web interface still offers a smoother overall experience.

Inside Discord, users can click the Animate button directly beneath an upscaled image. If you have remix mode turned on, you can write a custom prompt to guide camera movement and character action.

Videos can also be extended sequentially, and remix mode lets you tweak the prompt for each new chunk of time. To get crisp results, users can tap virtual upscale buttons on the finished video to export a higher-quality mp4 file.

This drop follows the recent rollout of Midjourney v7 and its draft mode, though the company continues to battle legal pressure, including high-profile copyright lawsuits from Disney and others.

Why it matters

We think end-frame control is the line that separates a tech demo from a reliable production tool.

Before this, getting an AI video to land precisely where you wanted was nearly impossible without hours of rerolls or heavy post-processing in external editing suites. Being able to anchor both ends gives motion designers and marketers actual control over transitions, morphs, and product reveals.

Opening the feature to external image links is also a very smart move. It means creators are not locked into Midjourney's aesthetic style just to use its motion engine.

That said, Midjourney is playing catch-up here. Tools like Runway and Kling have offered keyframe and end-frame controls for months. Plus, with Disney and major media companies taking Midjourney to court over copyright infringement, expanding aggressively into video will likely only turn up the legal heat.