top of page
Search

Midjourney's Video Creator: How Good Is Their First Version?

midjourney blog post image
A Midjourney generated image using Midjourney Automation Suite

Midjourney recently released its first video generator, V1. Many people wondered if it would be as good as its image generation. Initial tests show it is surprisingly better than expected, and it offers new options for creators. Let's explore what this new tool can do and how it compares to other video generators out there.

First Look: Midjourney's Video Skills

At first, some thought the video quality might not be great. But after hundreds of tests, it's clear Midjourney's V1 stands out. While other tools like V3 have their strengths, Midjourney's video generator is competitive in many ways. It creates videos at 480p resolution, and clips start at 5 seconds. You can extend these clips up to about 21 seconds, which is a good length compared to many other models.

Even though the resolution is 480p, the videos look crisp. High resolution does not always mean better quality, and Midjourney's V1 proves this with its clear output. It might not be as realistic as V3 in all cases, but it gains ground in other areas.

  • Midjourney V1 supports multiple aspect ratios, unlike some other models that are limited to 16x9. This gives you more creative freedom.

  • The base image quality from Midjourney helps its video output. Since Midjourney already excels at generating images, this foundation carries over to video well.

  • While some other generators offer very tight control over prompts, Midjourney V1 offers a little less prompt focus. However, you can still get great results by trying different prompts. This new feature is a strong entry into the video generation space.

Generating Your First Motion Clips

Starting a video in Midjourney is straightforward. You can begin with an image as your first frame. From there, you have two main options: auto animate or manual animation. Manual mode lets you add a specific prompt and choose between low or high motion.

A great benefit of Midjourney's video creator is that it makes four videos at the same time. This gives you multiple options to pick from without having to generate each clip separately.

Testing the Boundaries: Real-World Examples

Astronaut in Space

We started with an image of a man in a spacesuit floating with a galaxy background. Applying motion manually and setting it to high motion created impressive results. The character stayed clear, the edges were crisp, and the galaxy background moved nicely. Even details like reflections in the helmet were kept stable.

Midjourney also lets you extend clips. We tried an auto-extend, which defaults to low motion. This extended the clip to 9 seconds, already longer than some other video models allow. The quality remained good for this initial extension.

However, when we tried to extend clips multiple times for a very long video (e.g., beyond 9 seconds to 13 or 17 seconds), we saw some quality drop. The more the video extends and the character moves, the more likely inconsistencies appear.

Are you looking to create numerous video clips and manage diverse outputs? TitanXT's Midjourney Automation Suite can help streamline your workflow, making batch processing and organization simple.

The "Gymnastics Challenge"

Next, we tested a common challenge for AI video tools: complex human movement, specifically a girl doing an Olympic balance beam routine. This kind of motion often breaks video generators due to the complex physics involved. Midjourney V1 showed its limits here, with some body distortion and odd head movements. It still performs better than many other generators at handling such movement, but it shows where AI video still needs to grow.

Animal and Vehicle Motion

Testing a horse running in a field revealed strong details in the horse's movements and natural reactions like hair flowing. The system did well with multiple moving elements. However, when we tried to extend the clip and add a new element, like a man the horse was chasing, Midjourney V1 did not include the new element. This shows a current limit in its ability to add new specific details during extensions.

Car racing scenes also presented interesting results. Initially, cars sometimes spawned in odd ways or moved backward before going forward. After the first few frames, though, the racing action looked good. Even a prompt for a "multi-car crash" offered some compelling results, though the track itself sometimes changed.

Managing a wide variety of video ideas and outputs can be time-consuming. TitanXT's Midjourney Automation Suite helps you handle your video generations from different prompts, saving you time and effort.

What Midjourney V1 Does Well

  • Midjourney V1 creates natural, fluid motion. It avoids the "slow-motion" feel seen in older AI video models like Cling 1.6.

  • It does a good job keeping characters consistent and reducing morphing, even with a lot of movement.

  • The tool maintains the original image style throughout the video, which is not always the case with other generators.

  • Its support for multiple aspect ratios offers more creative options for users.

  • Compared to some alternatives, the cost per generation appears very competitive for the quality it delivers.

Current Drawbacks and Future Potential

  • The 480p resolution is a current limit. However, Midjourney may add upscaling features later, which could greatly improve video quality.

  • Generating videos uses a good amount of your "fast hours" on Midjourney plans. While larger plans get unlimited generations, they will switch to "relaxed mode" once fast hours run out, meaning longer wait times.

  • Prompt adherence struggles with adding completely new objects or complex elements on extended clips, requiring careful prompting or starting fresh.

  • Complex physics and advanced body movements (like gymnastics) are still challenging for the AI to get perfectly right.

  • Quality can degrade with many long extensions, especially when significant or varied movements occur.

Midjourney's first video generator is a strong start. While it has some areas to improve, its ability to create crisp, natural-looking motion from images is impressive. As with any new AI tool, user feedback is important to help it get better. If you use it, be sure to give Midjourney your thoughts.

Eager to explore Midjourney's new video features and optimize your creative process? Streamline your efforts and maximize your output with TitanXT's Midjourney Automation Suite.

 
 
 

Comments


bottom of page
Midjourney Automation Suite - Automate your image generation workflows on Midjourney | Product Hunt