Midjourney's New Videos: First Impressions and What It Does Well

Midjourney, known for its amazing AI image creation, now steps into video generation. This new feature lets you turn your static images into moving clips. We will look at how it works, test it with different types of content, and see how it compares to other AI video tools available today.

How to Make Videos with Midjourney

Making a video in Midjourney is simple. Start by selecting an image you have already made. You can also upload a new image. Once you pick an image, you can type in a prompt to guide the video or leave it blank. You also get some settings to control the video output.

Video Settings

• [P][B]Motion:[/B] This setting controls how much movement is in your video. You can choose from low, medium, or high motion. High motion often includes more camera and subject movement.[/P]

• [P][B]Speed:[/B] At the moment, 'fast mode' is available. 'Turbo mode' might be coming soon. This setting changes how quickly the video gets made.[/P]

After setting your choices, Midjourney starts creating. Videos are currently 480p and typically run for 5 seconds. You can extend them up to 20 or 21 seconds by adding 4-second parts. Videos render quickly, often in about a minute for a 5-second clip. You can also choose to extend manually with a specific prompt, or let the AI extend it automatically.

Want to make your Midjourney video workflow even smoother? Explore the Midjourney Automation Suite from TitanXT. It can help automate parts of your video creation process.

Testing New Midjourney Videos with Different Content

We put Midjourney's video feature to the test with various subjects to see how it handles different types of motion and detail.

People with Specific Actions

One test involved making an eye blink. The AI did a good job of creating the eyelid movement, which was impressive and cool to see.

Challenging Food Scenes

AI tools often struggle with food, especially detailed actions like cutting and eating. We tried to make a video of a chef cutting and eating a steak. Results varied, but one version showed the chef picking up a piece of steak and eating it with surprisingly good mouth movement. This was better than many previous attempts by other AI tools, especially since the chef was not in the original image.

Action and Nature Scenes

For action, we created a video of a robot being cut in half. Some versions had good effects, though one had physics issues with a head turn. Overall, one variation looked really good.

In nature, we tried to change a scene to a "thick forest" with a lama in it. While the scene change did not happen much, the motion of the llama was natural and good. AI video has been getting better at motion, as seen with older tests of a husky running in a daisy field.

Do you generate many Midjourney videos? Speed up your work with the Midjourney Automation Suite from TitanXT. It handles repetitive tasks so you can focus on creative ideas.

Comparing Midjourney with Other AI Video Tools

We compared Midjourney's video feature with Runway, Cling, and Hilu (version 2.0). We used a complex image prompt and then asked each tool to "bring to life" the image with high motion.

The Comparison Results

• [P][B]Midjourney:[/B] The videos often showed a dance-like movement. The output was strongly tied to the original image it created.[/P]

• [P][B]Runway:[/B] Also showed some dancing. While it featured good wind effects, its overall movement quality might not be as good as Midjourney's for this specific test.[/P]

• [P][B]Cling:[/B] Stood out with impressive flashing lights in the background of its video. Even though the main subject’s movement was simple (a walk back and forth), the background effects were excellent.[/P]

• [P][B]Hilu Miniax:[/B] Had an interesting feel. It put fog into the image well, but there was not much movement from the subject.[/P]

The other AI tools often created surprisingly good videos, sometimes even beating Midjourney in the initial image creation based on a complex prompt. However, Midjourney's video capabilities held up well. Cling did a good job with the lights on the suit. Yet, there was less overall movement. More comparisons with other tools like V03 and Pika are likely in the future.

Final Thoughts: Did Midjourney Video Deliver?

Midjourney has entered the AI video space, building on its strong image generation. Its first video feature shows potential, especially with handling tricky details like eating, and offers good control over motion. While there are still growth areas, like specific scene changes from prompts, it works well.

Do you want to get the most out of your Midjourney experience? Consider using the Midjourney Automation Suite from TitanXT to improve your creative workflow.