Midjourney Video is Live! Experience New Creative AI Tools

The world of AI video is changing fast. In just the past week, we saw a big shift, including new ads going viral, Midjourney releasing its video model, and two more top models from China. While our focus usually stays on other AI news, so much happened in video that we had to test it. Most new releases this week were about video. This update tells you what changed and shows examples, so you can quickly see how powerful these tools are now. They really do change what you might think about AI video.

Midjourney's Video Model Arrives

Midjourney's long-awaited video model is here, and it looks amazing. The artistic quality that Midjourney is known for now appears in video. This makes it stand out from other tools. Have you seen AI-generated videos like this before? Probably not. They offer unique and artsy results. This feature is part of the $10 per month plan.

How Midjourney Video Works

• You create an image in Midjourney as usual.

• You can now turn these images into videos.

• This works for images you already made.

• Right now, it is image-to-video only; text-to-video is not available yet. This makes sense because images were Midjourney's big strength.

We ran some test prompts through Midjourney's video generator to compare it with other models objectively.

Other Big Video AI Releases: Seed Dance and Hilo 2

Two other major AI video tools came out this week: Seed Dance 1.0 and Hilo 2. On text-to-video leaderboards, Seed Dance beats previous top models like V3. For image-to-video, Seed Dance 1.0 (from ByteDance, the company behind TikTok) is also number one, with Hilo 2 close behind. These tools can be expensive to use; running some tests cost us €15 very quickly.

Comparing Top Video Models

• [B]Midjourney Video vs. V3:[/B] Midjourney shows good anatomy for human figures and fantastic robots. It has a cinematic feel. For fast motion and complex scenes, like with a car example, Midjourney does well with reflections and wheel movement. Its strength is its artistic look, not necessarily realistic action. For detailed or hyperrealistic action scenes, you may need a different tool. Managing many attempts and keeping track of your best output can be manual. Consider trying TitanXT's Midjourney Automation Suite to streamline your process.

• [B]Hilo 2:[/B] This model struggles with human anatomy, often making arms look wrong. It often adds more camera motion. For cars, it performs well, with good reflections and dynamic angles, and realistic wheels. But for humans, it is not the best.

• [B]Seed Dance 1.0 Pro:[/B] This model is impressive, especially with human movement. In tests, a person walking on a beach could turn around and pose for the camera, looking natural. For car shots, reflections are very accurate, and it can even add details like burning tires. It ranks at the top for both image-to-video and text-to-video.

Text-to-Video Examples

Midjourney cannot do text-to-video yet, but Seed Dance and Hilo 2 can.

• [B]Seed Dance:[/B] For a "woman DJ spinning records on a rooftop in LA with a cheetah," the human anatomy is good. The scene looks solid, fitting for a music video. For a "sloth with pink sunglasses on a donut float in a pool," the sloth anatomy is among the best seen. It offers a very high visual quality.

• [B]Hilo 2:[/B] The "woman DJ" scene is more dynamic with a sweeping shot. The "sloth in the pool" is cool and fun, but does not look as good as the Seed Dance version. For detailed or hyperrealistic action scenes, you may need a different tool. Try exploring TitanXT's Midjourney Automation Suite to help organize all your generations and streamline your AI video creation process.

The two big winners are Midjourney for its unique artistic style and Seed Dance 1.0 Pro for its high quality in many cases, especially when compared to V3. V3 does include an audio component, which these other models do not. These new tools greatly expand what is possible with AI video.

New Tools for Images and Video Editing

Beyond video generation, there are new tools for image creation and editing. To handle all these new possibilities and create effectively, tools that automate your workflow are a big help. Check out TitanXT's Midjourney Automation Suite to improve your creative process.

Craya 1: A New Image Generator

Craya 1 is a new image generator. Its outputs do not look AI-generated. It often has an aesthetic film-like look, sometimes crisp, sometimes artistic, but different from Midjourney. It is a very artistic tool with a distinct style.

Hicksfield Canvas: Editing Images Without Changing Everything

Tools like GPT-4o for image editing are good, but they often change a person when you edit their clothing. Flux Context also changes the human, though sometimes subtly. Hicksfield Canvas offers a solution. It allows "inpainting," which means you can selectively edit elements of an image without completely changing other parts, like a person's face. If you want to edit an image and then turn it into a video, Hicksfield Canvas lets you do that with ease. This tool is excellent for creating ads that feature human figures.

AI in Action: A Viral Ad Example

An ad that went viral online was made entirely with V3. This ad shows that high quality is not always the most important thing for consumers. If content is interesting, small visual flaws might not matter. The ad's audio helped tell the story, and most people did not notice minor visual issues. This highlights a chance for creators interested in using AI tools. If you experiment and use tools like V3, you can tap into new market areas.

The creator of this viral ad shared his process. It took him two days to make. He had 300 to 400 generations but only used 15 clips. This shows that having a clear idea of what you want to create is important. You often need many generations to find the few perfect clips.

Quick AI Updates

Here are some other important AI updates:

• [B]TikTok Symphony:[/B] TikTok is adding new AI features regularly. TikTok Symphony turns images into videos, showing TikTok's support for AI-generated content on its platform.

• [B]AI Avatars in China:[/B] In China, AI-generated avatars are now used for live shopping, a popular sales system on social media. Some top live streamers are using avatars to work 24/7. One AI avatar stream got over 13 million views in six hours.

• [B]New Open-Source LLM:[/B] A new open-source reasoning large language model (LLM) from China is available. It is comparable to current models and has a context of 1 million tokens, which is useful for those who can use it.

This week brought many developments in AI images and videos. The speed of change is remarkable. We hope this information helps you understand what is possible with these new tools. Thank you for reading.