Explore Midjourney’s Latest V7 Changes & Try This Free AI Lip-Sync Tool

Let’s look at the newest things happening in Midjourney V7 and check out the Tavis Hummingbird lip-sync model.

Midjourney V7 Model Gets Better

When Midjourney V7 first came out, it had some problems. Remember the unusual poses and blurry background faces? Since then, there have been smaller updates like tiling and remixing features.

Now, Midjourney has pushed out what feels like a bigger update to the core V7 image model. Good news: those strange, rubbery characters are less common. The model is getting better at creating more realistic or intended figures.

However, the blurry background characters can still appear in scenes with many people. Details on main subjects, like clothing or equipment, seem to have more detail now. It’s a clear step forward for the V7 model, though there’s still fine-tuning happening.

Introducing the –exp Command

A new command has arrived: `–exp` or “experimental aesthetics.” This is similar to the `–stylize` command you might already use. You can even use them together.

What does `–exp` do? It aims to add dynamic elements and creative details while still following your prompt. Midjourney suggests trying values like 5, 10, 25, 50, and 100. Lower values (like 5-25) often show more noticeable effects, while higher values might have less impact.

Testing shows that `–exp` can add unexpected but interesting details to images. It can change the composition with different camera angles or elements you didn’t specifically ask for. It creates a kind of “weird that works” effect for many users. Values between 10 and 25 seem to be good starting points for many people.

Midjourney Editor Updates

The Midjourney editor has also seen some improvements. While many users prefer external programs like Photoshop, the editor’s retexture feature is worth exploring. It lets you quickly select parts of an image and then re-prompt them for a different look or texture.

You can also add and mask in other images on separate layers. The retexture tab then lets you generate new visual options based on your combined image.

Omni Reference is Coming Soon

The highly anticipated feature, Omni Reference (Cref), which lets you easily reference subjects across images, is still not fully released but is expected soon. Midjourney CEO David Holtz shared sneak peeks showing it can successfully maintain a consistent subject’s face while applying different styles and scenes. It looks promising for keeping character consistency.

Check Out a Free AI Lip-Sync Tool

Beyond Midjourney, there’s a notable new AI lip-sync model from Tavis called Hummingbird. This tool synchronizes someone’s mouth movements in a video to new audio.

The model works with existing video footage (it doesn’t work with still images). You provide a video and an audio file, and it generates a new video where the person or character appears to be speaking the new audio.

The results can be quite good, creating believable lip movements. Tavis offers a free tier on their website for 5 minutes of lip-sync video. For more extensive use, it can be more costly, so starting with the free option is a good idea.

Wrapping Up

Midjourney continues to release updates that refine the V7 model and add new creative tools like the `–exp` command. The soon-to-arrive Omni Reference also promises to be a significant feature for consistency.

Whether you’re trying out `–exp`, waiting for Omni Reference, or experimenting with new tools like Hummingbird, the field of AI creativity is moving fast. Stay curious and keep creating!