#7 The future of video is spatial and has Midjourney become too good?
The final substack of the year is here with number 7 of the bi-weekly innovation Substack. This week we’ll take a dive into the update of Midjourney, the future of video capture is spatial and the making of the McDo lipsync. Let’s go!
Midjourney v6 is live and it’s a blast!
We’ve written about past versions of Midjourney in our Substack recently. After the first version of MJ (V1, released in july 2022), the AI has been leaping forward with every new version update. The latest version (v6) has a greatly improved generation toolset, we’ll highlight a couple of major improvements:
Text generation: in previous versions it wasn’t particularly possible to generate text in an image. At least it wasn’t readable text but all scrambled letters. With the new version of MJ it is sort of possible to generate readable text. But it’s still quite hard to get decent results. After playing around with it, I've got the example below. It seems to work better when there are known examples of text in images that it was trained on. So for this example I tried making a book with the text “TV Guide”. It worked, but it isn’t perfect yet.
Heightened realism: the realism and resolution of the images is on another level. This new version has such a great realism when prompted right, that in my opinion it by far outperforms its competitors like Dall-e3. In the sample below we’ve created images with the same prompt but in all versions of MJ. The gain in details and overall quality is staggering.
Other samples via: https://twitter.com/i/status/1739406119725060484
Is MJ6 too good?
Some people have reported that the update of Midjourney has become too good in generating artworks that resemble the original artwork. It started copying too much. In the sample below someone used a prompt to generate a copy of the Mona Lisa. The results are extremely similar to the original painting.
The future of home video is around the corner with spatial video
A few months ago Apple announced to launch a new type of VR system: the Apple Vision Pro. With these new type of augmented reality style goggles, a new way of video playback called spatial video has emerged.
Spatial video is a type of video in mixed reality that is captured with camera apps that have a 3d mode like lidar. With apps like Wist by Wistlabs it’s getting easier to record such videos.
It’s still pretty rudimental, but the potential is there. Just a matter of time for this to become better.
Check out this video by Ben Geskin with a cool test.
The Making of the McDo LipSync Experience
We’ve had a great partnership with McDonald's Belgium over the past years. After a bunch of cool campaigns in 2023 we closed down the year with a captivating end-of-year campaign. The centerpiece of this campaign is a Christmas-themed digital overworld integrated into the existing McDonald's Belgium app.
This immersive environment, designed as an interactive advent calendar, offers users a series of unique experiences, each unlocking on a specific date. Among these experiences, the McDo LipSync stands out, showcasing the capabilities of AI technology in creating engaging user experiences.
Read the whole story here: https://www.livewall.co/blogs/the-making-of-the-mcdo-lipsync-experience-en