AI-generated video is getting better and better as professionals like my friend Ali are actively turning their prompts into actual films. Now, however, there’s a new player in the Gen AI video scene and it’s pretty impressive. It’s called the Dream Machine and it comes to us from Luma Labs.
Best of all, unlike the hyped-up tools from the big AI players (looking at you, Sora), you can actually use it. Anyone can sign into Dream Machine and start generating short video clips from text prompts — today.
How good is it? Check out some samples below, but before you do that, here’s a quick reminder about our next AI Fundamentals class, happening next week:
We've still got a few spots left in our June class open to everyone, and this time it’ll be laser-focused on tools and techniques for PR work. It’s happening June 20 at 1 p.m. Eastern. We’ve curated a custom toolset of apps specifically crafted to speed up PR writing, pitching, and creating on-brand media (think: articles, images and videos).
If you’re new to AI, or if you wish you could do more with it for media/PR/marketing work, this class will take you to the next level. This PR class is one time only, so don’t wait! Click the button below to sign up:
I mean who hasn’t wanted to generate a photorealistic depiction of cigar smoking dogs riding a motorcycle.
Or three women dressed as cats eating cupcakes?
What’s most interesting about these videos is the quality. While you’re still seeing some terrible artifacts — check out the hands of the lady on the right — the overall quality is more realistic than anything we’ve seen in a while.
What I think is happening here is the slow ascent out of the Uncanny Valley for video. Almost all images in the GenAI stable have a dreamy, unreal quality to them right now, be they stock photos of people drinking coffee or an illustration of a robot doing his homework.
Further, the leaps that we’re seeing in improved video quality are extremely impressive and, well, visible. Whereas we can’t quite feel whether on not a text interaction has improved, we can definitely see a difference in this video compared to earlier AI video generators.
Unfortunately, Luma just won’t listen when you ask it not to do things that reduce hallucinations. I tried the previous prompt again with one caveat: “three women dressed like cats eating cupcakes under a colorful umbrella. do not show their hands at all.”
I got this:
I guess it tried its best.
Regardless, you can try the Dream Machine now for free. It takes about a minute to generate videos and, barring a little weirdness, it shows us that this stuff is getting better and better.