Google just announced their new AI video generator during the Google IO event and it looks amazing it's called vo and the best part is unlike Sora you can actually go sign up for the wait list right now so let's go over these demos and how they compare to Sora and the other leading video generators they are also testing out other features like storyboarding and the website gives a taste of how that will look I'll get to that later first here's a quick clip from their keynote introducing it I really like one of the quotes Donald Glover has in here I won't play the full clip just the more important parts the core technology is Google deep mind's generative video model that has been trained to convert input text into output video it looks good we are able to bring ideas to life that were otherwise not possible we can visualize things on a time scale that's 10 or 100 times faster than before but that's what's cool about it is like you can make a mistake faster that's all you really want at the end of the day at least in art it's just to make mistakes fast so using Gemini's multi modal capabilities to optimize the model training process vo is able to better capture the Nuance from promps so this includes cinematic techniques and visual effects giving you total creative control everybody's going to become a director and everybody should be a director cuz at the heart of all of this is just storytelling the closer we are to being able to tell each other our stories the more we'll understand each other now let's take a closer look at these demos we'll start with this one it is really impressive all kind of talk over it as it's going since there's no sound but this is a full one minute long generation so the way it starts out is a really solid flyover shot of this kind of neon City these buildings look really consistent then it speeds up and this car comes in there is definitely some morphing and fuzziness here but the consistency of the car is really great and the driving physics are solid the part that's coming up is the most impressive to me but already the fact that it can go through these different scenes while maintaining consistency is far beyond other video models it is definitely a little blurry and not high definition the scene is changing really fast like the car is speeding through it and then this tunnel looks cool now right here when it comes out of the tunnel we see all these other cars with great consistency and high detail along with all the buildings in the background this is overall incredible especially considering how this scene started out and what it went through to get here right now this is the prompt they Ed so you can see it was pretty basic but they prompted each part of the scene that it went through and it nailed it and they have to reiterate that this video hasn't been modified especially given Google's history we don't know how cherry-picked this is you know there were probably a lot of generations to get this output but no matter how many times you ran a prompt like this through other video models it would be impossible to ever get anything close to this it is still a level below Sora but it's a big step up from the models we actually have access to and from everything I've read it seems like Sora will be very expensive if we ever even get access to it so this will be the best video model we can use which I'm really excited about let's check these others out this jellyfish has really solid physics and detail although they wrote deep ocean in the prompt this looks right up near the surface this time lapse of a water lily opening is pretty perfect other than the fact it cuts a little short but it looks great being able to generate all these videos is great but there's usually more that needs to be done from there to get them production ready and share them with others wondershare uniconverter has an entire of tools to help they've been one of the leaders in this area for 17 years and these tools apply to any video but I'll focus on how it relates to AI video since that's the topic of this one with a lot of actually honestly all AI video generators the videos that come back need to be enhanced or touched up so wondershare has two models to help with that the AI power D noiser can reduce noise and motion artifacts while improving the video resolution and Clarity then the frame interpolation technology can increase the frame rate without losing quality so it adds more fluidity so that's needed for pretty much all AI videos using the current technology if you're going to be posting them anywhere at least you can also add or remove watermarks from videos can efficiently compress or batch compress audio and video files without losing quality across essentially any format I'm going fast through all of this because there's so many Tools in here but those are some of the main ones that are most needed when working with AI videos in particular now this is a really helpful tool for anyone creating with AI image and video tools use the link in the description to go try out and you can also see all the other features I wasn't able to get to and thank you to wondershare for sponsoring this video this horse in the sunset is amazing all four legs are walking really accurately plus the head and the tail movement looks really authentic although there is still the tendency of generating things in slow motion which happens with all video generators now this is a really good shot of this spaceship again it kind of Misses part of the prompt like stars streaking past it so that's an indication of where this model is with prompt adherence but it's a really good output either way and I know I'm finding a way to criticize every one of these they are incredible but that's just the state of all AI video right now it's never perfect you need to acknowledge where the limitations are to be able to find the best ways to use them all right this Kebab on a grill is solid you know the flames and smoke look natural although you can get something this quality from other video models out there this panning shot in a mountain landscape is great again this could be done with other generators this golden retriever on the other hand this is an amazing shot the tail wagging looks really natural and the consistency in the scene is really good like how you can see through these leaves a little bit on the side right at the beginning the scene later and looks consistent with what it showed before that's something that's often an issue with other tools now here's one of a person that's where things always get the most difficult this is the only shot of a person they have in any of their demos and it has very little movement it followed the prompt really well and her face remains consistent throughout the shot but there's some morphing on her hand when she moves I'm guessing There's issues when you generate people that involve movement that's usually one of the biggest struggles for video models and I think they probably would have showcased more of that if it was good at it the only other video with a person in it is on their website and smoke is covering the whole face or this one of a person walking from behind that was in the keynote video so not really pushing it to see where the limits are with that there are a couple other videos on the site to check out this balloon person dancing looks really good same with this turtle underwater the light on its shell looks pretty realistic there's a POV of a mountain biker cruising down a canyon that looks great then this crochet elephant is really impressive that's one of my favorites out of these to sign up for the weit list go to this website I'll link to it in the description then click join our weit list and it's a short form to fill out I am really excited to get my hands on this interestingly they only showcased examples that focused on realism there's no cartoon or like 3D or abstract Styles so I really want to experiment and see what it can do there the only exception is on this post where they have a short demo of the storyboarding feature they add clation in the prompt it gives a thumbnail then they add a new prompt for the next scene and then they generate a song to go along with it that's really cool they're going to have their image video and music Models All in This one platform which is awesome this clip doesn't show what the result of these shots is though but I like that they're adding the storyboarding here I want tools that have more creative control and work towards storytelling instead of just making some cool visuals Runway ml is really good with this same with LTX Studio you know this seems basic compared to those platforms as far as control and storytelling but the quality of output from the model is much better and notably there's no mention of image to video at all so it definitely seems like it will only be text to video which is a downside and to reiterate from these demos it looks like by far the best text to video model will actually have access to but we don't know how that will work as far as generation times and what you're able to generate they only showed that one longer video that was a full minute every other example was 8 seconds or less so my guess is that's what we'll be able to generate I'd assume those minute long Generations take a lot of time and compute just a hunch but I doubt they'll give the full ability to generate those really long videos out of the gate overall this is a huge step forward and will open up all sorts of possibilities for creating AI films I am really excited to play around with this to stay up to date with all these AI advancements make sure to check out futur pedia.com in AI Innovations there's of course all sorts of other features on the site if you watch this channel you've probably seen them before but you can find the best AI tool for every use case and build out a profile where you save your favorites get custom recommendations every week there's also a whole curated database of the best AI tutorials on specific use cases and our newsletter where you can get tools tips and tutorials delivered straight to your inbox every week thank you so much for watching I'll see you in the next one