# Massive Updates to Generative Video AI

## Метаданные

- **Канал:** Skill Leap AI
- **YouTube:** https://www.youtube.com/watch?v=zif3b5l6ptw
- **Дата:** 29.11.2023
- **Длительность:** 7:56
- **Просмотры:** 19,302
- **Источник:** https://ekstraktznaniy.ru/video/12867

## Описание

This week, the top 3 AI company working on Generative Video release some major updates.

First on the list was Stability AI, the company behind Stable Diffusion.
Stable Diffusion is the most popular open source AI tech for text to image generation and it powers some of the top AI apps like Leonardo AI and Dream Studio.

First update is from Stability AI.  They have introduced the first open generative AI video model, which is based on the popular Stable Diffusion model.

Right now, you can join the wait list for an upcoming app. And if you are more technical, you can access this initial model through Github.

They also release a new image generation model inside of their App Clipdrop you can try now.  It’s called SDXL Turbo and it can literally generate images in real time.

The second company with a big leap in AI video was Runway. Runway is still the leading text to video platform with Runway Gen-2. They released a new update called Magic Brush.
With Magic brush, you can brush over a

## Транскрипт

### Segment 1 (00:00 - 05:00) []

the top three companies that are working on generative video AI all release some major updates and it's a huge leap for AI video generation so let me go through the list with you I'll show you some examples I'll show you the ones that are available to use right now but first on the list is from stability AI now stability AI is the company behind stable diffusion so stable diffusion is the most popular open- Source AI technology that generates images from text and it Powers some of the top AI apps for text to image like Leonardo AI dream Studio clip drop those are all powered by stable diffusion and this is a big deal because this technology is open source is from stable diffusion stability AI so now a lot of people could create apps with the power of generative video AI so Leonardo for example is one of the biggest companies out there right now is getting 15 million views a month on their website it last time I checked and that is powered by the latest version of stable diffusion models so now imagine what's going to happen in the whole world of video AI now that stability AI has introduced stable video and right now if you go to stability ai. com you could join the app wait list so if you click here it's just going to give you this form here to fill out so fill out this form pretty straightforward and for more technical people they do actually have this for testing right now and you could download the weights again far more too complex for what we're going to do here I'm going to wait for the app to make an updated video but if you come to this hugging face page it does have a link to GitHub and then you could download it here as well and here it also shows you that release so it says we're releasing stable video diffusion and image to video model for research purposes right now so this is some examples they have on their page now before we get to the second one let me also show you stability AI also released something called turbo so this is sdxl Turbo this is stable diffusion turbo let me show you it's available to test out you could do this right now so there's a website called clipd drop. this clip drop is one of the biggest apps in the world of text to image and AI image editing but look at this option right now this is almost real time text to image generation and a lot of this technology obviously is built into the video too let me show you what it does I'm going to type in a cat and let's see just in one second he made a cat running I'm going to type in running and look at that in almost real time in water okay one more second in water with a dog let's see if it adds a dog look at that now it's not quite good right some of those were not as good as the first option I had but this is just incredible this is almost happening in real time I'm just blown away I just been testing it for only less than an hour and it's really incredible how fast it's working it's still not good though so some of my images are not as good as the first generation when I'm doing these live updates but give it a try this is inside of clipd drop. now the second big company that introduced something really interesting in the world of AI video is Runway so Runway is still the leading text to video platform they have a tool called Gen 2 that is absolutely mind-blowing still I think is the best in the game and Runway inside of Gen 2 has a new Option called motion brush that I wanted to show you in this video is available to use right now for anybody and again why this is a big deal is Runway right now is valued at $1. 5 billion they're not going anywhere so as soon as they perfect this is gonna make a huge change in how we generate and even create videos moving on in the next couple of years so inside of Runway if you go to this image tab you need to upload a file so I created some images from M Journey I'm going to upload but you could also generate from text but this motion brush basically brings Steel images to life and that's what I want to show you here so when you upload an image right here motion brush click on that and I'm going to click anywhere to edit so you get this brush icon here and you could brush part of your image that you want to bring to life so let's say this cape right here I want to give that Cape some motion and maybe I want to give this flag some motion so that's all I'm going to paint just like that and let me just make sure I do a good job here and then on the bottom you have some settings and you could go ahead and change basically the motion so this is going to move horizontally left to right so you could change the speed basically these are all different speed settings and then I'm going to go ahead and press save here and generate from here again this generates pretty

### Segment 2 (05:00 - 07:00) [5:00]

quickly this generates 4 second videos but this motion brush now gives you so much control it's not just giving it an image or giving it some text prompt and creating a motion video with it this lets you change just part of an image and turn it into motion and let me go ahead and show you this in full screen and let me press play look at that now it looks like I accidentally brushed part of the face here which caused this kind of issue but you get the idea here it's not quite there yet so you're going to have to try it multiple times but how cool is it just to move part of an image and bring that to life and probably the most exciting update in this world is from P Labs so you could access Pika at p. art and right now I just joined the weit list but they rolled out something called Pika 1. 0 and it's going to take place right on this website once you get in but if you look at their demo reel just from everything else I've seen in the world of generative video AI this is absolutely stunning and the fact that you could actually do things like you could do with photoshop with an image right inside of Pika basically live edits something that's part of a video and here's an announcement we're excited to introduce P 1. 0 this is inside of Discord where Pika always was so it was available for free it still is available for free so if you want to use it you could jump in and basically follow them on Discord join their Discord here and then you could go ahead and generate images so they have these things called generate one two three and then you could see what people are generating here and you could generate your own but what's going to happen now with this new paa if I go to their announcements they are going to roll this out outside of Discord as well so the web version is going to have the first version of Pika 1. 0 I've been using it for about a month now inside of Discord and it's pretty good especially with bringing images to life it wasn't great at turning text into video but if I turn images into video it did a really great job and now with P 1. 1 this is going to be outside of Discord on a website which a lot of people like a lot more it seems they don't have to come into Discord to use this and it's going to be a much better model so this could be a complete Game Changer as you see on the demo reel the results are really stunning now the only other one missing from the party is kyber AI That's one of my favorite text to video generation platforms but as you could see these three updates from these three other leading AI platforms that are working on video this is going to be a complete Game Changer my backgrounds in film making so I'm really excited to see where this could take my creativity and then also use it for any type of commercial work and promotional work as well once it gets a little bit better now once I get into some of these betas I'll make a more deeper dive video with some tutorials I hope you found this useful and I'll see you next time