Update: Gen-3 is now publicly available. I mentioned a couple weeks... It was only a day! :) Runway released a prompting guide to help you get better outputs here: help.runwayml.com/hc/en-us/articles/30586818553107-Gen-3-Alpha-Prompting-Guide

By the time Sora finally becomes available we’ll probably already have an open source version or free one that’s just as good

8:21 He was reloading the baguette with cheese.

While it's really impressive how far we've come with AI video, I'd really like to see how specific we can get with it. Your prompts were kinda generic. What happens if I want a character to wear very specific clothing? Colours, material, fit? Can it do a coat with mother-of-pearl buttons, or do we get any old buttons? Image Gen looks impressive as long as you do generic portraits of people, but breaks if you do anything complicated and out of the ordinary. Try prompting for a handstand or someone dangling from a tree branch heads down. To be useful for serious production work and not being just a toy, you need to be able to have a granular level of control. I know I'm asking for a lot, since we're only at the beginning stages yet, but it'd be nice if you really stress-tested it, so we know where we're standing at the moment. Still, great work, Matt! 🙂

I think it will be better when the image to video option becomes available.

Low quality clipart video for people who really don't need to have anything particularly good. It'll do because it doesn't really matter because no ones pay much attention to it anyway. This will be great for those adverts at the side of blog posts that everyones trying hard not to accidentally click on so they only focus on the thing they googled in the first place. Marketeers will love it. As long as it moves, it'll do.

The band playing music beneath the ocean was pretty cool! Even the floating microphone looked kinda real lol!

comparing this to year-ago generations, it's a big improvement. Next year vid generations should be even better

These funky artefacts heavily remind me of image gen around DALL-E 2, Midjourney 4 and SD 1.5. As we've seen those ironed out and improved in later versions, now we're seeing better hand generation and legible text in the newer versions of those image gens - can we expect similar, along with improvements in temporal consistency, in these video gen models? Pretty excited for the next year, if so. That baguette video blew me away a little bit.

next time you see a demo, expect that they generated 100s of videos to make 1 video demo!

That’s cool, but I wish we had more control over the videos created. The videos are starting to look better, but the control is still lacking, which makes these videos not that useful for filmmaking projects. Also, we need *‘Image to Video’* if we want this tool to work with consistent characters, and to create scenes for films. I want to be the director of the film, rather than letting the AI have all of the control over how a scene looks.

11:02 Imagine if the man & woman hugging here were ‘consistent characters’. I hope it will become easier to create scenes with *‘multiple consistent characters’.*

17:58 the girl on the water is practically perfect, very difficult to see any problems. Impressive!

Huh, another AI company showing off their product and preventing the general public from trying it out.

I would like a video on why the AI videos do what they do and what the blockers are in making them more realistic - I mean can the output be put back through an AI that corrects all the issues?

Gen 3 said ‘not available in your country’. LTX was available today and it was like an interface for the usual crappy vid generators. A waste of time and early signup attention. Great vid as always 👌

The color palette it picks is pretty good. Also I like the contrast. I'd say in about another year we'll be looking at some pretty fantastic looking AI videos from all companies. Thank you Matt for putting in the time to show us all!

Matt, have them export a depth map video (as an option - more points) with the video, which could be used to composite in some additional stuff - like some animated 3d objects that you add in.

I feel like alot of the clunk could be resolved in the future with like a critique AI pass of sorts that can critique object perminance and physical interactions

So many of these videos look like they’re beyond CGI, they’re really captivating!

9:06 He just turned clipping off on his hand for a second 😆

Just tried this and it's amazing for music videos. When making a real film you're also creating a lot of takes to cherry pick the best one, so I'm okay rerolling until it looks ok

Thanks for your video on Udio. I am addicted to it now. The “copyrighted lyrics detected” is pretty annoying. Considering we can easily get cover song licensing and both the AI prompter and rights holder could be making money, it’s kind of dumb to restrict usage of other peoples music.

I just started using Runway Gen-3 a week ago and so far I've made a zombie video and now a dancing cats video. I think both turned out great. Yes image to video will make Gen-3 easier to use but if you know enough about cinematography and camera shots, and you're not too picky about the details, you can do pretty well with just prompts alone in Gen-3.

Runways costs 35$ per less the 8 minutes a month. And this is for Gen 2. The next plan is 95$ euro per month to have unlimited generations. This is a professional platform, not for end users.

The quarterback at 12:30 was sacked so hard he's looking out the earhole of his helmet as he fumbles the ball.

Now. What is the compute power used for these Gen 3 vídeos vs the compute power used for the Sora demo videos?

Though the generations are hit or miss, this seems really excellent for visually storyboarding your actual production.

bruh when kling finally becomes available to everyone, it'll wipe the floor with every other video gen out there

Imagine how amazing the next iteration of Walking with Dinosaurs will look with this tech.

Simply imagine the fun video editors are gonna have once a solid AI video generator arrives. Works will be done in a matter of secs. 🤩

🎶 Runway train never coming back. Wrong way on a one way track 🎶

I assume OpenAI is allowing others to release theirs before they release Sora so they don't catch flak for being the first. Then once the public is accustomed to decent video generation, they release Sora which will be probably better since they have more compute, and receive little to no backlash. At least that's my guess...

In other words, there is a long marathon left before this is really good... 😮‍💨

Wow, this sure shows how much further they need to go to get a full 10 seconds that's usable. But it does look like there's a few seconds in most clips that are pretty good.

And THIS is just an Alpha-Release. ALPHA!! And look how well it is starting off at that Level. We'll be able to get improvements in the days and weeks to come. And this remember is with all those Creator Media 'Restrictions' they impose on AI Scraping as well.

I animate and do lip sync from MidJourney images, so I wish I knew how this was for that. I don't need any 1997 screen savers LOL. So far it doesn't look as nice as Luma but I'm only a third of the way into your video and I will finish watching now

tbf you did request the rapper to be "signing into his microphone" lol

I really hope makes one for us who are VJs for concerts/festivals so we can have generative content on the fly

Call of Bread Restaurant Zone - Loaf and Load

All I want is to be able to tell Netflix “Generate season 9 of game of thrones” or “generate a thriller about…”

As a working video editor / Multimedia designer Ive been using many different video ai tools over the last year. This looks a little better than the current crop but Honestly Video ai is not there yet.. not for professional use anyway. And im yet to be convinced that Sora have not heavily manipulated their sample videos in post to please investors. I hope im wrong.

hi there, how many credits does it cost to run one of these generations on 3gen? Thanks

Tried Sora/Keling/Gen3/Pika, have to say those things are not very differentiated, at least from users' perspective Text-to-video is still relatively barrier for most of mobile users.(And that's why we are founded, making productivity equality by video-to-video method)

Just got an email giving access so I’ll get it a test drive(crash?) and post on X because that’s what people do. Waiting for Pika to make a move as well? Exciting times .

On that text test, I think you need to give it a more simple prompt that's a single word or two, and keep the background simple.

Marvel is watching runway. The lawyers are properly getting their paperwork ready.

looks like he was loading the baguette

that baguette ops looking good

Do you know if it has live audio reactive generative visuals? if not. do you know any A.I software that has?

its an alpha model they can tune it a bit like they showed in preview but the industry isn't ready for it

Thanks, Matt. I'm glad I subscribed to 'Stripcue To Coeplee Chickk Thenn'.

Can Runway Gen 3 take in starter images like Luma?

I'm convinced they are all Comfy UI workflows lol

Seems like video game footages were used to train this model, look how good it did with all the video game prompts

2:24 What is the name of song and where I can find it?

Could you please share just a few of your videos so that it would be possible to download them and analyse how they handle or manage the image to image (25 or 30 frames/second), just to see?

How long of a clip can you create with the Pro subscription?

As of this date, should one invest time in Luma or Runway. Which one is the best end result for the money/monthly credits?

Give it 6 months and hands will be solved just like images, give it a year and cohesion will be pretty much complete and length will be extended from seconds into minutes.

This helped me a lot! I've been waiting for a minimum quality level and I think we're just there by our fingernails. I was going to wait a bit because I couldn't decide on the tools and procedures. I think the best control and quality for me will be midjourney/lumalabs/elevenlabs/synclabs. If anybody has alternate suggestions I'd love to hear them. I'd wait for runway to have image to video but there's always a reason to wait and no guarantee of how it will compare to Luma. Time to make some cool stuff!

However, for some use cases definitely useable! 🎉🎉

Nice to see someone make the greatest AI videos since I arleady make the Best AI songs ever made ^^

Does anyone know how to Generate a video (or another generator) that can produce videos with no camera movement? I've set the camera movement to the lowest setting along with prompting "Tri-pod shot", "no camera movement" etc but nothing works and I need still camera video generations to edit in pre-existing characters filmed with a tri-pod.

very good video and I had similar experience from my first tests: you definitely need 95$ unlimited plan to use it because >90% of results are garbage. Also no image input. I would not recommend at the moment for normal usage. But if you want to experiment a lot and learn which things are working it can generate some interesting and sometimes just amazing output.

Hi Matt, would you mind saying what kind of computer system you are using to generate your videos. Just wondering how powerful a system you need to keep the generation time to sane levels. Thanks.

14:42 he’s an AI millionaire 🪙

It's unclear how many Gen-3 videos we can generate under the various paid plans.

what makes someone qualify as a creator to get early access? I thought you have standard subscription you got early access? I have standard but I don’t see gen 3 option, I am not on desktop right now though only phone

The bowling video is so uncanny that it made my legs go cold and filled me with a deep sense of dread 😨 Soon we'll have the name of a new phobia for these kinds of things 😂

18:17 omg that is Twiggy.

14:02 Guys, don't forget to _Stipcile to Cplehlee Chickk Then_

Is it possible to choose the aspect ratio? Like 9:16 instead of 16:9

Don't forget to Chickle Then! 😂

When do we get to see you play the banjo, Matt?

Thanks for this. I've been using Pika but this looks like it will be useful as well!

is there an ai that can help me generate some basic clip art style for a billboard advertisement?

Maybe they won't release it until the music industry's lawsuit against the ai music makers is over? These companies gotta be scared that they will get really sued for their copywrited fair use training.

Your timer during the generation makes it look like it took an hour and 47 minutes

It looks promosing - for me the most useful workflow is when I generate imagies and than animate them - I hope they will add this feature soon.

Matt - what is that matrix code "lamp" in your bg? I'd love to get one 🥺

i want to see you do an AI video for typography if it exists.

Needs image to video (I find that gives better results on most AI Video generators). Result's look ok and have come a long way in short space of time but still a long way to go imo. Any idea on cost of this?

Can it do higher resolution than 720 p?

More interested in the image to video.

I created 3 beautiful MTV for Kitaro without even using Sora or Gen 3, check them out. 😍😍😍

I have a feeling Amazon and metis is gonna eat up all these AI companies

Crazy fire 🔥

mreflow still freaks me out every time lol

Very interesting so far. AI video and games really tickle my fancy.

The real life results were mostly pretty bad... i estimate that only next year we will get text to video models that will really create good results rhat can be used to create high quality short movies f.ex.

Sure text to video has come a long way but what's the use case? Entertainment?

what about image to video? did you try that?

Awesome content thank you im only worry about the fact that we will have to generate each video multiple times i dont want to burn my monthly limits in just few days

its ALWAYS looks good when the platforms partners post some demo, until you check it out by yourself and the results is sucks hahaha

Copyright is automatic. The Berne Convention

Have to say, GEN-3 has been relatively disappointing considering the price point. I found it hard to get good results of 10s when generating fantasy art and animation. It's fair, but not as great as I was hoping for based on the demos.

The craze of sora is now over

Is dancing another passion of Matt besides A.I? In the other video u used people dancing in Magnific Relight lol

What's gotta be scary is....anyone that has ever taken psychedelics ....anyway. This all looks familiar when it glitches.