No video
AI Video News! Gen-2 Text to Video Generator AI is PUBLIC!
Рет қаралды 34,218
Күн бұрынRunwayML Gen 2 became available to the public today after being in beta testing on Discord since February. It is the first publicly available text-to-video model on the market and it has thousands of users.
▼ Link(s) From Today’s Video:
✩ Gen 2: www.futurepedi...
✩ Yesterday's video: • The First High Res FRE...
► MattVidPro Discord: / discord
► Follow Me on Twitter: / mattvidpro
-------------------------------------------------
▼ Extra Links of Interest:
✩ AI LINKS MASTER LIST: www.futurepedi...
✩
✩ General AI Playlist: • General MattVidPro AI ...
✩ AI I use to edit videos: www.descript.c...
✩ Second Channel: / @matt_pie
-------------------------------------------------
Thanks for watching Matt Video Productions! I make all sorts of videos here on KZfaq! Technology, Tutorials, and Reviews! Enjoy Your stay here, and subscribe!
All Suggestions, Thoughts And Comments Are Greatly Appreciated… Because I Actually Read Them.
-------------------------------------------------
► Business Contact: MattVidProSecond@gmail.com
@Pepius_Julius_Magnus_Maximu... Жыл бұрын
LMAO They made it public literally the day an open-source alternative was released, THEY ARE AFRAID
@MattVidPro Жыл бұрын
Makes you wonder...
@spinninglink Жыл бұрын
I literally thought the EXACT same thing when I saw it. Support open source Y'all!!!
@Gerlaffy Жыл бұрын
AI will be the first field to effectively democratise and make technology free and open to all
@mikepriceakasmalls Жыл бұрын
What is the open source version?
@bethaapple2102 Жыл бұрын
But still charge $15 for 2+min per month? Like what? More like nah. Skip…
@deciphrai Жыл бұрын
Timestamps courtesy of Deciphr AI 👨💻 0:01:13 - RunwayML's Gen 2 is accessible to everyone 0:01:55 - User interface and free trial 0:02:17 - Pricing for Gen 2 0:04:31 - Duration of video and upscaling 0:04:53 - Initial images and inspirations 0:05:18 - Testing out the grass prompt 0:06:44 - Impressive results with Gen 2 technology 0:11:01 - Review of Gen 2’s image-to-image and image-to-video features 0:11:31 - Discussion of open source generator Potat1 0:12:31 - Gen 2's attempt at generating a KZfaq outro
@malfunkt Жыл бұрын
It's really cool! Though only 125 secs for 12USD/mo. It would be great if we could get multiple smaller previews of the output and only get charged for the full-render.
@armondtanz Жыл бұрын
Absolute killer :(
@davidpaxton6402 Жыл бұрын
I think your calculations are a bit off. 1 second uses 14 credits. That's only 49 seconds of video. So on that plan 1 minute of rendered video will cost over US$20!!!! $30+ if you're an Aussie!!!!!!! 😮😢😢😢
@malfunkt Жыл бұрын
@@davidpaxton6402 Just going off the plans on ML's site, with standard 1-year sub at $12/mo you get 625 credits which equals to 125s of Gen 2 rendering.
@armondtanz Жыл бұрын
@@davidpaxton6402 Are u sure? it just says 205 secs in top corner - max = 4 seconds, therefore 4 seconds is removed and you now have 201 secs... Ive just got $15 pm plan
@davidpaxton6402 Жыл бұрын
@@malfunkt l recheck that. I was going off the plan when only Gen-1 was available. Maybe they altered the plan. Nonetheless it's still too expensive. AI video and image generators require a lot of test runs to get satisfactory output. I'd rather prefer a $15-20 per month sub while it's in its infancy. It's definitely not good for commercial use.... yet.
@MrJaggy123 Жыл бұрын
MattVidPro : "This is so horrifying!!!!" Also MattVidPro: *plays it again and again and again* ;)
@AmazingArends Жыл бұрын
Why do you think people go to horror movies?
@dwsel Жыл бұрын
I had a hard time recognizing these shapes, so I actually laughed the whole time 😂
@legacylee Жыл бұрын
Being one of those people who had beta access to GEN 2, this is bittersweet. I went from having unlimited generations to limited, but it's still my favorite AI tool to date. Using multiple tools to dial in the look and feel of the videos has opened a world of creative generative content. Been playing with Wonder Studio, then using Midjourney images and Gen 1 and 2, a little Photoshop, and other tools has been a fun workflow!
@edwardbradshaw6850 Жыл бұрын
I’ll wait until it gets a bit better to jump in.
@vp62ift Жыл бұрын
Where you able to get more then 4 second duration?
@legacylee Жыл бұрын
@@vp62ift no unfortunately the longer clips are only for Gen 1 right now, I already spoke with them
@roblikestotalk4984 Жыл бұрын
It’s not called bleeding edge for no reason lol. Thanks for staying up to date as much as HUMANLY possible
@nahiddotai Жыл бұрын
My initial thoughts are these short videos are great for intros and outros of movies, music videos, short films and that type of creative content. It's no where near as good for creating actual movies or well put together content, but AI moves fast! I'd love to see how it evolves and what Gen 3 and 4 from RunwayML will be able to do. For now, I wouldn't be paying this - impressive nonetheless
@yellowcottagetales Жыл бұрын
This kid will have a great broadcasting voice as it deepens as he ages. Good now, but will be pro quality for sure.
@MattVidPro Жыл бұрын
Thanks! I'm 23 right now, but man do I wish my voice was deeper. It hasn't got deeper since I was about 14 lol.
@yellowcottagetales Жыл бұрын
@@MattVidPro It will get deeper.
@guepardiez Жыл бұрын
It's never too late to become a chain smoker.
@giri.goyo_yt Жыл бұрын
Gen-2 is going public! *I hear the sound of many credit cards going into debt...*
@iamYork_ Жыл бұрын
When it goes open source and can be run locally is when the boundaries will get pushed beyond a toy...
@IKP1000 Жыл бұрын
Great update video, thanks. I got the alert message last night and immediately rushed to try it. First impressions: using an image as part of the prompt was not a success for me at all, but just using simple text prompts, such as "fire-breathing terrifying aliens flying over an alien landscape" produced some great results, and "locking" the seed really helped with consistency over several creations with similar prompts.
@carlosalbarran8322 Жыл бұрын
Making the model opensource was probably the reason Runway made it public as well. I dont think it is a coincidence.
@GorilieVR Жыл бұрын
Woah i typed the prompt: "space flower blossoming with galaxies and vivid color" and result is crazy cool 😊🌌
@ixiTimmyixi Жыл бұрын
Thanks for the shoutout Matt. I can't wait to see what you make next!
@giri.goyo_yt Жыл бұрын
I'm a child. I did think "Your export will be saved to Matthew's Assets" was hilarious. That confessed - great announcement and demo, Matthew!
@nomadvizion9424 Жыл бұрын
By the way, Runway already has a text to audio generator called Music LM.
@dimitryprokof4847 Жыл бұрын
Thanks for the video! I think we live in a society so excited to get into such quick creativity that we forget how still beta and early stage this Gen2 is. It truly looks plain useless with its results but we all get mesmerized by writing something and having a few frames moving, looking forward what will happen soon with this technology!
@phizc Жыл бұрын
Honestly, if I were paying $0.12 per second for any of those videos, except the first one (wind in field), I'd not be happy. Good thing they have the free tier. Obviously it can produce some good results, but finding them would get expensive quick.
@iamoneabe Жыл бұрын
Thumbnail update time ;) "GEN-2 is PUBLICY AVAILABLE" And a good honestcomparison video between GEN-2 and Kaiber would be great as I'm not looking to them both..
@Dron008 Жыл бұрын
Too creepy. I tried to generate a cute kitten chasing a big dog but couldn't even see a cute kitten, just some mix of cat and dog. And no chasing at all.
@coloryvr Жыл бұрын
At the current level this is nothing I woud pay for....BTW: The horrifying part was very big fun to watch.
@MichaelLaFrance1 Жыл бұрын
"I think I'm gonna get sick," is the best reaction to AI video I've ever seen.
@Bamazon1990 Жыл бұрын
Ai is improving by leaps and bounds but I think we are still a long way off from getting the creepy out of it, as in when Ai actually knows wtf a natural human looks like drinking coffee etc
@timsnodgrass4637 Жыл бұрын
Interesting start and an amazing leap forward in technology but based on the pricing I saw and these clips it seems like it needs to grow a bit more before it's ready for prime time. That said I look forward to what I'm sure it will become in the near future.
@FusionDeveloper Жыл бұрын
The eye one you did, turns my stomach for some reason. I had to look away the last few loops.
@MrPablosek Жыл бұрын
Eventually open source will surpass the closed one, as it has done in fields of image generation and music generation. And in the field of text generation, it's getting very close to chatgpt, main limiting factor is the hardware to run it.
@TheFitchFiles Жыл бұрын
8:47 that's definitely a new Tool music video lol
@RMCanimationOFFICAL Жыл бұрын
Finally! I’ve been so long to public like 3 months!
@21EC Жыл бұрын
Thanks so much for letting us know..I had no idea it already was out !
@TaskSwitcherify Жыл бұрын
You know the hallucinations of ChatGPT? Can you imagine how glitchy this is going to be for the first year?
@omar92 Жыл бұрын
cant wait for opensource
@JEAGERlST Жыл бұрын
This stuff terrifies me. You'll no longer be able to prove you've actually recorded something in person.
@armondtanz Жыл бұрын
Matt on his eyeball creation "ewww worst thing ever" Me: take my money , where i sign up!!! Ps. I thought their best feature was vid to vid? Or was that sum1 else?
@MattVidPro Жыл бұрын
It is, but text to video is more... traditional? thanks to DALL E, Midjourney, ect.
@RufusJacson Жыл бұрын
Don't want to dis this too much, as it is amazing that it can generate anything at all, but I've just tried 4 different very simple prompts and it created videos that weren't even close for 2 of them and then created what we're effectively still images for the other 2... It's exciting but WAY TO EARLY to suggest they're delivering on their claims!
@nuclearthrone2497 Жыл бұрын
I had to laugh so hard on that eye creature video 🤣
@VaibhavShewale Жыл бұрын
so we get only 85 seconds and no extra free credits every week or month? sed
@eranfeit Жыл бұрын
You said in the video that you can run it locally. Do you have more information about it ? Thanks Eran
@superjaykramer Жыл бұрын
Matt you are in the uncanny valley! buddy. Really enjoy watching your vids, you're the go to guy for AI news.
@youMEtubeUK Жыл бұрын
Ultimately it's going to be cheaper to go open source and just spend the money for a beefy GPU and run Nvidia omniverse, generative AI images, davinci resolve machine learning for your creative needs. Midjourney about £100 a year and this what £120 and others will all add up...
@davidpaxton6402 Жыл бұрын
Yes agree. RunwayML is way too expensive. Da Vinci can also do this post production edits for free and it's cinema movie quality. The market will go elsewhere.
@arielcurra7647 Жыл бұрын
I tried generating a bear eating spaghetti,it mixed the bear with the spaghetti,it was scary although kind of cool
@metakron Жыл бұрын
Oh look at that. They made it public just when an alternative was developed. Who knows why? 😊
@SMmania123 Жыл бұрын
So looks like monopolies fold under pressure lol. So its free to use now?
@studiotime6671 Жыл бұрын
Just saw the email in my inbox and immediately thought about your last video lol
@erikburzinski8248 Жыл бұрын
RunwayML's Gen 2 failed hard on my test which it to attempt to get it to do a proper body transformation from an adult into a child it didn't even try. (the test is meant to show flaws in its human generation and I can get it to work on stable diffusion however there aren't enough charters to use the same prompt here.
@CrystalBreakfast Жыл бұрын
They think this is ready? Like ready enough to charge people to use it? Nah, man. This is not ready. Needs a lot more training, making it usable for free would help them gather more training data from public use.
@21EC Жыл бұрын
yes :) omg I thought it's gonna take a while before we get this released for everyone ! :) im so happy ! holy shi% btw i just created a cute creature from another world and it looks so amazing !
@ubhelbr Жыл бұрын
10:38 BE NOT AFRAID
@KirmiziKara1 Жыл бұрын
So next year WE will.see.first.holllywoood Blickbuster Film with AI
@TheChromePoet Жыл бұрын
I'll give you about 72 hours before this video becomes obsolete.
@21EC Жыл бұрын
YESSS !! omg I thought it's gonna take a while before we get this released for everyone ! :) YESYSYEYSEYYSEYES Im os happy ! HOLY SHI% BTW I just created a cute creature from another world and it looks so amazing !
@cesarsantos854 Жыл бұрын
Tried it. First one was good. Second was pure horror. The third so so.
@21EC Жыл бұрын
BTW honestly Gen1 was meh..I wasn't impressed by it but Gen2....damn..just wow! now we're talking, the future is here and now we gonna see all kinds of incredible AI videos being generated by people, exciting times !
@21EC Жыл бұрын
Bummer..this costs money after like 5 generated vids, is there any way to buy just one month of use only ? (for $12)
@reezlaw Жыл бұрын
I feel like the "close up" part is ruining your prompts
@nangld Жыл бұрын
That software really need a way to redraw the messed up parts. I.e. select these additional eyes and just "fill these with yellow texture"
@AmazingArends Жыл бұрын
Yeah, saying it's "pretty good" is not good enough. Video is a lot harder to fix than a still image.
@Alexafloofaiart Жыл бұрын
is this going to be like Dalle you know how you can use it in different ai websites?
@metakron Жыл бұрын
I can't wait for gen-2 to become an IOS and android app
@AlexCrews Жыл бұрын
You do have an app for iOS I tried using the app it won’t let me sign in
@Ben_D. Жыл бұрын
ya... When my feed tries to show me a vid about "crazy new tech revolutionizes everything" and I see that the vid is two weeks old, I just give it a pass. Two weeks is literally a generation these days.
@editedbymaths Жыл бұрын
Thats the reason - They dropped it.
@peacefulman2196 Жыл бұрын
Realistic video at 1080p require a lot of rendering time, about 1-5 minutes per frame on powerful hardware. Ai wont be able do realistic videos at high resolution for now. Theres no magic in this world. Probably can do some combination from existing simple videos, like you can do it in adobe premiere. Another problem: AI cannot track 3D obcects and do not have feeling for depth.
@ibrahimayad6732 Жыл бұрын
Sorry, man. But after seeing the result of the galloping horse, I closed the video. but I will never use GEN2 in the mean time untill it is perfect
@heredownunder Жыл бұрын
I can’t watch any AI KZfaq videos less than 2 weeks old
@hoangnam6275 Жыл бұрын
Same feeling when seeing that vid
@FusionDeveloper Жыл бұрын
Awesome, but I'm not paying. I will just use whatever I can for free because i don't need to do anything specific, which is, just for fun. Really awesome outputs though!
@Joe_Joe_Ultra Жыл бұрын
3:30 seconds of introductory content. Recommend condense of that mess to like...19 seconds would've been nice
@Nobmaser Жыл бұрын
I got it for only $1 with 300 credits
@metakron Жыл бұрын
HentaiAI subreddit: its free real estate
@Maisonier Жыл бұрын
What about anime?
@Yozo_official Жыл бұрын
10h 644 likes 197k subs
@mikepriceakasmalls Жыл бұрын
I like your channel buddy
@MattVidPro Жыл бұрын
I appreciate that
@mikepriceakasmalls Жыл бұрын
@@MattVidPro I’m really getting into this AI stuff. Keep it up
@anguswilliam2141 Жыл бұрын
Should probably get AI to do your hair.
@MattVidPro Жыл бұрын
true
@Someone7R7 Жыл бұрын
Potat 1 is better
@spinninglink Жыл бұрын
4 second videos eh. And the prompts i were trying weren't even giving me close to what I wanted. 👎
@morpheus2573 Жыл бұрын
Runway ML should not be charging for the privilege of making deformed monstrosities.
@etuzex6166 Жыл бұрын
I would say they're pretty crapy images and people are jumping on it like on NFT's back in days. Paying money for lagging pictures😅 what a fraud.😂
@aziza7107 Жыл бұрын
Matt, just let it go. Get a buzzcut or shave it all off. You know what I'm talking about.
@brockhampton3078 Жыл бұрын
nah mind your own business
@Drugs543 Жыл бұрын
No way!
@foxdenfinds Жыл бұрын
it will only get better and soon we don't need youtubers like you as we just prompt you with text :P :P :P lol JK
@AM-pq1rq Жыл бұрын
awesome!
@MattVidPro Жыл бұрын
Thank you! Cheers!
@egretfx Жыл бұрын
your head when you are showing the platform isn't even needed to be there, that's just narcissism no offense tho:)
@Sjlasse Жыл бұрын
finally
@contentfreeGPT5-py6uv Жыл бұрын
Yoo i play with this Is awsome
@phanhung3232 Жыл бұрын
can i use it free forever sir ?
@jopansmark Жыл бұрын
Not so impressive after Modelscope and Potat1
@lRlMlGl Жыл бұрын
Well jesus
@Brazil-loves-you Жыл бұрын
It's paid = I don't care about it
@kurpop Жыл бұрын
yooo
@madcatlady Жыл бұрын
I have no free points left wish didn't have to subscribe for a year
@MaxChromaColor Жыл бұрын
You can choose monthly, then it is 15 per month however, and you get certain number of credits, then you can pay $10 to buy more credits.
@madcatlady Жыл бұрын
@@MaxChromaColor but can you only do a month
@MaxChromaColor Жыл бұрын
@@madcatlady yep you just have to select monthly payments instead of yearly. They say its $12 but that is only if you pay a full year its a discount... the monthly subscription is $15 if you choose to pay per month.
@madcatlady Жыл бұрын
@@MaxChromaColor cool
@erik4444q Жыл бұрын
I tried it. It is absolute trash. Pretty much a combination of all the dumb things that don't work with image generation. We keep doing this AI hype that will not deliver the promise that you're promoting.
@newplayer1313 Жыл бұрын
Let's see how your comment ages
@brekol9545 Жыл бұрын
lets goooooo
@BeginningInfluence55 Жыл бұрын
Really finetuning would be the key. Stable diffusion 1.5 generated very unpleasant, cheap images on its own (but can generalize very well) but you can achieve an arbitrary quality by finetuning it on high quality images. The custom checkpoints are all tied to a specific use case, but in this use case they are magnitudes better than the base model. I imagine same if you could finetune these text2vid models.
@PigOnPCIn4K Жыл бұрын
Hey I was thinking exactly what you were too, I was so frustrated after seeing good results from it and I've been using it successfully with some extra models free! I'm posting a short with what I made tomorrow
@phen-themoogle7651 Жыл бұрын
SPAROTRCDRO1
Harley Quinn with the Joker
Рет қаралды 25 МЛН
Theoretically Media
Рет қаралды 115 М.
adumb
Рет қаралды 388 М.
AI Video School
Рет қаралды 24 М.