Getting Started with Gen-3 Alpha | Runway Academy

Learn how to use Text to Video with Runway’s newest video model, Gen-3 Alpha. Try it now at runwayml.com
academy.runwayml.com

[@High-Tech-Geek]

At least Luma gives us 30 generations free per month. You guys are charging $0.25 per SECOND and you all know how many times you have to reroll a clip to get something close to useful. Throw us a bone.

[@oddfellow8366]

Seriously!! 😂 $1 per generation, I use maybe 1 or 3 out of 10 (if I'm lucky).

[@wholeness]

Another token grab by Runway at $.50 a generation for terrible stock footage😂!

[@southbridgeforestHOA]

what is a good alternative? I agree, the footage does not match prompts and seems like stock footage.

[@Artof3drendering]

img to video needed

[@zenmasterjay1]

Your subscription system is lame. For starters..... Now At least let people use gen-2 for free.

[@leonardoskoromnyy8839]

It generates garbage only! It must be free during Alfa! I want my money back!

[@doctorstrangebird1]

we need image to video also

[@OliSavageBeatz]

Yea

[@nartz5361]

dont worry, at this rate, i guess this year will release

[@azhar-1999]

Yeah that’ll be cool. Also the next cool thing will be video to video. Imagine that!

[@OliSavageBeatz]

@@azhar-1999 yea would get Crazy

[@doctorstrangebird1]

@@azhar-1999 oh yees :)

[@Repossessionn]

What a joke of a pricing

[@Repossessionn]

From my testing it is crazy how cherry picked these results are. I can barely get 1 perfect.

[@MR_AK558]

Is not free I have no money 🤡

[@lewpsyt]

Sora is free?

[@High-Tech-Geek]

@@lewpsyt Luma gives you 30 free generations every month. The Runway pricing is about 60 times pricier than Luma. Insanely expensive. Luma is about $0.25 per minute. Runway is $0.25 per second. Greedy.

[@MR_AK558]

@@lewpsyt sora is available to public??🤡

[@СвободнаяПалестина]

No​@@MR_AK558

[@Kuefler]

@@MR_AK558no

[@USM3C]

I just subscribed, really disappointed there's no character reference/upload available for Gen3 Alpha 😢

[@danielkahbe964]

Props for finally releasing it! Finally, some progress instead of shiny trailers.

[@Dragons_Armory]

⚠⚠⚠ WARNING, Gen 3 has almost NO functionality. There is no image to video what so ever and just a pathetic 1 text prompt window. Yes there are advanced sliders like Gen 2 But there is NO preview and no image to video option at all. What's more 9/10 videos you want to generate are very unstable and prone to mistakes (characters models glitch, anatomy all over the place, lighting and even BG became inconsistent.) currently it's so bad it's even worse than a scam. Worst 15 dollar commitment I ever got myself to. Don't give them a penny until they give you something worth your money.

[@southbridgeforestHOA]

what is the alternative right now? or do we just wait for sora from open ai?

[@Nursultan-r2d2]

still trash one, i can literally recognize ai or not

[@bartul_st]

gen3 without option to use image to video is total disappointment

[@Kaotonix]

as a producer and artist i've been using gen3 for real world artistic purposes. My most recent single's music video was created with gen3 and for the first time ever it's actually worth it. I spent about $30 in credits. I'll be making more music videos with it if anyone wants to follow along

[@JohanJeornburg]

Interesting, how do you feel about the AI music generators like Suno AI/etc?

[@Kaotonix]

@@JohanJeornburg good question, I think it’s getting really good but only in specific situations just like with video. It’s great for creating instrumentals, sounds for sampling/sound packs, but I’ve been making music for more than a decade and working on my craft and it just cannot compare currently. I see this changing very soon, but in the current state it is still a tool vs generating a really good polished song

[@magmazyprime8572]

Will this be available for free later?

[@rocus80m]

It already is but only for 3 vids and only in the base model

[@OliSavageBeatz]

Pleas Gen 3 Image to video🙏 Gen 3 is Amazing but Expensiv i think. But 70$ for Unlimited yea is much money but Worth it

[@MemePopperz]

Please make it free to try out for us normal users...i am really impressed but very disappointed that it is not free

[@DasHnezzEdits]

my guys, are you really asking 15$ for 6 gens?

[@TheTinker394]

At double the price of Gen2 We need preview for Gen3 asap

[@ashendoski]

Don't waste money on this; it's not worth it.

[@fabriciofrecia5949]

if anyone who's reading this knows how to write a prompt to turn lights off inside a room, it would really help me out. I have tried dimming, turning off, turning out, stopping, but the room remains lit, thank yoouuuu

[@rocus80m]

ask chatgpt buddy ;) try ask to give you some proper prompts, but give it a role first.

[@Sumkilla]

@@rocus80m also maybe micro train chatgpt on the gen3 prompt guide, which i'm about to try

[@manishchetry2116]

It's already paid so no

[@OliSavageBeatz]

Keep it up. In 1 2 Years this will geht so many Feature i hope and gen 4 ect. Looks Great. Hiw mutch Credits for 1 Video ect with Gen 3 ore gen 2. And pleas Image to video and this things

[@yewtwobur]

This technology is definitely not ready yet to warrant the costs being asked for individuals trying to learn. Simple prompts that work for one image to video creation do not work on the next one (getting the camera to stay completely still/no movement at all) Artifacts appear out of nowhere and fly through the creation. The learning process is confusing to someone like myself; beginner). I am very tempted to just pay the money for the unlimited plan but even that is confusing….it says unlimited generations in explorer mode? So that doesn’t mean unlimited generations because you still only get the same amount of credits ? 2250? I have used many of my credits by mistake also. 3 simultaneous generations of the exact same thing were made because I could not see that the video generation was started- I was in a different section than the one I was used to. Pretty easy to add a failsafe I would think. Part of this is me learning as I go and making mistakes which is normal but it seems like for a standard plan you may get 1 or 2 usable creations. 10 second clips? I think for a paid plan and especially higher tier plans you should be able to get longer clips. How about a lower quality preview of the image to video creation before you are charged? It is exciting though and I will stick with it as it evolves. I think within about 6-12 months most of these things should and will be resolved.

[@wholeness]

When Portrait Mode for Gen3?

[@AlejandraAlvarez-u2w]

I have some clips that were shot vertically. I want to keep my original video intact but add sides using generative fill to convert it into landscape. I tried using Gen-3 Alpha to fill the sides, and even uploaded images done in Photoshop to guide the generation, but the results were terrible. Am I using the wrong tool? Any guidance will be appreciated.

[@tommel302]

Image to video please

[@thepresence2095]

image to video please

[@saurabhbhardwaj6753]

BEware: if you are on a free account- use the exact prompt as shown in this marketing video and see the output. It will be no where close to what they claim here. I had run out of credits and subscribed to it and it was pathetic and stays pathetic. Nomatter how much detail you provide it, it will not give you what you want in a video unless u just want some panning on a simple image. In simple terms, it is still worthless. I am cancelling my subscription.

[@playexploreryt]

TF I CAN'T EVEN TRY FOR A MIN

[@lewpsyt]

Is not free 😢

[@AIcinematica]

It begins 🎬🤖

[@shadrackkiban4716]

How is this even possible 🤯

[@saurabhbhardwaj6753]

I have tried the exact prompts as shown here, my output is pathetic, this is just marketing!

[@GlitchtubeMeme-mu7uq]

It’s not free

[@JarosawTabor]

Diagram workflow to video animation. That'd be awesome.

[@denizcarta6805]

Dear Runway Team, you have plans to add "Normal Map" to your Edit List. Because its the future of Greenscreen Lighting. An other website called Beeble AI Relight will realese this feature soon. I really hope Runway will add this feature too

[@bbrother92]

Normal Map" and aces and render layers

[@artjar4138]

This is made of random images. How long and complicated must the prompt be for the AI ​​to do it according to my vision 100% in every detail.

[@TrunxKraft3000]

🤦🏿‍♂️

[@galefraney]

Is Runway unable to create text motion graphics? I asked for a simple prompt: "still static camera captures text that is dripping down from the bottom of the letters on a green screen background, text says "Rain Rain Go Away", solid green screen background" ... Runway seems to begin the render, I can see the progress bar going up to about 97%, then the render for this video goes back to 2% and seems to continue rendering, then at the end I get an error message. I've tried many different ways of writing the prompt, including not putting the text in quotes, etc. What's wrong with my prompt or is RunwayML gen 3 unable to render text ... I notice that "turbo" is not available for text prompt, only image to video ...

[@saurabhbhardwaj6753]

nomatter what ur output will never be accurate. I purchased it and now cancelling the subscription. at the end I used the exact same prompts as they show in the video here and u will never get an output close to what they show here. It is a scam!

[@Heisenburg17]

holy shit, gen 4 and rip videographers

[@masaodbalataee2422]

Why doesn't anyone talk about the video resolution? Is it 4K or what? Isn't the most important thing here is accuracy?

[@modgrip805]

It's fucking terrible.

[@NeuralNote-wp7xx]

Been trying this out... so far it's OK. Maybe I need to improve the prompting, but it's not doing a very good job with fantasy type of text to video prompts.

[@rocus80m]

Any examples? I would love to see the real capabilities

[@NeuralNote-wp7xx]

@@rocus80m Not sure if it's allowed to post a link to the video, but if you click my username (Neural Note) you can find my channel. The video going up at 12:00am PST tonight ('Dominion of Mortals') is my first use of GEN-3. I used it in various parts of the video... but really, if it wasn't for the fact that I spent the credits on them I would have not used them. The video quality is 'ok', but nothing close to the prompting I was entering. I followed the Runway guides, but could not get good quality shots for my subject matter. Here are the GEN-3 timestamps: [03:24.25 - 03:42.18, 04:23.22 (FPV DRONE) - 04:59.16]. For comparison, LUNA footage can be found at [02:38.01 - 02:47.16, 03:42.18 - 03:52.20]. In fairness, LUNA was from an Image, but I had pretty good results from text prompts as well in terms of accuracy.

[@rocus80m]

@@NeuralNote-wp7xx thanks brother

[@cucciolo182]

15 bucks for 15 videos, but only one of them was good. I also spent more time modifying my prompts due to the “ restrictions."

[@The-Spondy-School]

Is this even available NOW or just more... coming soon chatter?

[@denaco]

but is 1080p not available? i find 720p is too low res and looks bad

[@BicDin]

Things will probably get better. It's just a matter of time before it supports 1080p and allows you to choose any size of the image without having to crop it and maybe even add a blend of multiple images to create a single video.

[@southbridgeforestHOA]

I am trying out the Gen2 and the accuracy of the prompts sucks. I say 5 happy neighbors and it shows 8. I say 2 people join a group of 3 and it just shows 3 people. Is gen 3 any better?

[@FABIOGATI]

Considering the high price and without image to video, it doesn't seem very convenient, especially if most of the videos it generates are incorrect.

[@e.v.k.3632]

With 10 seconds it's gonna take a long time to make a movie

[@mad-cyantist3159]

10 sec for now.... the tech just got out of nursery.

[@ThanosDestroyeryearsago]

Don’t make a whole film with this tech yet. It’s not at that point.

[@FilmSpook]

@@ThanosDestroyeryearsago A whole *short film,* yes. 👍🏾👍🏾

[@AI-Art-Explorer]

i like how use friendly it is compared to gen 2, also image to video would be nice to.

[@AndrewShaw-q9e]

far too expensive

[@OnkelRobinjo]

The world is actually sleeping on this, it's insane. Let them sleep is what I say.

[@RdAttri108]

Is that dangerous for vfx artists?

[@ahtashambinasif3577]

We need Img to Video Please..

[@UnchartedWorlds]

Ok OpenAi Sora can close shop now, these guys are beating them 😁

[@silly.catmemeoftheday]

Realistic but this isn’t sorw quality w

[@blurtam188]

Can you get the video to loop?

[@NEWMOVIESAI]

Great!

[@Singularityseries]

Let’s go 🎉🎉🎉

[@tacticar]

The result is unbelievable.

[@wholeness]

You mean the cost 😂

[@wagnersouza9680]

Acabaram de sepultar o Sora

[@najlepsa]

So much better than luma.

[@gu9838]

yaaaaaaaaaaaaaaaaaaaay thanks for releasing it today i REALLY needed it. i love you guys!

[@tubestreamkyki]

You generate 2000 videos with GEN3, there can be 1 useful. What a great video generator!!!

[@southbridgeforestHOA]

that's what I feared. the prompts are useless. I tell it 5 neighbors and it gives me 8 people in the video. am I doing something wrong, so is it just not good? Seems like it's not generating any video, rather just pulling up random stock footage