Placeholder Image

字幕列表 影片播放

  • as always I always forget that this Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop!

  • This'll stop!

  • Stop this!

  • Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, stop, stop, stop, Stop!

  • Never forget I'm gonna do this this dot This got to stop this stock song.

  • Never forget Binstock.

  • Somebody compose that song for me.

  • Hello and welcome to a very special summer edition of the coding train.

  • So I'm kind of on a little bit of a hiatus right now.

  • You might have heard me talk about where this very light content happening in July and August, coming back full steam, full force in September.

  • So I want to talk a little bit about schedule and stuff and what's happening on what you can expect.

  • But before I do any of that in order for you to fall along with the main content, that is the content of today's lifestream, which is a very special excited content.

  • Very excited about you do need to download some software.

  • So I'm just gonna tell you what to downloads to.

  • Megan said it going in the background while I babble, babble on and on about other stuff.

  • So I'm gonna open up a Web browser, something like this.

  • Go, go, go, go Ugly.

  • I don't know how to pronounce that chrome a comb.

  • A.

  • I think it's pronounced Web browser, and I'm going to ask you to navigate Thio if you so choose, um runway ml dot com So go to runway ml dot com Venue one who head over to this link under download beta.

  • And here it should start downloading automatically.

  • And there are versions for Mac OS Windows.

  • Lennox.

  • Anything else?

  • So whatever you're offered, by the way, I'm talking to somebody over here.

  • Normally, I'm just in this room talking to myself, but I am talking to Chris Valenzuela.

  • How'd I do on the pronunciation?

  • Pretty good.

  • One of the founders and creators of Runway, who is here in the chat Say hi to Chris in the chat to answer your questions and to control my questions as I fumbled my way through trying to learn about runway.

  • So I'm gonna come back to talk about what runway is and what I'm plan to do in today's live stream.

  • But I just wanted to at least set you on Thio, downloading this if you want to, and I'm going to give you a coupon code in a little bit to get some extra credits as well.

  • So but before I do that, So why you're downloading that?

  • What's happening here?

  • Where are you?

  • Where am I?

  • Where are we with what is that?

  • What's what is the meaning of the universe?

  • That is not a question I will answer.

  • So this is a little, uh, neighborhood YouTube channel called the Coding Train.

  • And every time I say coding train, I have to blow a train whistle.

  • It's in this contractor signed a while ago and, um, where I make computer programming tutorials and otherwise waste time reading things like random numbers.

  • And most of my activity happens during the school year because I also work at a school New York University teaching a program called T P.

  • And also that's a graduate program and undergraduate program called I Am A.

  • They're both part of School of the Arts, so summer is a little bit lighter for me in terms of content anyway.

  • But this summer's air particularly unique summer in that this room that I'm standing in right now is being closed down, and I'm moving the cameras and equipment to a new room on our in the Brooklyn campus sometime this month or next month.

  • So I'm gonna lose some time where I don't have a recording studio and we'll be setting up a new recording studio, which hopefully some better stuff and back full force in September.

  • However, there are many videos still outstanding that I am working on editing together and releasing number one is and this is very exciting a full documentation off video about all of the equipment and software in this room that I'm using.

  • So that's probably come out in the next couple weeks, which is good timing, stirring that I'm closing this down.

  • So this, uh, coding train studio which has been here since I don't know when I first started, I gotta look back and figure out what was the first live stream I did from this room.

  • The first live stream on the coating train, I think, was 2015 or perhaps earlier than that.

  • But I was I was doing everything from my office downstairs for a while anyway.

  • That's not super important.

  • Um, so that's coming out.

  • I have a two part series on how to program your own processing Java library, which I know everybody on the Internet is just waiting.

  • That's gonna blow up YouTube.

  • That's gonna be, like highest viewed video ever, where I show you how to build a job, a library in eclipse because that's like all the rage.

  • Now that's whatever is doing, right?

  • Right?

  • Okay, take a deep breath here for a second.

  • That's coming out of that.

  • The edited version of the tick tack toe coding challenge that I did on a previous live stream is coming, and also if I could get my act together because I forgot to record some stuff for that today, a edited version of the Euro evolution steering, car racing coding challenge things.

  • So that's also coming.

  • So there's a bunch of content that will be coming out in the channel.

  • But this might be the last live stream until September, or I might pop back for a surprise live stream like this one on some topic, like runway.

  • Okay, now I'm looking at the check.

  • Now let's say you happen to be supporting the coding train as a member on YouTube or a patron on patron, and I think I am going to putting this out there.

  • The university.

  • I'm not sure about it, but I might shut down the Patri on and move over entirely to YouTube memberships.

  • You hesitate to have everything all on one platform, but there are some advantages to that.

  • Stay tuned to information about that, but you might be wondering why I signed up for membership.

  • And suddenly now there's no content for two months.

  • Well, I suggest that you inside up, if that's really an issue.

  • But one of the things that I will be head down working on for the next six weeks is a dish second edition of the Nature of Code book.

  • And I've talked about that in previous live streams.

  • So I might do some member only live streams with work sessions on that book and share some on PdF.

  • So that's, uh, that's what I'll be doing in terms of the member community.

  • Um, in the next couple months, Um, I'm looking over at the chat, but I think that's the Mains sort of announcement.

  • E things that I wanted to talk about.

  • Nature of code books, summer schedule tell you to download runway.

  • I don't know anything else.

  • Take a momentary break here.

  • I know what it means to take a break.

  • I mean, I'm just gonna take a break from talking.

  • Anybody have any trouble downloading runway?

  • If so, let us know in the chat.

  • Now, I will tell you that coding trade is sponsored by water.

  • It's not exactly wet.

  • And there are different kinds of flow, like laminar flow and turbulent flow.

  • I think what I'm drinking, it's more of a turbulent flow.

  • Refreshing, actually, I make That's my joke.

  • It's very lame, but it is my joke.

  • There's no sponsor for today's livestream.

  • However, I do want to mention that I am actually an adviser to the company runway that makes Runway.

  • So this while this isn't exactly like a paid product endorsement, there is that relationship there, which I feel is important to disclose.

  • So this is something like a sponsored video in a way, by runway itself, and you can find out more about the company runway itself in the people behind runway.

  • And I know they're doing a lot of hiring, so maybe you want to apply for a job with runway all of the runway website.

  • Uh, see if there's any questions that anybody had any cloths.

  • Vogel Rights.

  • I have runway running.

  • Wexler 80 to thank you for your membership of the coding train.

  • Uh, you have a little comma waving emoji icon thing.

  • I forgot what that means.

  • A couple?

  • A couple.

  • Couple, Two months, six months.

  • One year.

  • Who in the chat has the longest?

  • I think if I go here to YouTube slash the coding train slash Join?

  • Um, it'll tell me.

  • Doesn't tell me anything.

  • No, me because I'm already a member because I'm not logged in.

  • Who knows?

  • But, look, I'm live comma waving.

  • Can those Me?

  • All right.

  • Um, so is okay.

  • Cook cookie crumb is asking.

  • Is runway okay for beginners?

  • Because I'm a beginner.

  • Yes.

  • I said that way too loudly when we said we're Conley.

  • Yes.

  • So I think there are on the runway is a really interesting thing, and I think that it would be best for me to start using it to explain it, but because I think there are aspects of it that the whole system, by its nature of the way that it's built as a piece of software is meant for beginners, but it is a tool that allows you to explore pretty deeply the world of machine learning models, some of which are quite sophisticated.

  • And you can wander down a lot of advanced it tricky and complex rabbit holes.

  • So I think there's a wide range here of kind of understanding and required not required, but that that is part of using runway.

  • But what I like to do in the coding train is have everything.

  • I'm making my assumption that you've never seen it before.

  • I would say the only pre.

  • They're really not any prerequisites for today, which is pretty rare.

  • But though eventually at some point I'm gonna move towards using runway with maybe processing and with p five, Jess, and then you might the pretty records.

  • It would have some familiarity with those platforms and basic coding knowledge.

  • But I'm here to help.

  • Um, I am here to help, and yes, this is appropriate for beginners.

  • All right, so I think of somebody gave me a super chat, which I have to acknowledge.

  • Charlie, England.

  • Thank you.

  • Um, thanks for that.

  • Let's see.

  • All right.

  • Um uh, any update on our mysterious bug that we're talking about earlier.

  • It's fixed.

  • Oh, exciting.

  • I was I was all exciting to debug way.

  • Have a new member.

  • Welcome to new member ball Simpson.

  • Thank you for joining the coating trade ring.

  • Make all the noise things like, you know, I should automate all this stuff.

  • Whatever.

  • Okay, so let's get Let's get going.

  • So what I'm going to do?

  • So I This is this is something slightly different from my usual life streams in the sense that this is entirely themed around Runway.

  • And ultimately it may not make sense or being necessary to edit this into smaller, shorter videos.

  • But just in case it does happen to make sense to do that because there's a lot of me babbling, and I'm sure I'll waste some time later chasing down some weird buggers piece of documentation or reading random numbers.

  • I'm going thio assume, actually that I'm gonna make two video tutorials extemporaneously improvising.

  • But one will just be an introduction to runway as a platform and signing up for a runway account, and then the other will be maybe doing a coding challenge where I try to get the results.

  • The output of the machine learning model into a P five jazz schedule.

  • I'll get to that later.

  • And really, let me just mention if you have questions about Runway asked him in the chat because you could have a whole completely separate conversation going on without me.

  • With Chris, one of the founders, creators of runway here.

  • And maybe they're also even some other people from runway watching high people for runway.

  • I'm probably gonna screw this up, So careful what you wish for.

  • All right, here we go.

  • Um, all right, let me just cover one other thing, because how around is asking Do you have to pay to keep using it So I will get into this As I look at Runway Runway is something that you can use.

  • You don't need to enter a credit card.

  • You can use it without paying any money.

  • However, there is some caveats to this.

  • The one of the things that I'll be doing is running the machine learning models with runway in the cloud, and that requires paying a fee by the minute or by the hour by the minute.

  • For cloud GPU usage every runway user when you sign up you automatically get $10 of free credits, and as I go through the sign up process, I'm gonna give you a coupon code.

  • It's just coating train, which will give you an additional $10.

  • You'll get $20 credits for free.

  • You actually do quite a bit with $20 credit to be surprised.

  • And then towards the end of the line street, maybe we'll talk about ways that you can actually run the machine learning models locally on your laptop if you're a laptop, has a GPU.

  • Obviously, there's something.

  • There's quite a bit of limitation to that, depending on what computer you're using.

  • But that would also allow you to run it without paying any money whatsoever.

  • I get that pretty much right.

  • All right.

  • OK, so, um, like you without the beard.

  • But I'm wearing the beard.

  • Where is that a thing you say wearing the beard.

  • That's weird.

  • I don't think you wear a beard.

  • You do whatever.

  • All right, So let's go Thio, the home page of Runway, and I am now going to start artificially, as if I'm starting over, introducing you to runway and then going through the steps to download and sign up for runway, and we'll see how that goes.

  • Uh, me, I think what I'll do, e think in the new studio.

  • By the way, I won't have this turning the cameras on and off thing which I actually quite sad about.

  • It brings me a lot of joy, Eli, right?

  • No, the beer wears you, which I think is a very profound statement.

  • Okay, by the way, I just have to highlight this.

  • Look, See, this is the home page of Runway ml consume in here.

  • We see this is some sort of computing machine running the runway software.

  • This is probably like the dense pose model, maybe, and then, but I just want to, like, highlight this.

  • Go over here into the corner here.

  • Look at this nature of code by some strange person in Daniel Sheaf mon.

  • Okay, there we go.

  • Nice to see that there's some other books.

  • There's a lot of life's little Easter eggs in here.

  • If you keep looking, here comes everybody.

  • This is a great book by Clay Shirky.

  • I think I might have read all of these books here.

  • We'll see.

  • All right.

  • Um great.

  • Alka reports that signing up in entering coding train code went smoothly, which is great.

  • All right.

  • I have no plan for this.

  • I mean, everybody is aware of that, but I just feel the need to say that one more time since Chris.

  • Hello and welcome Thio.

  • A tutorial Siris on the coding train about a piece of software called Runway.

  • So what I'm gonna do in this particular video is just show you the runway website and going to download the runway software.

  • I'm gonna sign up for a runway account and we kind of click through and show you the basics of how it works.

  • Now, what is even runway and why might use it before I even get to that, though I feel like it's important to say a couple things.

  • One is runway is not something that I've made.

  • Runway is made by a company also called runway.

  • They happen to be some of the founders of runway and creators of Runway are alumni of a program where I teach That makes sense.

  • I t p and I have an adviser to the company, but this is a tool that you can use starting today If you download it for free with some with a coupon from the coding train.

  • Let me tell you this one again one more time.

  • Just I won't keep doing this over and over again.

  • But I just needed a little, like, warm up.

  • That was a warm up.

  • I said too much.

  • I don't need to go on and on about this way too much.

  • Okay.

  • Hello, and welcome to a new tutorial.

  • Serious on the coating train using a piece of about a piece of software called Runway.

  • So what is runway?

  • How do you download and install runway and kind of tinker around it?

  • That's all I'm gonna do in this particular video.

  • Only clear runway is not something that I've made.

  • Runway is made by a company, a new company called Runway itself.

  • It's a piece of software you can use.

  • You can download it for free.

  • You can use it for free.

  • There are aspects of it that require cloud GPU credits, which I'll get into later.

  • And you can get free credits and a coupon code that you'll find a description of this video.

  • But really, I want to just about what is so excited about it.

  • I'm planning to use it in the future.

  • In a lot of future tutorials and coding challenges and teaching things that I'm going to dio on.

  • Guy also should just mention that I'm an adviser to the company runway itself.

  • So I'm involved in that capacity.

  • All right, So what is runway right here?

  • It says machine learning.

  • Four creators Bring the power of artificial intelligence to your creative project with an intuitive and simple visual interface.

  • Start exploring new ways of creating today.

  • So this this to me is like the core of runway I am.

  • Somebody was a creative quota.

  • I'm working with processing and P five jazz.

  • You might be working with other pieces of software.

  • That's just commercial software coding environments.

  • You writing your own software and you want to make use of recent advances in machine learning.

  • Read about this model.

  • You saw this YouTube video?

  • It's bottle.

  • Can you use in your thing well before runway, one of the things you might have done is find your way to some get hub Repo that had, like this very long Rimi of about all the different like dependencies you need to install and configure and then you've got it.

  • Download, visit, install this and then make build this library, and you can really get stuck there for a long time.

  • So Runway is an all in one piece of software with an interface that basically will run machine learning models for you.

  • Install and configure them without you having to do any other work.

  • But press a button called Install, and it gives you an interface to play with those models, experiment with those models and then broadcast the results of those models to some other piece of software.

  • And there's a variety of ways you could do that.

  • Broadcasting through http request see messages and all these things might not make sense to you, which is totally fine.

  • I am going to poke through them and show you how they work with an eye towards at least showing you howto pair runway with processing and how to power one way with P five Js and also show you where there's lots of other examples and things you can do with the other platforms and stuff like that.

  • All right, That was pretty good, huh?

  • All right, so I'm going to.

  • So the first step you should do is click here under down the fur.

  • So the first step you should do is click here under download Runway beta.

  • It will automatically trigger a download for Mac OS windows are Lennox.

  • I've actually already downloaded and installed runway, so I'm gonna kind of skipped that.

  • But skip that stuff and just actually now run the software.

  • So I'm gonna go over here, and I have, uh I have the runway icon already.

  • Here is part of my What is this thing called the Mac?

  • The doc knows what clique that run run away.

  • And now it's saying, Welcome to Runway.

  • Sign in to get started.

  • Okay?

  • So if you already have an account, you could just sign in with your count.

  • I do already have an account, but I'm gonna create a new one just so we can follow along with the process.

  • So I'm gonna go here, create an account.

  • I'm gonna enter my email address, which is Daniel at the coding train dot com.

  • Then I'm gonna make a user name and password.

  • His reign will be choo choo.

  • You think that's gonna be taken already?

  • My password or let's pause for a second generate a password not currently sponsoring today's live stream.

  • But, hey, Dash Lane, I think you can sign up.

  • It's actually dot com slash Go to trade or something.

  • They're responsible.

  • Previous life stream could find that their, um their password generator.

  • So we're in here.

  • Password changer, tools, password, history, password generator there, Otis.

  • Um, any requirements of the password I should know about?

  • Let's do like Oh, wait, this is not such a good idea.

  • All right, hold on, everybody.

  • That's it.

  • I'm going to generate the password that you can't see it.

  • A copy password.

  • I'm going to copy it.

  • I'm copping the password in and it also put it down here.

  • Please hold.

  • Okay, Um, and there, I'm gonna put you back here.

  • Let's go.

  • Now that I've put in my very strong password, I'm gonna click next, and I'm going to give my details.

  • Daniel Schiffman, The coating train.

  • Create account.

  • Uh, and it's giving me a verification code to Daniel at the coating train dot com.

  • Let me grab my mobile device.

  • That'll be the easiest way for me to read that email.

  • Oh, it's even got a notification.

  • Google thinks that An email from runway is a very important emails I haven't set up on Lee.

  • Give me an obligation for important emails.

  • Uh, I'm gonna say six.

  • I don't think there's any issue with me.

  • I guess maybe, though, I will shut this office.

  • Well, uh, just just for security sake, if you can, like, determine what keys I pressed based on audio signals than you're doing.

  • Very well.

  • Uh, validate.

  • Okay, um, put you back on.

  • Their account has now been creep.

  • Account has now been created, and I can click.

  • Start.

  • All right, So this is that I'm just gonna, like, minimize the browser so we can be clear on what we're actually seeing.

  • That's part of runway with this year.

  • Nobody in the password, I think.

  • Well, that's that We turn that off.

  • Was that the fast word that I just revealed by accident is going remarkably well.

  • That was totally with that.

  • My room.

  • My bathroom.

  • Take out now or I'm gonna change the password.

  • I'm sure I can do that.

  • Manage my account.

  • Um, And it profile.

  • Password change.

  • Password.

  • Current password.

  • New password.

  • Beat you all.

  • Hackers are everywhere.

  • Update great.

  • Changed my password.

  • So now my account is secured again.

  • Um, I'm logged in.

  • We go back to a runway and logged me out.

  • That was nice.

  • Sorry about that.

  • Everybody okay?

  • Close this.

  • They're gonna close the password generator.

  • Oh, looks like I got some sort of super chat, which I will now check on.

  • Thank you.

  • Oh, X rated, too.

  • That's very nice of you.

  • Just to say thank you.

  • Um, okay, here we go.

  • All right, Is everything you imagined the coating train would be just me trying to, like, not give people my password by accident.

  • It's basically what, 90% of the time it's all right.

  • All right.

  • So once you've downloaded installed runway and signed up for an account law, didn't your account, you will find this screen.

  • So if you've been using runway for awhile, you might then end up hear clicking on open workspaces.

  • Because workspaces are a way of collecting Ah, bunch of different models that you want to use for a particular project into a work space.

  • But we haven't done any of that.

  • So the first thing that I'm going to do is just click on browse models, and if you've never run runway and click on browse models.

  • This is a very exciting moment for you because there are a lot of models which you suddenly can play at your fingertips that you could just play around with.

  • And I haven't even planned this.

  • So I think we should just pick one somewhat randomly.

  • I'm going to use style Gan.

  • Later.

  • Anybody have one that they see that they want to request?

  • I would love to use dense post would be kind of fun.

  • Um, attention, Gan.

  • What's the one is the one where I could draw.

  • Oh, yeah.

  • Is that spade cocoa?

  • Yes.

  • So that I think that would be a good one.

  • Thank you.

  • Soon Dar 10.

  • All right, So the first thing that I might suggest that you do is just click on a model and see what you could do to play with it in the runway interface itself.

  • Because one thing that's really wonderful about runway is as a piece of software in their face, you could explore an experiment with the model to understand how it works, what it does.

  • Well, what it doesn't well do.

  • Well, what?

  • It does it all before starting to bring it into your own software, your own project.

  • So I'm gonna pick this spade Coco model, which I have never looked at before.

  • It very legitimate me.

  • I have no idea what's gonna happen to click on that.

  • And now here I can find out some more information about the model so I could find out.

  • What is the model do?

  • It generates realistic images from sketches and doodles.

  • I can find out more information about the model.

  • For example, this is the paper on that describes this model semantic image synthesis with specially adapted normalization.

  • Sze trained on coco stuff Data set.

  • Remember when someone asked, Is this big?

  • Is this is this to draw for beginners?

  • Well is for beginners in that you're a beginner.

  • You could come here and play around with it, but there's there's a lot of you can go very deep too, if you want, find the paper, read through the notes and understand more about what this model, how it was built, what date it was trained on, which is always a very important question to ask whenever you're using a machine learning model, so I can also we could see their attribution is here.

  • So this is the organization that train the model.

  • These are the authors of the paper we could see Went what?

  • The size of it when it was created, its CPU and GPU supported and then very important down here.

  • But I'm actually gonna highlighted here we can click on the license.

  • So this is creative Commons attribution noncommercial share alike four point license.

  • So I'm not gonna pretend to be a lawyer here.

  • I'd be able to give you proper advice on the beyond the scope of what I'm doing.

  • Here is the nuances of what all these licenses mean and how you can use them.

  • But you can find out and do that research yourself right here in the runway interface because to go under gallery and we can see just some images that have been created so we can get an idea.

  • This is something.

  • This is a model that's themed around something called image image segmentation.

  • So I have an image over here.

  • What does it mean to do image segmentation?

  • Well, this image is segmented, divided into a bunch of different segments.

  • Those segments are noted by color, so there's like a purple segment a pig segment on light green segment.

  • And those colors are tied to labels in the model essentially that know about a kind of thing that it could draw in that area so you could do image segmentation in two ways.

  • I could take an existing image like an image of me and try to say like, Oh, this segment it.

  • This is where my head is.

  • This is where my hand is.

  • This is where my hand is.

  • Or it could generate images by sort of drawing on a blanket into staying.

  • Put a hand over here, put a head over here and it looks like at least example we're seeing here is kind of like a, uh what do you call those things?

  • Art, Not a landscape.

  • It's a bowl of fruit that has a name.

  • How come I can't think of that word?

  • Old chats going to say it instantly.

  • Uh, art bowl of fruit.

  • I can't think of his word.

  • Um, I'm going crazy.

  • Still life?

  • Is that what I mean?

  • I think it's a still life.

  • It's not what I mean.

  • Maybe it's just still life.

  • I thought it was something else.

  • Whatever um, All right, I'll get edited out of local version of this video.

  • Okay, so that's what a big segmentation is, at least in the way that I understand it.

  • Um okay, so let me go here.

  • Now.

  • What I want to do is I want to add this sort of.

  • Now I want to use this.

  • So what have I done so far?

  • Thank you, James.

  • Craig.

  • And, uh, well, these super chest today, this is not usually happen.

  • YouTube, like promoting super chats.

  • Thank you, James.

  • Craig, I really appreciate it.

  • You that gets a train whistle.

  • Oh, okay.

  • What a time to be alive.

  • All right.

  • Where was I?

  • What have I done so far?

  • Downloaded runway.

  • I've poked around the models, and I just clicked on one.

  • Now I want to use that model.

  • I want to play with it.

  • I want to see it run.

  • So I'm gonna go here to add toe workspace.

  • It's right up here at the workspace now.

  • I don't have a workspace yet, so I need to make one, and I'm gonna call this workspace Red, actually.

  • Just want to leave It.

  • Has this name red current multiculturalism.

  • Several automatically name workspace for you and, uh, did it did it breaking news.

  • I'm being told that these workspaces air named from colors and art ISMs fruits, fruits, fruits and our ISMs.

  • So maybe you got blueberry, um, abstract Expressionism or something?

  • All right, so it, uh, created No, no, I'm gonna name it, though I must say coding train, live stream.

  • So I'm gonna do that when it create.

  • Now, I have a work space, and you could see this is my work space.

  • I have only one model added to this workspace over here, and it's kind of highlighting up for me right now.

  • What to do?

  • I need to choose an import source.

  • So every machine learning model is different.

  • Some of them expect text input.

  • Some of them expect image in.

  • Put some that might expect input.

  • That's from its arbitrary scientific data from a spreadsheet.

  • Then the model is going to take that input in, run it through the model and produce an output.

  • That output might be numbers or it also might be an image.

  • Or it might be more text.

  • So now we're in the sort of the space of a case by case basis.

  • But if I understand image segmentation correctly, I'm pretty sure the input is going to be both an image and the input and the output are both gonna be an image.

  • Right?

  • So let's make a little diagram here.

  • This white board has shut off, but let's make a little diagram.

  • So we have this, uh is this model called again?

  • Spade Spade coco.

  • So we have this machine learning model.

  • Presumably, there is some neural network architecture in here.

  • Maybe it has some convolution, all layers.

  • This is something we would want to read that paper to find out more run away.

  • It's gonna allow us to just use it out of the box.

  • And, you know, it certainly would always recommend reading more about it toe learn more about how to use it.

  • So my assumption here is I'm in my software that I want to build.

  • I want Thio, maybe create a drawing piece of software that allows the user to segment down an image.

  • So you could imagine maybe I'm gonna, like, kind of draw something.

  • That's one color.

  • Look, I could use different colored markers.

  • Look, I'm going to fill, you know, I'm gonna fill this image in with a bunch of different colors.

  • One of these markers don't work.

  • This is not important.

  • And then I'm going to feed that into the model, and out will come an image.

  • So we have input, and we have output.

  • And again, this is going to be different for every model that we might pick in runway.

  • Although there's a lot of conventions, a lot of the models expect images as input and output images.

  • Some of them expect Texas input and output, an image or images, input and output, text, et cetera.

  • And so on and so forth.

  • So here, under choose input source, I'm gonna collect segmentation.

  • Oh, wow.

  • Look away.

  • I'm not in the right thing.

  • Well, let me get back to where I was before.

  • Let me actually just remove it.

  • Well, actually, I click file.

  • This is fine.

  • Uh, I don't know why I'm obsessed.

  • I'm gonna try.

  • I'm gonna delete it just by the way, one thing you could do is you could delete a model, but I'm gonna delete it, and then I'm just gonna add it again.

  • Case you missed that part to get to see it again.

  • I'm gonna add toe workspace, Cucumber, multiculturalism.

  • No, I don't wanna knew.

  • I want to go to my work space.

  • Oh, Adam.

  • Model.

  • So what did I What?

  • Did I miss you?

  • Let's see, I'm discovering the runway interface.

  • I click here recently.

  • Used spade Coco, add to work space.

  • There we go.

  • 00 you know, it must have to save my workspace because there was nothing in it.

  • All right, no problem.

  • We're gonna do it again.

  • Read The nice thing about this is we get to have more automatic workspace names.

  • Let's see.

  • So I'm gonna say coding train live stream again.

  • I just wanted it to be I just wanted to be back to here in case we had it.

  • Now, I guess I can't not do it without thinking that we're gonna end it all right?

  • Now, what I want to do is choose the input source in runway for the model.

  • So something that's gonna produce a segmented image and so that could be coming from a file.

  • It could actually come from a network connection, which I'll get into maybe in a future video.

  • Or you could explore on your own.

  • I'm just gonna pick segmentation.

  • Now this is, like the greatest thing ever because what just happened is runway.

  • Your image segmentation is a common enough feature of machine learning models that Runway has built into it an entire drawing engine so that you can play around with image segmentation so you could see these are the colors for different labels.

  • So maybe what I want it looks like it's a lot of transportation stuff.

  • So maybe what I want is let's try.

  • Let's try drawing some people.

  • I'm gonna do that.

  • If we do it at a diversion, this will get sped up.

  • How am I doing?

  • It's drawing to people.

  • I don't think this is actually how people look in a really image.

  • We're gonna be interesting to see what it does and let's put on airplane flying above.

  • I wonder if I don't need to Do I need to be interesting thing about joining?

  • Does it matter if I do it in this shape of the airplane, or is it just gonna figure it out because it knows what the shape of an airplane is?

  • That's actually a much better airplane than I imagined myself drawing what else?

  • Um elephant.

  • Who?

  • Oh, boy.

  • There's a lot of other stuff.

  • It's not transportation, wine, glass.

  • It's gonna be a very large wine glass with two people with a airplane and a wine glass flying overhead.

  • Okay, are we doing now?

  • I'm gonna choose an output, and I just want to do preview, right, because preview right now, it's like I'm not actually I don't need to export this, and you need to use it somewhere else.

  • I just want to play around with it in runway itself.

  • So I'm gonna preview and, uh and here comes the most Not.

  • And now here's the thing.

  • Okay.

  • Now I have selected my input, which is just the segmentation interface of runway itself.

  • I have selected my output, which is just a preview.

  • Now it's time for me to run the model.

  • And here we go.

  • Run remotely so remote.

  • GPU enabled.

  • And you could see just by signing up for runway, I have $10 in remote GPU credits.

  • It will be interesting to see how much just running this once actually uses.

  • So one thing I mentioned now, if you want to get additional credits, I can go over here.

  • This is like the sort of icon for my profile.

  • I can click on this.

  • Um, I'm going to go now to ah, here, Um, get more credits.

  • Is that where I want to go?

  • You go to get more credits, and this is gonna take me to a browser.

  • Paige and I could have certainly pay for more credits, but I'm gonna click here, and I'm going to redeem credits by saying coding train right here.

  • So if you would like to get an additional $10 in credits, you can do this, and we can see now I should have $20 in credits so I can close that.

  • Minimize the browser and go back to, um here.

  • My workspace.

  • My model looks, um Where do I want to go?

  • This is this is workspace.

  • Here we go.

  • There we go.

  • Back here.

  • So this icon up here just so we're clear, this icon up here is your work spaces of which I only have one with one model that's connected to remote GPU.

  • And if I want to look at other models, I would go here to this icon.

  • All right, now, never depressed.

  • Run remotely coming.

  • Could we have a very long time.

  • You think What?

  • Oh, it is so beautiful.

  • I cannot believe it.

  • So that's what a spade Cocoa machine Learning Mile Jared.

  • It's really interesting to see the results here, so you can thank me knowing nothing about this model kind of how it works and what to expect to get some pretty weird results with it.

  • Probably if I were a bit more thoughtful.

  • Maybe if I even like, filled in the entire space, right?

  • Probably I left so much of it Blank also included, like a giant wine glass with two people.

  • It's very kind of creepy looking, although this I think this sort of resembles me in some strange sort of way and we can see here.

  • Look at this five cents.

  • Here's my five sex image, but what's that all?

  • I can live paint, okay?

  • And apparently I can live paint.

  • All right, that's crazy.

  • Let's try doing that.

  • So the reason why I took a long time was it was just kind of booting it up.

  • So one thing I should mention is the reason why they took a long time.

  • It was like spinning up the server and everything to start actually running the model.

  • But now that it's running in real time, it can happen much more quickly.

  • So let's try adding, I want to add something like But what if I want to add an umbrella?

  • Let's add an umbrella to this person and there we go on like this one.

  • This person's gonna have a suitcase.

  • What's a click on that I'm drawing down here?

  • Look at me like not knowing how computers work.

  • Uh, this is the This is the inputs I need to draw here, throw a little suitcase.

  • What else?

  • Well, let's put on some eyeglasses this to make This person looked more like me, and I keep doing that.

  • It's funny how my instinct is to do that.

  • My glasses here, those air, great eyeglasses, tennis racket, cop, knife, bull.

  • I kind of want to start over.

  • Um, we start over.

  • How can I clear this like probably this Oh, fill bucket.

  • Oh, that's what have been.

  • That's useful.

  • Is there a clear the whole thing?

  • But I can undo or trash very clear.

  • So let's try filling it.

  • So what would be a good thing to fill it with?

  • That's like background dirt fence.

  • Let's try floor would.

  • So let's try filling it with wood floor.

  • Oh, Whoa.

  • Then let's put let's try to put like some fruit.

  • Um, so what's there like apple here?

  • So let me put an apple gonna paint that.

  • Well, this is looking much better.

  • Now let's put like an orange right next to it for the couple oranges to make a little bowl of fruit.

  • Wow.

  • This is crazy.

  • Wow.

  • Uh, where's my search?

  • Um, now hear this.

  • Let's see what else?

  • One of their fruits.

  • What is there?

  • A banana is probably banana.

  • Okay, so let's put a banana.

  • Okay, this I gotta stop.

  • That's pretty amazing.

  • So again, here was just a little moment later of being being a little more thoughtful to think about how this model actually works.

  • And what what might know if I knew If I looked at the data set, which is the fairly well known I imagine Coco Image data set That's probably going thio.

  • Give me even Maur information to think about what it's going to do well, but you can see how it's able to sort of think about a sort of little seat a little little pile of fruit here on a wood background on this.

  • Looks like more like cloth.

  • Like it's sitting on a table.

  • Pretty realistic.

  • Andi.

  • Yes, Charlie.

  • England points out, which is correct.

  • This is continuing to use the GPU credits, and we can see that, you know, Still, though, I've even with doing a bunch of life painting, I've just used 10 cents there.

  • So you could do a lot with the free $10 just in playing around.

  • Simon said, spade Coco, I'm just curious here to Google a spade coco data set.

  • Or maybe that was from like, it's the cocoa data set.

  • Yeah, Coco data set.

  • Is this data set?

  • So I would encourage you also to check out this You're alone with a lot more information.

  • What's probably about the data that was used to train this?

  • Particularly segmentation.

  • All right.

  • Um, okay, let me take a minute here.

  • Let's see if people have any.

  • Let me take a minute to see that.

  • Any questions that people want to ask in the chat that I can either answer myself or ask Chris about?

  • Yes.

  • And so, um, I noticed that Chris just wrote in the chat.

  • Some models only work on remote GPU.

  • So if you want I don't know if this, you know, if offhand, if this is one that will work locally?

  • No.

  • So this is one that only works on remote.

  • Cheap.

  • You other models were worked locally.

  • I think I'm gonna try to run pose net, which can run locally.

  • And so it won't use any GPU credits for that.

  • Um, and then some models will work locally, but require another installation, another dependency, which is something called Docker, which I could talk about her install at some point.

  • But I know that Runway is working on making that process even easier to be able to run stuff locally.

  • Um, okay.

  • Can I lower resolution to speed up?

  • That's an interesting question.

  • So I hope I can see here that the resolution is set to 6 40 by 3 60 Is that something that can change with this model?

  • Good.

  • No, but the model I couldn't change the interface.

  • Yeah, so?

  • So this is a good question because you know the way most machine learning models work in particular, if they're fixed inputs and fixed outputs, not like a kind of sequential model.

  • The output is fixed, the input is fixed.

  • Requires a particular resolution because the neural network is expecting a certain amount of numbers to come in.

  • So you can do to some extent you could do re sizing and re sampling if I go back to my diagram here like I could do some re sizing and re sampling here.

  • Or I could do some re sizing re sampling here before to prepare the input or after I get the output.

  • But the model itself, the input image has to be a particular dimension.

  • And the output is only gonna come in that dimension.

  • Some some models are released, I think with, you know, like a different version, doesn't it with which may be output different dimensions.

  • I think I've seen that.

  • But in this particular one, it's fixed dimensions.

  • Okay.

  • Um, all right.

  • I couldn't, like play with this for ever.

  • I could play with this forever, but I think I've got a knot.

  • All right.

  • Um all right, so now let's let me think about this.

  • What else do I want to show here before I I I think what I'll do is to separate, like, sort of like tutorials.

  • One using Poe's net with processing and then one using this style.

  • Gan arraign Bos with p five gs.

  • But I don't think it's just in sort of this, like general tutorial about the platform as a whole.

  • What else do I want to say?

  • Um so let's look at so input wise, I could also.

  • So I chose to do segmentation here, but I could also use a file.

  • So if I wanted to open a file on the computer, I could do it that way and then output.

  • If I change to export, I could also actually export that thio of riot, different formats.

  • But of course I could also right here just under preview, I can click this download save button, and now I am saving forevermore that this particular image of the file now what's really important here, actually more important here is under network.

  • So if what I wanted to do let's click over here under Network. 00:52:46.390 -

as always I always forget that this Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop, Stop!

字幕與單字

單字即點即查 點擊單字可以查詢單字解釋

B1 中級

現場直播#182:RunwayML介紹 (Live Stream #182: Introduction to RunwayML)

  • 0 0
    林宜悉 發佈於 2021 年 01 月 14 日
影片單字