Skip to content
TrackPodcasts
technologySep 13, 20261:16:46

🐙 Lunch & Learn: Let’s Create AI Videos! | Tina Huang

Tina Huang

About this episode

🐙 Lunch & Learn: Let’s Create AI Videos! 🤖 Sign up here to get accompanying workbooks, summaries, and notifs for future lunch & learns: https://www.lonelyoctopus.com/email-signup ✉️ NEWSLETTER: https://tinahuang.substack.com/ It's about learning, coding, and generally how to get your sh*t together c: 🐙 Lonely Octopus: https://www.lonelyoctopus.com/ Check it out if you're interested in learning AI & data skill, then applying them to real freelance projects! 🔗Affiliates ======================== My SQL for data science interviews course (10 full interviews): https://365datascience.com/learn-sql-for-data-science-interviews/ https://365datascience.pxf.io/WD0za3 (link for 57% discount for their complete data science training) Check out StrataScratch for data science interview prep: https://stratascratch.com/?via=tina 🎥 My filming setup ======================== 📷 camera: https://amzn.to/3LHbi7N 🎤 mic: https://amzn.to/3LqoFJb 🔭 tripod: https://amzn.to/3DkjGHe 💡 lights: https://amzn.to/3LmOhqk 📲Socials ======================== instagram: https://www.instagram.com/hellotinah/ linkedin: https://www.linkedin.com/in/tinaw-h/ discord: https://discord.gg/5mMAtprshX 🤯Study with Tina ======================== Study with Tina channel: https://www.youtube.com/channel/UCI8JpGrDmtggrryhml8kFGw How to make a studying scoreboard: https://www.youtube.com/watch?v=KAVw910mIrI Scoreboard website: scoreboardswithtina.com livestreaming google calendar: https://bit.ly/3wvPzHB 🎥Other videos you might be interested in ======================== How I consistently study with a full time job: https://www.youtube.com/watch?v=INymz5VwLmk How I would learn to code (if I could start over): https://www.youtube.com/watch?v=MHPGeQD8TvI&t=84s 🐈‍⬛🐈‍⬛About me ======================== Hi, my name is Tina and I'm an ex-Meta data scientist turned internet person! 📧Contact ======================== youtube: youtube comments are by far the best way to get a response from me! linkedin: https://www.linkedin.com/in/tinaw-h/ email for business inquiries only: [email protected] ======================== Some links are affiliate links and I may receive a small portion of sales price at no cost to you. I really appreciate your support in helping improve this channel! :) Follow this podcast to get Tina Huang’s insights in audio format, perfect for learning on the go. Tina Huang on YouTube: https://www.youtube.com/@TinaHuang1 Disclaimer: This podcast is an independent audio adaptation of content originally created by Tina Huang. It was made by a viewer who values her insights and aims to make them more accessible for audio-first learners. This is not an official production of Tina Huang, and it is not affiliated with or endorsed by her. All rights to the original video content remain with Tina Huang. If you represent Tina and would like to discuss this project, please don’t hesitate to get in touch. --------------------------- --------- Keywords: chatbots, coding tips, ai apps Learn more about your ad choices. Visit megaphone.fm/adchoices

Get every episode summarized

Each time Tina Huang publishes, we email you a written briefing from the transcript — the topics, who appeared, and any specific claims, with the ad reads skipped.

Email me new episodes

Free for 3 shows. No card needed.

Hosts & guests

Transcript ready

1,022 searchable segments. Every word is indexed and playable.

🐙 Lunch & Learn: Let’s Create AI Videos! | Tina Huang

Tina Huang

0:00
1:16:46

Full transcript

Tina Huang🐙 Lunch & Learn: Let’s Create AI Videos! | Tina Huang. Machine-transcribed; use the interactive transcript above to jump the player to any line.

Hey friends, how's it going? Hey Christian. Hello Hussein. Hello. Hello, peace. Good weekend, Tina everyone. Happy weekend everybody. Happy Saturday. Hey Ivers. Hello friends. Alright, so today we are going to be testing out videos. I downloaded some like, have you guys played with AI videos before? It's out of curiosity. Have you tried it out before? Your mic is off.

Hmm. It looks like it's fine. It indicates that it's fine. Is my mic working? No, yes maybe. Yes, no maybe. Is not working really? Hmm, interesting. It does show that it's completely fine. My mic is working. Okay, cool. How did you guys get access to Pika? I haven't been able to get Pika for, it's like I'm still on the wait list. Yeah, I was kind of disappointed. I was really hoping that would be able to get it before this livestream. But, alas, I have not. Is it good? Is it better than runway? You think? Hmm. I mean I can just boost up the gain.

Yeah, boost it up the gain. Okay, cool. Is that better? Just got access yesterday. Ely, because I got it. I've tried. Yeah, let me check. I'm pretty sure I still don't have access. And I said I don't look a while back. Is it better than runway if you guys have tried that? Pika. Hmm. And we just double check that I still don't have access. Nope. Sad. Still don't have any access. Hmm. Let's see.

Nope. I'm just going to go. Dang. Well, if somebody let me know how it is that we, and I was like, wait, Pika. I still have one on the way. This. Sorry. I'm just like double checking because I wasn't. I'm pretty sure I was. Dang it. Okay. Well, anyways, hopefully at some point, it will work out. But yeah, let me see. Is there any comments if anybody says that? Just send you a discord invite. Oh, can you do that? Let me see. Can you do that? Like just go on discord. And you just send me an invite. Let me see back here.

Okay. Which who are you? There's like a bunch of people. Um, Deja, do you have a different username? Okay. Well, in any case, why don't we just get started and fingers crossed that we managed to use Pika as well? Okay. So let's first try out. Let's first try out. Okay. Let me share my screen. Should have this. Okay. I feel like I'm such a mess today.

Yeah, I feel there's just some days where I'm like not able to figure things out. It's like me today. Well, you did not need to see that. That is not the correct thing. Why is it even showing that? Close that. Close to what the hell is happening. What is happening in the meantime? Please entertain each other. That accused story, though. You don't need to know. I mean, you can look at it. Yeah, maybe you should look at that. It's I cannot. It's um, I was going somewhere. You don't people yesterday. Um, well, let's see. Nothing seems to be working today. You know what? Give me one moment. Let me like get my shit together.

Okay. I think that should be good now. Okay. Cool. Sorry about that. That was a lot more difficult today for some reason. Um, all right. So I was going to start off with. Um, you must be using a fruit company computer. What? Pica is like this really. There's like new hyped up video generation thing that I was hoping to get access to as well. Yeah. So yeah, I wanted to test out like a few of these have you guys heard about runway before. So runway they recently come up with this gen two. So they use that gen one. There's a few things that you can do on the side bar. Um, you can generate images. So you can generate images. Um, sorry, you can generate videos. You can also generate images. But you can generate videos.

Um, you can also use it to edit videos as well and generate audio too. So I figured that we can try both like generate video using text image or both. So we can use like text. So I want to say like. I sure you know what? Let's start off with an image. So I have like a few images that I saved here. Let me know if you guys want to see any different ones, but I have one of Rockley which. I'll upload here. Um, your video is cute. So it's going to be one minute to do so. I'll show you guys the motion brush as well while we're doing that. We can try to like see like animate. You can use like this little thing. I want to call it. This.

To like generate where it is that you want to have the animation. So say like right now we want to have the animation on the legs. So we'll see how that goes. You can just like mark where it is that you want it to do it. And then you can also do different things with the horizontal and vertical axes. So let's say like we want horizontal like that. And then on the vertical side let's do this as well. We can also do main for the proximity. So as it's a kind of like coming back at us. So let's save that and then we can. Generate that as well. Let's wait a little bit to see what it comes up with. So here we have like a okay it's here. So here we just have one that is just generating video by itself. And then the other one we have just trying to animate the legs and see what happens. I have actually not tried that. Oh dear.

Okay, that was not great. Interesting. Okay, well hopefully the one with the legs will be better. I did do one like with myself in animating my hair, which turned out pretty decent. But what the fuck? That's scary. Let's see comments. A hair racing experience, which program are you using for video generation? If you guys want to try it out as well. So you actually get a lot of credits for free. It's called runway. So let me just share that with everybody. Also link everything afterwards in the descriptions.

So you can try this one out. Well, that was a fail. It's not great. All right, let's see. Okay, what is this one going to look like? Oh my god, I like scared. It didn't do anything. Okay, it had like a little minuscule. Oh, okay, it has like a little minuscule moving the legs. Say like we said it by four seconds. What would happen? Okay, I want to show you one that I have done, which does work quite decently. So I'm going to show you guys one that does work. Let's see. So this one of me.

Very fresh. So I had like a little like movement, but you know, not that great. But we'll try like this. This one I was able to get it to work. So for the motion brush here, I was trying to animate was my hair. Oh shit, that was my eye. I think you need to see like my eye becoming really weird. So you can I'm just animating the hair. Say it looks like this is the only part that I will try to animate. Okay, cool. So horizontal is going to be a little bit of horizontal. And then you can also do camera motions. So you can do like a roll over and like pan over something like that. So yeah, that's I'm just going to try this out. Fingers crossed it actually works this time.

Any suggestions for the text input by the way? You're doing already better than that. Similarly, as I'll buy Pikachu, there's a human character. Sometimes the effect is scary with the characters. I'm all really crazy or something like that. Yeah, like for human characters, it is a little bit more challenging. So I also have like some other pictures that we can try out. Maybe that will work. Yeah, in the meantime, let's try another one. So I have one of Ditto. The Pokemon that is. So we can try this out. I guess like, do I really need to watch this kind of? Do this. Just get all of that. I don't actually know if I need to do this. Okay, I have a feeling that the mouth is going to do something really weird. So I'm actually going to erase the mouth part.

Like this part, I'm going to erase. It doesn't do something really weird. We can do like horizontal. I think we can do like vertical. So it kind of like goes downward. I think we can do it zoom. And then. Here as well. Let's do like a roll in a pan. So I was generating this one. Okay, okay, okay. This is going to work. Yeah, yes. Oh, I got it. Wait, it actually. The one that I generated before didn't play. This is actually pretty decent. Okay. All right. This is what you guys think. Relatively decent. It's like the hair and eye animation. My brain is too underdeveloped for this. Like it. Well, thank you. My beautiful shiny hair. Does Chattanoid also can be used to do this. It doesn't do image. It doesn't do video generation now. But it does do image generation.

That's pretty cool, right? Like I think it's pretty cool. We'll do like one. Let's look at Ditto. Tina is winky. Is it both. Oh, yeah, it's like one of my eyes. Okay. We can do some like text-based ones. And it's usually a lot better at text-based ones. But this is kind of like really interesting. I can totally see a lot of like animation based stuff. And I'll show you guys a little bit later. Some of the new things that are coming out, which is not necessarily available now. But what people are working on the results they have. This space is just evolving so quickly right now. And seeing like full animation, like AI-generated videos, like movies and stuff, I can totally see something like this happening very very soon. Yeah, I would not be surprised to see this. I try to use it with just text on my images disappearing. Wait, I can share with you guys some of these images. If you like. Okay, let's look at the Ditto.

Oh my god. I look, I feel scared. Oh no, what the fuck happened? Wait. What the fuck? It just did. Wait, I thought that I thought that it was supposed to animate things except for the mouth. But it fucked up and animated only the mouth. Oh no. Okay, last one. And then we're going to do text-based stuff. Okay. Make this a scary. Can you create a video just by touch? Try something controversial. I'm going to, I'm going to, this student's going to get taken down if I do that. You can though. You actually can, I think. I don't see why it wouldn't let you do that. Imagine doing non-stop motion like video or chat to PT or you just tell it to move a hand bit by bit. Well, if you think about images,

like a video is just a series of images, right? So if it's able to generate consistent images of things and then you just like stick them all together that would be able to create animation. One of the issues that AI video generation faces now is that if you try to generate a bunch of different videos, the place of the person is not consistent. So that's kind of the issue that it faces right now. But I do, there are like people who are working on that. Oh my god, this is like so creepy as well. I'm like scared. I'm just scared. Okay, this is not as bad. This is okay. Right? It's a bit slimy looking. But this is okay. Okay, let's do some like text-based stuff. Let's say, clownfish, swimming in the ocean, or say like clownfish. Yeah, it's starting with something easy.

You can also add style. So we can do, want to do anime style? Let's do anime style. Okay. Generate. Yeah, I think this one's not too bad. I don't know what happened. It died on me. Okay, cool. Um, why is it not? Are you guys able to see the new image? Why am I having so many technical difficulties today? Can you even see the ditto image? Today is not a good day. Well, how about you? Oh, you guys came and see the ditto. Oh no, what is happening? I don't know. Today is a really weird one.

Okay, let me try again. Um. I'm just trying to change it around. Okay, there you go. Okay, so here's the ditto stuff today. Damn. What is happening? Technical difficulties, my friends. I'm just trying to change it around. This is the ditto image. Yeah, that's the ditto image. It's not that bad, right? I feel like it's not that bad. This one is fucking scary. Look at that. It's like so scary. But this one is okay, I think.

I hope you guys are able to see that now. I think your, my, your share is okay. I think I think it works now. Can you create a video, Athena having something controversial? Please don't, please don't do that. Please. I have seen people do that with other people though, which is very concerning. Um, right? It is just like breathing. Really. This is the thing of nightmares. Okay, so for this one, I wrote clownfish swimming in the ocean. Um, that clownfish looks, that clownfish kind of looks terrifying. Fingers crossed. Okay, it's not terrible. It's a little bit horrifying the way that the clownfish looks. But why don't we say clownfish. Octopus hiding in a reef.

Let's try this one. Put in the comments, please. What, what should I try? Let me know what I should try. Did I use acid? Realistic. All right, Ewoks slowly roasting. What is an Ewoks? I must try. Is that a bird? I can't, oh, it's from Star Wars. Okay, I've actually never watched Star Wars before. Terrifying. Okay, that Ewoks slowly roasting over an open fire. That sounds terrifying. Maybe a slob in a scuba, maybe a slob in a scuba suit. Let's do that. Oh, I like that as well. Dinosaurs are baking cookies. Okay, let's try those. Let's try a few of these. What is the slob supposed to do in a scuba suit? I think it needs to have an action.

Maybe walking around in a scuba suit? Could do that. Okay. A slob in a scuba suit walking around. Or swimming around. We're ocean themed today. Let's try something that's not anime. Digit art. Let's do that. Okay, let's do another one as well. Be hysterical. Dinosaurs baking cookies. I like this one a lot too. Let's try it one more here. And then we'll move on to our next tool. Okay, Octopus.

Let's see how it does. Okay, I think that's fine. Stop particularly hiding in a reef, but it's amongst the reef. So that's fair. Yeah, okay, what would you guys rate this? Like one one to five in terms of how good it is? I guess maybe we need to do another comparison as well. I think for me, I am impressed. Like it's clearly not working that well, but I am impressed that's able to do any of it. Because often high on with new technologies, it's really like the first time that you do something. It's the really hard part. And then you just kind of optimize on top of that. So let's do this. Let's let it do its thing a little bit. In the meantime, I wanted to show you guys some of the new ones that don't have code yet.

Oh, yeah, I'll show you this one a bit too. They don't have code yet. But for example, they have this one called anime anyone. It consists in controllable image to video synthesis for character animation. That is pretty cool. Right. That is able to do that. They also have a video explaining how all of this works. If you're interested, also link it in the description after the video. I'll email you if you sign up for the AI newsletter, which is an email week. I'll email you some of the resources too. So it's like this idea of you're taking a reference image and then you're giving a pose to do. And then you're able to animate the video. That is pretty impressive. So I'm interested to see this actually come out. There is nothing like you can actually play around with it yet.

I'm very... This is like a lot better. Humanoid. And then fashion video synthesis as well. You can take a person and just have it dress as different people with not different people. You can have a single person and then you can dress it in different outfits. This one is also pretty cool. Yeah. It's like dance based stuff. You'll notice a lot of these new kind of research. They're Chinese based. So it's two for intelligent computing. This one is by Ali Baba. A lot of them, this video generation, is from China. Hello world. I'll thank you. I appreciate Evan Gambler. Thank you for supporting the channel. Can it generate transparent video? The one for Runway cannot at this point. No.

I can't wait until we can remake movies with actors we want and changes we want them. So Metta did recently, maybe like a month back, it released this thing where it shows you like... You can have like AR influencers that are able to do like slight movements and things like that, which is kind of interesting. So you can do certain things right now. It's not perfect, but I definitely can see that happening. Okay, let's check back on our little... Ooh, our sloth. It's not super interesting, but it is a decent image and movement and then dinosaur baking cookies. Oh, this is cute. Okay, I like this one. It's like little arms moving. Yeah, this is cute.

This is a really... I think this one's pretty good. Is it making... Oh, although I don't think it's making cookies, I looks like fried chicken. It looks like some stew and fried chicken is what it looks like to me. Interesting. Oh, I like the animation in the back. Okay, this one is probably the best one I've seen so far. Okay, let's move on to another one. So this is called stable video diffusion. So it's by this, like... You can try it out using GitHub and then they do have a go hugging face that you can try out as well. So we can try to do one of these. It does take a little bit longer, which is why I already tried it ahead of time. A stable diffusion one. A stable video diffusion. So this is the reference image that I gave it. And this is what it did. I feel like it's still the face, right?

In terms of a lot of the other parts, it's actually not that bad. It's like the face and the hands. That is difficult. There are these memes though. I would say like runway. Would you say they're about pretty similar? There are like these memes. Like this one is pretty good. Yeah, it's just going to take a little bit of time to do. As it's queuing. The face goes to blur. Yeah, I would have goo with that. It's kind of like... It's the face and the hands. I think people is hard. Like the dinosaurs is good. It's pretty decent. Cookies doesn't look that cookish. But it also released beta version iGlass and they are cool. Ooh, I heard that for Oculus.

I did work at Oculus previously, but I did hear that they were able to put into Oculus. It's more like AR based now as opposed to VR based. Looks like the dancing demos are more often girls than octopi. Yes, I do believe so. Sansa ignition horizon and warm hues shimmering sea. I like this one. Let's try that one. Try to text face hugger on a dinosaur. It's really good considering it's a free test app. Yeah, all of these are free. Runway gives you a lot of credits. So technically there's a paid tier, but it's not necessary at all. Like there's a lot of... They give you a lot of credits and these are all completely free. It's kind of like the demo versions that they released. Just like I would expect AI to be studio tools are expensive and resource heavy to make perfect face glasses, my guess. Give me one second. My doorbell just ring. Let's see if this comes up. PRB.

I am back. It's good. Yeah, let me give you the links for these as well. So I give you the one for runway. Let me give you the link for this one. Steve. Whoops. Okay. This one. I did in fact just walk out midstream. I did. Hey Tina, you heard about AI music. I have. I have. So I in my Gemini versus GPT-4 versus GPT-3.51, I did test out the one with music, which is actually decent. I feel like if I were to do a ranking like chatbot stuff is the best. Like text-based stuff is the best. Then music and then, okay chatbot stuff. Um, music slash images, not sure which one is better at this point.

Something that likes, I think music slash images and then video. Like video is kind of like the next, the next big thing. Yeah, I think video is 100% the next big thing that's happening. Very, very interested in how that's going to turn out. Just by just speed of how these tools are coming out in the amount of improvement that's happening though. Oh my god. I feel like by mid next year, we're going to be able to generate like full on animation. Based videos, that would be my guess. So definitely important to keep up with everything that's happening in the air world. A lot of things are happening all the time. Okay, let's see if it was able to generate something. No, it was not. Why it be slow? Okay, well, you can test it out if you like. Um, it is kind of slow right now, but they do have like demo versions. So it's pretty cool. It's pretty cool. It's completely free. So can't have that much. Can't do that much. Expectations can't have that much expectations, right?

Okay, yeah, so this is the one I showed you was able to animate characters. If you click on the code base, there's nothing. It's just docs with nothing in it. Yeah, so I'm going to check back maybe a little bit later to see if there's something that we can try out later. Um, animate anyone. Yeah, this is animate anyone. Oh, yeah, this is, oh my god. This is nudity. I haven't checked up this before. We're going to change this one. That was nudity. Oh my god. Let's try like with an actual model. This one is called magic. The fuck I swear there was no nudity in the reference photo. This is called magic enemy. I'll temporarily consistent human image animation using diffusion model.

So it's able to animate. So this is like another model that they have. You would upload a reference image and then for it to do something like running and it's able to animate say like the Mona Lisa that's running here. Maybe we'll try like one of these as well. So this person running to which is pretty decent, right? But I'm going to try with another image and hope that it's better because I wanted to try with something myself. I want to go look at the comments. I'm going to learn to make these guys. As you reveal that Tina has been AI generated this whole time. Could be. Could be sad loss opportunity. Could make you laugh as well as demo the technology. What? What is it? Code imagine. What did you say? That is going to be interesting. It really slow.

Try to be slow. What? We're not going to talk about this. All right, moving on. What do you do for fun? Apparently create AI images of people in myself. Can you make video on how people in college should approach you without making money? With making money to buy a MacBook, that's very specific. After college or doing college? I feel like I have quite a lot of career based videos. A lot of it is just experimenting and then getting a lot of internships. That's how I would say the best way of approaching this. It really is. If you just experiment, like if you go to college and go to professors and ask for internships, usually they convert to being paid. I think that's the easiest way. That's what I did at least. Don't look at the comments. They's hugger on a dinosaur. What does that mean? Is that bad? Is face hugger on a dinosaur? That sounds like it's bad people stuff.

Like suspicious. Midranny, Aldo, Tex and Heizim and Mowley uncover results behind the notice. What's the link to this hugging face face? Let me give that to you as well. There you go. Yeah, the ones that they do have is pretty good. I am going to try to upload an image. Maybe an image of not myself. My goat deer. Why don't we animate an image of link click? This is one of my favorite. This is one of my favorite animation. Okay. Oh no, error. Why? Let's try again. A running motion and then we'll try again with the guy from link click.

His name is Cheng Soucher. Oh no, that didn't work. Choppy Mitch here. There. Maybe this time it will work. I guess we're just getting error messages today. Well, that was sad. But yeah, at least the ones that they showcase seem like it was pretty good. It does seem like it is cherry picked though. Why is it? I want to try one at home. Maybe it's the Mitch. I'm trying another image. Okay. Let me try the one running. I guess I'm just getting errors now. Damn. Okay. Well, there's another one that you can try out. This is an interesting one called story to motion. Did design infinite and controllable character animation from long text.

Again, very interesting. All these seem to be Chinese stuff. So I think Chinese is very much leading the AI video scene. Story to motion is a new task that takes a story. Talk green area generates motion and trajectors that align with the text description, which is really cool. So essentially you're able to take a task and tell it to do something which would relate. So here we have dancing in front of a modest building, dancing around trees, eating hot dog from a food cart. We'll have a video demo of this. So let's watch that. The prompt that they have. Charger beauty generates the prompt.

As proof, we'll do it in 7.5 seconds of dancing. Amongst the trees. Walking towards the food cart. Eating hot dogs. Goes to the puddle. Jumping over to puddle. Oh, careful. If this was for real, I would be. If this was like actual.

Wait, are you guys able to see that? Okay, yeah, you were. If this was. If this is actually possible, that's really cool. I can totally see being able to maybe like NPCs. Like NPCs and video games. I could totally see that. That people are yeah NPCs and video games and just generating characters doing stuff and video games and stuff. Damn. That's pretty impressive, isn't it? That's like if it's for real and augmented reality, I can totally see that as well. Like people just walking around you like fake people just walking around you doing things. This is scary how good it is. Artists. Yeah, if it actually works, I'm very excited to see if this actually works out. But do you guys see what I mean? Like even though it's not super great right now, the speed of how things are progressing. It's really crazy. What could come out of it? Yeah, damn.

Oh yeah, the screen is a little bit. Yeah, this is like so crazy. How long is your pad for AI? Artists made a plaster figure and blowed in its nose. Try to put a one piece character in one of the animations to see how to limit his construct or some famous anime. So I tried to do that one. It gave me an error. So we'll try again a little bit later. Maybe it'll work then. I'm not impressed. Why you not impressed? Why you not impressed? It's not impressive. Tina, you should do more squats. Yeah, you're probably right. I should do more squats behind them. Yep, it is. Oh wait, it's closed source. It's on GitHub. I'm assuming it will be open source at some point and they would have like a hugging face space. But there's currently no code. So I don't know if it's going to be open source. But why would you have something and not? I guess they could just be hiding the model. I'm not sure if it's going to be open source or not. We shall see. But right now you can use all of these things.

Do you plan on making your own AI products at all? What kind of AI products? Code, I'm trying to animate a goose with a butter knife and it's giving me errors. I'm never impressed. I'm the negative Betty of the channel. Okay. What can we do to impress you? We want to see AI Tina. Yeah, I want to see AI Tina too. Like maybe AI Tina can do all the things. Can do all the things that normal Tina. Tina's not playing nice with my suggestions. What's your suggestions? I'm sorry. I missed when I'm reading through the comments, I kind of miss. Miss things as well. You know, wouldn't it be a riot of the AI made a face hugger attack? You want to do a face hugger. Okay, Alien movie. Okay. All right. I promise we'll do that. I promise I will do that after I show you guys one more thing that I think is very interesting. Or like two more things that are very interesting. So this is again another one short to long form video diffusion model of gendered

transition and prediction. It's very again, Chinese based Shanghai Artificial Intelligence Lab East China Normal University Shanghai, Jialtong Monash universe. I don't know if that's Chinese. Niaan technology. Work done during intership by Shanghai. That's pretty cool. So work done is during intership by Shanghai AI laboratory. I think they do have a paper. So they do have these papers, which you can check out. But again, I don't believe there's any. Oh, wait, this one does have code. Okay, cool. Is the official implementation. Stable diffusion. Is there like a space that we can do it? Okay, let's move on. And then we can see we can manage to get this to work. But apparently what is able to do is much better animation than what we see on runway.

Input image, output image, water. Output image is able to combine them together and make something. So you can produce a spider man. And then something else to do that as well. So yeah, that's another one. So I did want to show you some, you know, how I get I would say like it's a bit finicky, right? I would say usually things are a bit finicky when you're looking at video stuff. But there's some people who are really good at doing this. So for example, this one. Somebody actually managed to make you should definitely check Twitter or X. If you're interested in AI stuff and just air videos and things like that and hers at progress. Following these people is generally they are able to come up with really cool things that you might not be able to do yourself. So maybe making it into a cartoon. That's pretty cool. That's really, really, really cool.

Another one is this one. So you can look at these that are here. You're able to take different outfits and then just put different outfits and heard of people. If you just go on X and just write AI video, I video AI creator and these things, you will. Yeah, you'll be able to see a lot of like cool things that people are doing using your video. Okay, so the last thing I want to show you guys is called motion. So this one, I'm definitely. It's on discourse is kind of like mid journey still, but what you can do is you can take. Something here and then you can say like motion, you can give it a prompt like climbing tree.

You can choose from different characters. For example, we can choose like a cat. And in place here. This is also completely free by the way, all of this is free. So it's just generating something here. As we wait, we shall see. A itina coming soon, I'll be the first on a jump on a itina. Definitely. First on a jump on it. I don't know if I should feel scared being too happy of how fast we're getting. How do you guys feel? It's honestly speaking, how do you guys feel about how quickly everything is progressing? Just a few months ago, we were all like marveling over AI generated images, right? And then the next big thing now is AI generated videos. Everything happening so quickly. How do you feel about that? Like are you excited? Are you slightly concerned? Very concerned? Looking forward to it, not looking forward to it.

If a itina could host lunch and learns at some point, that would be very cool. Like some of you want AI to rise up. Oh yes, please do like the stream. Soon in future, I'm not impressed with the pan guy. Cypher had to read that. Van gambler, AI Tinas hosting a lunch and learning the top corner is the most impressive AI I have seen. Same. Same. I love it, but I studied AI. 70% concerned. Cypher, wondering how long before we start making friends with AI? Oh, have you guys heard about the chat bots that people have been getting emotionally attached to? If you've seen the movie, her, it's definitely a thing now. Like people who are talking to chat bots getting emotionally attached to them and things like that. So also, I was actually thinking if I should do a video about this in China, the application of AI is a lot more widespread than it is here.

Like AI influencers is very common now. AI influencers just being able to sell products and things like that. It's actually quite common these days. A lot of things are AI generated now. AI videos being there. Yeah, so I was saying China the progression is not necessarily faster than it is in the US, but I do anticipate that it would become faster. And at least the application of it will be faster than it is in North America. You just generally see that overall. Like in China, you don't use physical money anymore. In China, everything is pretty centralized. And the reason for this, I think, is just because here it's slower to get because democracy. But there, somebody makes a decision that, oh, we're going to use all of this now. And it's so centralized that the dissemination of it is really, really quickly. And I feel like Chinese people are generally also more accepting of these new technologies. And compared to people in the US, for example, that's how I see things.

I actually will be going back to Beijing soon. So I was thinking if I should do like some videos about this, like for example, self-driving cars, it's been a China for a very long time now. You can see them on the streets all the time. Robotics and stuff. You can see them everywhere too. I'm a mix of emotions lately about it. I haven't seen her by believing it. Yeah, it's like this guy who falls in love with an AI. Tina, what would you do if you knew you were going to lose your job? I wanted to start a career in AI. What kind of skills would you develop? This is a really good question. I actually want to talk about this with I personally am very excited, but I'm also kind of afraid in several different ways, right? From like a selfish personal perspective. I'm like, oh my god, there's AI videos that are going to be coming out. What's the point of having people doing my job by creating content? Right? But at the same time, I also see like so much that can be done using these technologies. So I'm pretty like confused in that part.

But yeah, that's like a selfish one. And then the other concern that I do have is everything is being done so quickly in policy is 100% not keeping up with this. It feels very, very concerning. We're very much, I don't say like there's a lot of similarities now with the Cold War. I mean, I wasn't alive during the Cold War, but it seems like that. You know, when you get big powers clashing with each other, trying to compete against each other, companies competing with each other, and getting more and more things that are coming out and regulation. I mean, they managed to not have a nuclear fact. So war. So, you know, thank God for that. But I don't know about about this anymore. Yeah, like I am pretty concerned about these. I with that being said though, I rather ironically, the only way that you can deal with this kind of thing is to invest in learning about AI. It really is. Like that's the conclusion that I came to is the conclusion that at least from other people,

it's the only one that people came to. Because if you want to have a save for how these things are going to be developed, you need to learn about them and have your voice heard. Because people are not going to stop developing these technologies. Like no matter what, they're going to keep developing these technologies. It is, it is what it is, right? Tina, okay, I haven't seen her. It changes the foundations of society. I study at AI University and I'm scared. TPH. Are you scared? Can I see the application you talk about down below on YouTube? Yeah, if you sign up for all link below as well, but in general, I usually link resources for the weekly AI events and stuff. So if you sign up here, you'll get that directly in your inbox. While I may, it takes me a little bit longer to update descriptions. We're honestly, I might forget to do that.

If you sign up here, definitely it's, it's being emailed to you every time. No spam, just event, notifs and resources. I was a live during Cold War. It didn't impact regular people in my country. And right now it's nuclear. It's nuclear to cause of constant new threats and readiness. Okay. So really interesting. Because I was hearing that people were like super scared doing a Cold War. And then children were all being like shepherded around telling them what to do if Russia decides to bomb you. Like something like that. But then again, I wasn't alive during that time. So I couldn't say, okay, what about now? Would you say the threat is more so than it was before? Oh, thank you, Lala. Lala Papa. I study and I'm less scared. See, that's what I think. That's what I think. Because I think the more you understand something and you study it, the less afraid that you become because you understand how things are progressing. You at least get to make that choice about what it is you're going to do as opposed to if you're just like not doing anything, then you're going to end up, you know, if you just choose not to do anything, you're going to end up hoping that whatever comes your way is going to be beneficial for you.

And we don't know if that's going to be the case or not, right? Like we don't know if it actually is going to be beneficial or not. So it's better to the best way to be prepared is to learn about these things and understand what is happening. Like keeping up with the overall trends of the industry is really important. That's also why it's a difficult topic, but I like talking about how it impacts careers and things like that because it's such a big component of people's lives. It's also why I like to read a lot of papers about it. It's the same though. People have different point of views. Most people don't understand cyber warfare. I grew to Tina fear and ignore is there to old enemies very much. If I learned how long until AI is mostly shit, mostly hit memes. Thanks for the things for the comedic relief already is already is. Have you guys seen. Not to like fan fan defiers, but I do want to, you know, kind of show you guys something that you can check out.

Like this. I would say is the thing that terrifies me the most when I did watch it. I felt very scared. And I think it's kind of it's a little bit. I don't think it's the most accurate thing. It's more so you know how like they usually like to show you things that are very. Let me show you guys. Yeah, they usually like to like show you things that are a little bit more kind of scary than it usually is to to get you into action, which you know, I think that's fair. But there's like this video called like slaughter bot. Oh, I think it's not letting me show it, but I'll like link it link it below. That is probably like one of the scariest scenarios that I can I can think of at least. On that note, let's see about our shit posting memes. Okay, I want to see did it end up working out. Okay, yeah, I wanted to see this one creation one.

We had. Okay. This is one that I did previously. Where my friends were to go. Which one did I do it under? I don't know which one that I did under. Do you guys remember was it creation? Three. Okay, I don't know where it went. Oh, here it is. Yeah, so this is you kind of have a character. So if you do map out the character, it's able to do things. I guess it's kind of climbing a tree. Yeah, which is pretty impressive. I think that's pretty impressive. What was I going to say? I think in the future, if we map out each of us, then it would be a lot more accurate. This is what it's telling me like you're generating something from scratch. Maybe it's not as good as doing it.

But maybe the future is that you're taking like 3D models of yourself and then being able to use that to generate things like dancing, like walking through walking through different places. The thing that we saw earlier. Yeah, the thing that we saw earlier about being able to walk with this. I feel like I have a feeling that's what's going to happen. Like people are going to start uploading like different versions of themselves, like different viewpoints, and then be able to generate these AI characters. Yeah, I think like people who was like Kendall Jenner, these people, they actually sold their image and likeness to meta so that they're able to become AI influencers and things like that. So I feel like that's probably what's going to happen. I have a question for you guys. Like, would you do that? Like, would you be one of the early adopters or something like that? Or would you wait until later to do it? I have a feeling that it's going to happen regardless. Yeah. I have a feeling it's something that's going to happen regardless.

Or would you feel like you would be one of the early adopters with something like this? I need to study. Same. I haven't seen her. It's the same though. People have different point of views. Most people don't understand if I learns her most. Everyday life was normal. People went to work, cold water, and the news. I was child. I was fine. Okay. Okay. I need to study. You're my teacher. I need to study too. People were born to see playing, go to spaceships. I saw no internet, internet changing, everything. I think AI may be one more thing to change world, but these changes may happen more in each lifetime. The development of things is definitely accelerating. The amount of innovation is 100% of accelerating over time. You took a very long time for us to figure out how to make fire and print things. Right? But now it's very, very quick.

Sex level technology when it goes into effects and modern day loneliness and AI attachments as well as VR. The real offload early always gets more money when hired. Okay. So let's see early adopter and is going to be early adopter miles, stuck owns the metaverse. Does that mean early relate later adopter in a way, yeah, the real upload early always gets more money when hired. Okay. You see that's that's only what I mean. It's like if something's going to happen. The only choice you have is to keep up. Right. Like whether you're concerned, whether you're not considered, whether you're excited to only correct thing to do is to keep up. There is a book if you're interested in understanding how it is to like this kind of cycle. It's a little bit depressing. If I'm going to be honest, I go, I think it's like actually depressing. Because a lot of things, you know, people have to be like kind of depressing in order to trigger other people to act, but it's called, um, D. What's it called? It was called like the containment problem.

If you're interested. I think I don't know. I forgot what it's called. Containment is not the, oh, sorry, it's called the coming wave. Here's like a link. If you want to check it out. Kind of depressing. But I do think it's trying to be as is trying to like tell you some stuff. All right. Cool. Well, okay. We're out time. I'm going to run and go get some food. How do you guys like this kind of livestream? Like, can you could you let me know? Like, do you like this kind of livestream? We're just like looking at stuff together and talking about the implications and things that we can do. Um, it's like more chill, I guess, than me just making a video. Also, can you write in the chat suggestions for what we can check out in the future? Things that you would like to do. I would play Mangalo, play around a bit by get bored. I'm not a creator. So it's just for fun. Say, well, would you upload an AI version of yourself?

If you get a lot of, yeah, like, would you, would you do that? You think? That's what's scary. Would you ever, would you ever know what is real anymore? I rather own my own double. Fair. I was 230. I'm here. I was still here. Go and sleep. Thanks for being here. I appreciate it. But you go to sleep. I wonder my living in a fake world. Okay. Does learning Python help me learn about AI? Absolutely. I actually have a video coming up about how to learn AI in 2024. Although my editor's on break. So I'm not trying to edit myself. Hopefully it looks okay. Yeah, absolutely. Like if Python is what you should learn in terms of coding wise, absolutely. If you're going to learn AI, I think the first step of learning AI is going to be just learning how to use Python in order to use the APIs, the models that are available. And then if you want to dive deeper into these models and understand them and even making them and find who need them, then you start learning more about machine learning the maths behind it and things like that.

That's how that's the approach that I think is the best approach in terms of being able to create things and also being able to learn more deeply about stuff. Early adopter to be smarter than others from comfort of my computer chair and a semi matrix like environment. There are already double gainers. Yes, there are. It's Python really that good though. It means the fairly easy language, not even a true coder programmer. Then very much so that's probably the easiest language for you to learn. I love it. Okay. So you like the just kind of like we chill and just talk about stuff and learn things. Check things out together. It's my first. Yes, we want more live stream. Okay. Chill, chill, containment, not that would be more education less drama streams. Okay, fair. So something like this, where we check out stuff together. Maybe would you guys like we could like go over certain topics like career and things like that. We could do that.

Can you recommend any channels for books for learning about AI, I love your live streams. That's going to be a new video. I'll link a lot of resources. I promise lonely octopus. Yes, lonely operatives teaches AI stuff as well. Becoming a January. The next cohort. Simple tricks using AI to help with general task. Yes, that's what I'm going to be making. Like definitely I want to do live stream on that like certain things that you can do using AI and actually applying it. All right, guys, I'm going to run. Go get food. I'm hungry. Thank you all so much for joining and I'll see you guys next time. Yes. Okay. Bye. Bye.

More episodes

More from Tina Huang

View all episodes →