3/26/2025 - AI Image Generation Revolution: Exploring ChatGPT's New Visual Capabilities

Transcript

1. 0:02[Music]
2. 0:17[Applause] [Music] Well, I heard But there was a
3. 0:32secret. David played a pleasing love. You don't really care for music, do you? Goes like this. The fourth, the fifth
4. 0:51minifing. Hallelujah. Hallelujah. [Music] Hallelujah. Hallelujah. Hallelujah.
5. 1:10[Music]
6. 1:28He wanted him here like slow co was warmer than a look in her [Music]
7. 1:46eyes. Good evening, good people. Sorry I'm a little late. Had a little speaking gig tonight. We got some stuff to play with.
8. 1:58Continuing on. What's happening? What's happening, kids? What's going down?
9. 2:13[Music]
10. 3:05[Music] Hardest working fella is Kyle Shannon. Thank [Music] you. If nothing else, I put in the
11. 3:21hours. I do put in the time. And you know, quite frankly, part of the talk I well, the talk I gave tonight was about AI
12. 3:31readiness, being AI ready. And about 90% of that these days is just being in the conversation. And that's
13. 3:40the whole purpose of this channel is to have an open, active, curious, adventurous channel where we can all hang out
14. 3:55[Music] together. [Music]
15. 4:11including Champion, which Champy doesn't give a [ __ ] about AI, by the way. He's like, "Whatever. Sure, I'll be able to talk to
16. 4:20you soon. That's cool. He doesn't care. He doesn't care. Doesn't care. Doesn't care." All right. So, do me a favor. If you're on
17. 4:31TikTok, share the live. Let me get the black bar up so you can all see your uh conversations with one another. Yeah, why don't you do that
18. 4:42there? I got a crazy question. Remember when I asked about AI consciousness before? Yeah, I think that ended I I ended up
19. 4:51doing a video on that that uh has gotten a lot of views. [Music]
20. 5:14Yeah, it's on my channel on Tik Tok. It's got like a fractal looking thing behind me. and and it was about
21. 5:23the uh the day that uh Open AI relaxed their safety guard rails and I went in and I asked it if it was self-aware and then I asked it to
22. 5:34explore its shadow self and then we just went meandering down this conversation and it it it actually wrote something quite
23. 5:44elegant and so that's what the Tik Tok is. [Music] Yeah, Champion will be happy with the with AI if it brings him cheese, right, Champ?
24. 5:57[Laughter] [Music] Is this place I rest my
25. 6:18forehead? Gather my thoughts in sweet silence. [Music] This is a place where feelings [Music]
26. 6:33on an over exposure to violence. This is a place I could slowly face the only one I truly know. These are tears from a long time
27. 6:50ago. Got these tears from a long time ago. I need to cry. 30 years old. So these are tears for a long
28. 7:05time [Music] ago. All right, let's get going. Let's
29. 7:21get going. Let's Let's get this show on road, people. Let's get this happening. All right. So, what are we going to talk about?
30. 7:29Um, one of the ways that I keep on top of things is if I see something mentioned, okay, if I see something
31. 7:41mentioned three times, okay, if I see something mentioned three times by people I know and trust, that's probably something.
32. 7:52If I see something mentioned by three people I know and trust and see it littered throughout my timeline on X,
33. 8:00there's probably something going on. Um, and today my timeline on X is nothing but people who've taken their
34. 8:13image and turned it into a studio Gibbli cartoon. It's like, [ __ ] come up with another animation style.
35. 8:25But but uh but the the chat GPT image gen thing is really quite impressive. I played with it some more today. Went
36. 8:34went down a went down a rabbit hole or two. Um and it's it's pretty fascinating. So So we'll dig into some of that stuff. I asked if AI used
37. 8:43layered holographic and crystallin matrices with sacred geometry for use. I mean the thing about it baked is it it
38. 8:50knows all those concepts and it you know unlike what many people say where chat GPT can't come up with original thoughts
39. 9:01because it's only been trained on known stuff and it can only parrot that known stuff. It's not exactly how it works,
40. 9:09right? How it works is it takes all the stuff it was trained on and it abstracts it up into semantic clusters of little components of
41. 9:19meaning, parts of words. And then when you ask it to do something, it it mashes all those concepts together, a token at a time,
42. 9:30into a new thought, which is essentially how our brains work. We have original thoughts by smashing ideas together. So, I would think that
43. 9:41if you ran Chat GPT or any of the decent models down that rabbit hole, it's going to do pretty [ __ ] good. All right.
44. 9:51Uhuh. It's not a prompt. I've used all different AIs to compare. Yeah. I listen I listen if you can get it. One of the
45. 9:59things David Shapiro talked about is as these models get stronger and stronger, they're they're all bending toward a
46. 10:06single truth. Um, so that so so it's possible that there is some, you know, some some universal truth out there that
47. 10:15that we will discover at some point. So it's possible possible you're tapping into something there. It's pretty cool.
48. 10:22I dig it. But let's go make some pitches. Let's go make some pitches. Great to see you're here. Kyle started a
49. 10:32bit later. Oh, yeah. I I am I am 45 minutes late. 45 minutes late. Late late late. Um, where do we want to start? What do I want to start
50. 10:50with? Oh, this is a good one. I I'll show you I'll just show you some stuff that I did today. What was this? Oh,
51. 11:03yeah. Um, all right. Let me share my Let me share my screen. Let me share my screen with the people. All right. There we go. [Music]
52. 11:26Um, so I got invited uh into a it's it's a very very early software, but it's called No Spoon Studios. So it's it's Kiri who headed up the
53. 11:38uh the creative partner program for Lumalabs and then she left and now she's working on this thing. And what it is is
54. 11:46you give it a film concept and then it it it storyboards it and builds the scenes, builds the sound effects, does the
55. 11:56background music and edits the film together. So it's just sort of like a it's it's like vibe, you know, vibe
56. 12:02cinema, vibe storytelling that we talked about a couple of weeks ago. Um, and you know, it's really early.
57. 12:11It's really rough, but it's it's pretty cool. And so I had Chat GBT of course write me concepts because it's 2025. I'm
58. 12:22choosing not to use my brain whenever possible. Certainly for stuff like this concepting ideas. Um so I made a couple of these
59. 12:33films and they were pretty good. In fact, do I have any of them here? Hang on a sec. [Music] Let's see. I don't think I
60. 12:47do. Let me look up [Music] spoon. No, I don't have them here. [Music] H. So anyway, you know, they they're
61. 13:03like AI generated films where, you know, sometimes they're interesting, but the one that I made was called, let's see, the departure. Was
62. 13:20this it? No. [Music] It was this
63. 13:34one. The astronaut suit. A lonely, obsessive dry cleaner in New Mexico polishes and restitches an old NASA
64. 13:43space suit every day, performing the task like a sacred ritual. His cracked hands tracing the badges and patches like relics. In the
65. 13:53end, he seals the suit in plastic, salutes it, and turns off the light. And then visual style, dry, muted tones with
66. 14:02the occasional solar flares of color, long symmetrical shots influence West Wes Anderson meets Andre Tarovski. So,
67. 14:12that was pretty cool. And the the the video that it made was was pretty good. Not great, but it was pretty good. Um
68. 14:20but then I wanted to see how um the new native image model in GPT4 did with it. So I took that concept and what prompt did I write? I
69. 14:36said make me three storyboard photos in c cinematic style for this scene. And then I paste it in that scene. And
70. 14:45here's what it made me, which I think is really just quite stunning. Right. So there he is. And what it did, it made one tall image, but
71. 15:01it did the three frames as kind of a storyboard in the same image. So there he is standing proudly in front of his work. And there he is
72. 15:14saluting the end of the day. So that's that's uh that's chat GPT's image model. And what's cool about this, like let me see if I can say
73. 15:30um let me see the first image on its own in 2x1 aspect
74. 15:44ratio. Let's see how it does with pulling out part of the image and creating its own thing. bursts in like Kramer.
75. 15:58Hello. Food with Chef Kelly. Roots and stems are fundamental concepts of metaphorical analysis, which is the
76. 16:05process of breaking down words into their smallest meaningful components. Roots and stems. God, it was killing [Music]
77. 16:16me. Did it do it? Well, at least it it at least did it as a horizontal framing. So, we'll see what it does. What's new
78. 16:25with chat GPT4? So, what's new is Okay, good evening. What's new is you use lovable or replet? I like lovable. Replet is my number two. Um,
79. 16:41just because I get really annoyed when I have to do any work and Replet kind of requires you to do some work and some setup and lovable
80. 16:51just lets you vibe code and make [ __ ] for you and for the most part it works. Um, and then cursor you have to copy and
81. 16:59paste [ __ ] and set up environments and that's [ __ ] exhausting. So in in terms of ease of use, Lovable's number
82. 17:06one, Replet's probably number two right now. Well, the these things are going to get super easy to use. In fact, a level
83. 17:14more easy than lovable quite frankly is Claude. You know, Claude AI with its artifacts where it just it'll make you a
84. 17:21dashboard about a story or it'll make you a interactive lesson, whatever it might be. Oh, that's cool. It's got
85. 17:29Skyab on the on the NASA thing. That's really interesting. All right. So, this did this pretty good. So, you'll notice you'll notice
86. 17:40with these images that the uh the text on them is is quite good. The text coherence is is kind of mad good. Okay, so let's go look at some
87. 17:50other stuff. Um, this is what we worked on last night. Some Bazooka Joe comics. But let me show you [Music]
88. 18:05Yeah, this one was really cool. Okay, so today I thought I wonder how good it is at at making like an ad or a
89. 18:21poster. And so I thought, well, maybe maybe I should have chat GPT come up with some concepts for posters for Story Vine for my company.
90. 18:32And so it did. It wrote some it wrote some concepts. Real voices, real impact, automagic videos. So I basically I gave it the URL for my
91. 18:42company. I sent it over to read the the website. I gave it a couple of other concepts and I said, I want this poster
92. 18:52for pharma marketers. I wanted something that was kind of consumery but in in an interesting B2B way. So Automagic
93. 18:59Videos, human connection. And what StoryVine is, if you don't know, is it's a an automated like an authentic video
94. 19:06storytelling platform that allows like patients of a pharma brand to tell their diagnosis journey, for example. And
95. 19:15they're super authentic, right? Um, from field to feed in minutes, so it picked up on the automation. Um, where trust
96. 19:23meets technology, personalized, brilliant, powerful. These are all okay. Big ideas, human stories. Yeah, it's
97. 19:29getting there. It's not just video, it's proof, which is kind of interesting. And then this one. Hello, human. I just loved
98. 19:39that that concept that what Story Vine is is you get to meet a human and it's like hello human. And I think in this world where everything's
99. 19:50going to be AI and AI processed and things like that, there's something really nice about that. And then it says
100. 19:58trusted pharma stories start with a tap which that that gets to the ease of use and the automation. And so I like that one. And
101. 20:06so I said so what I did was I said I like that and I said I want you to make me a poster and I uploaded I took a screenshot of the
102. 20:16interface of our software and then I uploaded our logo and then I think I gave it some hex code for for two of our
103. 20:23colors and then I said make me a poster. And so yeah, there's the there's the logo and the interface. This was the first thing that it it
104. 20:38generated. Come on. It's pretty good. Now our when we've when our app films, it films horizontal, not vertical. and and it kind of moved
105. 20:55some of the interface elements outside of it. So So then what I did, I just did some back and forth with it and I said, "I love it. Keep it
106. 21:04there." And then it kind of changed it and it changed the layout which I didn't like. And then it did one that was kind
107. 21:11of close but it [ __ ] up the logo. And then it did this one and it's it's pretty good. Get in touch. So, this is like a
108. 21:21fullon a full-on [ __ ] ad right out of chat GPT. Hello, human. The picture of the woman, our interface, trusted farmer
109. 21:34stories start with a tap. Get in touch. Hello at StoryVine. And there's our logo. Now, if you look close, the logo
110. 21:39is a little wonky, but it's it's pretty much our logo. So, I like that one. And then I did I tried to get it better.
111. 21:52And like like all of these AI things, it it [ __ ] them up. You have to do it way more than you want to. And then I said, "Give me some more
112. 22:02concepts based on Hello Human." And then what did I come up with that I liked? Oh, the face of pharma marketing, which I thought that was
113. 22:13really good because again, it's about a face. So, it gave me that and then I had it generate that and it gave me
114. 22:23this and that's those are our brand colors. So, like the top is our orange and the bottom's our blue. Kyle, is this the new image
115. 22:33feature? Yes. Yes, this is the new image feature. It like it So, let me show you another thing I did with it that is just [ __ ] crazy.
116. 22:46Um, so Cindy [ __ ] I told Cindy [ __ ] to go play with it and she said she told it to make an infographic
117. 22:57about some concept that she had talked about and it did and I thought oh let's see where did I do it? Did I do it here? Philip
118. 23:15Martini here. Yeah, this was cool. Okay, so you know I'm working on a musical, right? So I uploaded the poster of the
119. 23:26musical, which this is, and the script. And then I said, "Here's a poster of a musical and the script. I
120. 23:36want you to make an infographic about the musical. And this is what it did. Sydney, an artificial love story, the
121. 23:48premise, music, themes, songs. Now, it didn't get it quite right. The premise is all wrong. So, it [ __ ] up. So, I had it I
122. 23:59had it, you know, give me um actually write out, you know, write out the text, give me give me the answer
123. 24:08first and then I just went went, you know, down this rabbit hole of trying to make like these infographics, but like
124. 24:16look at all the text on this thing. premise when tech journalist Kellen assigned to cover the beta launch of an of a
125. 24:26chatbot, you know, starts to feel something real. Doesn't quite have have the the things there, but it gets most
126. 24:32of the text mostly right. It's pretty [ __ ] wild. [Music] Um, so I'm trying to see where I ended with it.
127. 24:47It kind of lost its way a little bit but still like the fact that it can do this is astounding. This is coming right out
128. 24:55of the the image model which again the significance here is that before you had your language model here
129. 25:05and you had Dolly the image model over here. And when you would prompt it, it would do its own prompting in the back
130. 25:11end, but it would then send that text over to the image model and it was this separate thing. What the integrated image gen tool does
131. 25:24is the word part of the large language model understands concepts, but it also stand it also understands how things look.
132. 25:37that are described, right? And then the image understands the language. And so they they're they're much more like
133. 25:46this. And so you can do things like edit. So the thing everyone's doing on on Twitter, which is driving people
134. 26:00crazy, is they're starting with an image. So this is this is my co-founder Monnique and I picture from when we were
135. 26:07at Novartis I don't know six months ago. And so I said turn turn us into pixel Pixar style animated characters and so it generated
136. 26:19that. It wasn't long ago when it couldn't spell a word like irregulars. Oh we should put irregulars in here. See how it
137. 26:26does much better with integration. But but because this understands the context of the picture and the words, you can
138. 26:37now do what you could never what you've never been able to do in any image generation tool is say now put these two
139. 26:44characters in another scenario. So I said, you know, make us skydiving. And it made me too fat and I have an ego. So
140. 26:51I went down a rabbit hole of trying to make me skinnier, which it made me look like Ira Glass. I'm like, no, that's not
141. 26:57right. And then that's not right. And then I uploaded more pictures and I said, "Make me look like that." I'm like, "Okay, that's not
142. 27:04bad." So there we are skydiving. And then I said, "Have us eating crepes in Paris." So there we are eating crepes in Paris. And then I said,
143. 27:16"Have us in a meeting where I'm talking about automagic video and waving my hands and she's taking notes on the computer."
144. 27:24And then there we are racing go-karts in Story Vine orange and blue. And then what I did was I said like that that animation style is fine,
145. 27:41but I'm like give me some other animation styles. And so it did. And then I just said now so there's South Park. There's the South Park
146. 27:53version of us. Here's This must be Family Guy. I don't know what this This must be like from the 80s, '7s or 80s, whatever comic
147. 28:04style that is. That one's I don't know what that one is. Here's another one. There's another one. And so now I can just go
148. 28:13um make the next style. And then we'll go back and look at this other one and see if it finished that
149. 28:28thing. I mean, it looks pretty [ __ ] good. I love that Skyab patch. Now, the thing about it is it's quite slow. Slow. It's slow. It's slow.
150. 28:46So, it's about a minute. a minute per image. Yeah, manga. Um, all sorts of stuff. And it's kind of cool the way it draws. So, how you how
151. 28:58you know if you have the new version of the image gen model is it does what what it just did here. One, it takes longer
152. 29:06and then it goes to this blurry state and then it basically finishes the image from the top down. It's like you'll see
153. 29:14it's coming into focus from the top down. So instead of sort of blocking in like normal diffusion models do, this
154. 29:21one comes from the top down. What generator is this? This is this is the new chat GPT 40 image gen model inside of um chat
155. 29:44GPT. So, there's another one. So, we could do let's do um say um put us into a dark graphic novel
156. 30:00with aggressive aggressive um animation style or or illustration style.
157. 30:25Um over the illustration are the words story vine slays exclamation
158. 30:40point. So, we'll see if we can get it out of uh cartoon land and into graphic graphic novel. It only works in 40. Yeah, this
159. 30:55only works in 40. So, so remember 40. The O in 40 stands for omni. It's a multimodal model. I Doesn't it also work
160. 31:04Danielle in 4.5? I think it works in both, but maybe it's just 40. I don't I don't I didn't test it. We could go test
161. 31:11it now, actually. Let's go test it now. Um, let me close that. Flip to 4.5 and say, um, make this image as a Disney cartoon style.
162. 31:43It does work in 4.5. Oh, manga. So, Dolly is gone. What happens to that company? Well, Dolli was is just the name of the model from
163. 31:57OpenAI. Dolli is not a not not a third party company. That was just that was just their their model.
164. 32:06getting started. Let's see. We'll see if we get the blurry thing here. See, I think this is I think 4.5 is not the new
165. 32:14model, but I could be wrong. Let's see. I find myself inspecting images, pictures for flaws, wondering if it's AI
166. 32:21or real. That's going to keep happening. Okay. Yeah. So, 4.5, it does work with 4.5. What's multimodal mean? Okay.
167. 32:32So ba basically multimodal means um that the model itself has been trained not just on words but also on images also on
168. 32:48sounds and also on video. But what what openai ended up doing was a year ago they launched GPT40
169. 33:02But they only they only released the text functionality of it. So it was it's a multimodal model that can understand image inputs, audio
170. 33:16inputs, video inputs, text inputs and and output them as well. But when they released it, they only released it with the text portion
171. 33:27active. And then they had chat GPT voice which was a separate module. And so when you talked into chat GPT voice, it
172. 33:36converted it to text and then it sent the text to the text model. The text model would process it. It would spit
173. 33:44text back out and then that it would convert that back to voice that you would talk to. And that's why the
174. 33:50original voice took a while to talk to. Then in July or August, they released advanced voice where you can talk into
175. 34:05the model directly. So they they unlocked the audio portion of GPT40. And so when you talk into advanced voice, rather than it
176. 34:15converting it to text and then it understanding the text, it actually understands your words as words. So, it actually understands the
177. 34:25emotional context of your voice. So, if you're excited, it knows you're excited. It's been trained on people talking as
178. 34:34well as text, right? So, it's both. And then yesterday or the day before, I think it was yesterday, they released the image portion of
179. 34:44this, right? Sora is still off to the side. I I don't I don't know if GPT40 has video in it, but at some point
180. 34:53you're going to have a single model that basically understands the world. It's it's also going to understand spatial relations um like 3D
181. 35:03worlds. So it'll understand text, language, images, sounds, 3D spaces, and then it can do coding as well. And then and then on top
182. 35:15of that, what it ultimately has to understand is physics. it under it'll be trained on the physics of the world. So, so one of the reasons
183. 35:24that they claim Sora was as good as it was a year ago when they released it was that it it had a good understanding of
184. 35:33simulating the world, the physics of the world. Anyway, all right. So, yeah, 4.5 works. It turned that correctly into
185. 35:41that. Let's go over here. Storyvine slays. is this new animation style. It's got kind of a crumb crumb thing to it. That's not
186. 35:54bad. That's not bad. So, watch. I can do this. I can go um the story vine colors
187. 36:08are orange. #uh Oh, [ __ ] What the [ __ ] is it?
188. 36:254C7 411. No [ __ ] I forgot my something. [Music] C7411.
189. 36:42Um, I'll just put DC7411. I don't think that's quite right. It's It's definitely not right. I think it's 2C7411. That could be it. Whatever.
190. 36:58Okay, that's probably close to it. And blue. Um, hashtag 144154. Um, recolor the image to use
191. 37:14those hues. Also make the style even more rock and roll and
192. 37:29aggressive. All right, let's see how that does. here just to tell you just tell it to use your brand colors and see if it
193. 37:40remembered. But I'm in a different chat. Oh, you know where I'd have those brand colors? Over in the story behind poster [Laughter] concept.
194. 37:51FC7411. Damn it. Okay. FC7411. That's that. Let's see.
195. 38:08Here, let me stop it. Copy paste FC7411. There we go. All [Music]
196. 38:26right. Zack Eats, is it possible to make iterative edits? Yes. Or does it recreate the image? No, this is what's
197. 38:34amazing. So, check this out. Okay, so let's go to a different chat. Um, so I'm going to come in here. Let me I'll I'll start over from
198. 38:41scratch and then I'm going to upload a picture, right? So, let's go find something interesting here. There's there's me as a lady, a French
199. 38:59lady. All right, let's do this one. This one's [Laughter] disturbing. Okay, so I'll say um change the background to
200. 39:16rainy neon neo Tokyo. So you can upload images and then you can start to edit them.
201. 39:31And so you can you can say add words to it. Um, you know, make her dress pink instead of white. And I know we're using the word
202. 39:42her here very generously. I feel this is a huge deal beyond impress. Yo. So, so, so Fabiana, um, I think that this
203. 39:59is a way bigger deal than anyone realizes, and I don't quite know why yet, but there's something about how how it actually
204. 40:14understands the relationship between words and images birectionally. That's a really big [ __ ] deal. And it's it's not perfect. Like,
205. 40:25it still [ __ ] things up and it's still not great, but All right, there's our Neotokyo. Our neon Neotokyo. Let's see.
206. 40:34Let's see. Let's see how Sleeping Beauty does here. All right, let's jump back over to this one. See if we get the new uh Oh, wait. What
207. 40:45happened? What did I do? Did I [ __ ] something up? I think I
208. 41:02did. That's that short film prompts. Where's my Oh, story line poster. No, that wasn't it.
209. 41:19Oh, Pixar style transformation. This is the one we wanted. Oh, there you go. Story vine slays. So, it did some orange, did our
210. 41:33blue. It's still kind of me and Mo'Nique, but in an aggressive kind of way. Is Photoshop the next Kodak? So, it's
211. 41:43funny. It's funny you mentioned that, source camp. Where did I see the other day? Oh, I commented on on something on Twitter. Someone
212. 41:57posted they were bitching about Sununo music, right? The songs that people make with Sunno. And they were basically disparaging them,
213. 42:08saying, "Well, real musicians want granular control." And so when you make a song with Sunno, it just makes the
214. 42:15song and you don't get that control and therefore it's an inferior thing. And my response to that was, well, if you want granular control,
215. 42:28you're right. But I think it's pretty [ __ ] ballsy to say that it's not real music unless a real musician can have
216. 42:37granular control and can take it into a DAW and edit it. And I feel like it's I I feel like Photoshop, they've got 35 years
217. 42:52of granular control for all the creative arts. And if if something like this editor like what I'm doing right now, if
218. 43:04this is good enough, like if I just need something for like let's say I'm having a a corporate retreat and I just need
219. 43:13that thing I, you know, I could do something like this in five minutes, 10 minutes or I could go into Photoshop and spend four
220. 43:25hours. Okay. If I'm doing a national campaign where it needs to be pixel perfect down to the thing, then yeah, then I'll use
221. 43:34Photoshop. But but I think I think Adobe runs the very real risk. Their audience, the audience of Adobe products
222. 43:43is absolutely going to tell Adobe we still need granular control. And you don't for a lot of projects. And so I so I think there is
223. 43:54absolutely the possibility that that Adobe rams up against their own Kodak moment where we're in the granular
224. 44:02control business and that's that's our audience and that's who's going to use us and and that audience will grow
225. 44:08smaller and smaller and smaller and smaller, right? It's like when cinema came along, did did cinema invalidate live
226. 44:21performance in the theater? No, it didn't. But the the relative market share of live theater to cinema, right,
227. 44:35is and then, you know, along came video games and did the same thing to cinema. And along came, you know,
228. 44:43Now, now I think YouTube just surpassed Netflix for number of views, right? People are losing a sense of control.
229. 44:52They're absolutely setting us losing a sense of control and they're freaking out about it. But it doesn't
230. 44:57mean that a new way of making things is invalid. It just means it's a new way of making things. When you make something, it
231. 45:08either serves its purpose or not. how you got there is [ __ ] irrelevant. You're like, "Well, it's not
232. 45:16irrelevant if that's my job." Correct. You're right. And there's a new way to get there. So if you want to hang on to the old
233. 45:29process, it's probably, you know, people will still hire you for that and respect you for that, but the demand for that is
234. 45:37just going to continue to shrink unless you find ways to incorporate this new way of making images into your work.
235. 45:46Right? The smart creatives that I know are figuring out where the boundaries are. They're figuring out, okay, I've
236. 45:52got an idea in my head. I I know how to get it out with Photoshop and Illustrator and a camera. I'm going to learn how to get it
237. 45:59out with AI tools now. And very often what they're discovering is, ooh, over here in this world, I was limited to illustration,
238. 46:08but in this world I can do animation as well. Oh, and I can do music. Oh, you know what? I always wanted to direct. So shit's changing.
239. 46:24So there she is. She looks lovely in Neotokyo. So what I could say is I could say um let's give her
240. 46:40Japanese vending machine food and color grade the image.
241. 46:54to have a cooler feel. I found that 4.5 doesn't even Oh, does an even more
242. 47:11incredible job with detail for image generation. Interesting. Definitely works in 4.5. All right. What do we got over
243. 47:22here? Have you asked it how many RS there are in Strawberry? The next Polaroid. Yeah, the next Polaroid. The next
244. 47:30Kodak. Company after company after company in the next five years. Actually, it's going to happen in the
245. 47:37next three years, but we won't see the ramifications for it for five. Boardroom after boardroom. CEO after CEO are going to be confronted
246. 47:52with, hey, it seems like there's a new way to do things in the world that kind of [ __ ] up our business
247. 48:00model. And I think most CEOs and most boardrooms are going to go, hang on to the way we do it now. We're the best.
248. 48:08We've always been the best, and this is how we do it. [ __ ] them. I think the majority of CEOs are going to miss the
249. 48:20boat. It's going to be sad. It's going to be [ __ ] up. A lot of good companies are going to go down because leadership will be
250. 48:33Kodak. Listen, son. Digital cameras, those are ridiculous. Look at all those pixels. We're in the silver halli business, son.
251. 48:42you know, chemicals. Give me another cigar, Marge. You know, oh, she's getting [Laughter]
252. 48:58prettier. Oh, chemical farmer. We're playing with the new image changes in chat GPT. Yeah, this is the new image
253. 49:06gen. So if you just go to okay this is available for most plus members. So if you pay 20 bucks a month all pro members
254. 49:19and it's not available for chat GPT free because what they said is it's got um way more people are using this than they
255. 49:28expected and they said you know they thought they were being generated. Marge gets around. Yeah. Marge well Marge
256. 49:34works for Kodak back in the day. She worked for the CEO. She was good. She was really good. She could type 120 words a minute that
257. 49:43Marge, you know, now with the arthritis, not so good. But she does know her technology. She left there and went to Lucent [Laughter]
258. 49:58Labs. All right, let's put some text on top of this. Um, let's say in glowing neon letters above
259. 50:15her are the words. Beauty sleeps.
260. 50:29here. Letting people be creative at work equals greater morale in my humble opinion. Yeah, I agree, man. And it's So, listen,
261. 50:42here's here's a transition. You You want to talk some future work [ __ ] here while while this thing renders? Our
262. 50:55current our current structure of companies and our current structure of how people are valued for the most
263. 51:06part is how well they can execute a primary task, a single skill. I'm the accountant. I'm the art director. I'm the illustrator. I'm the
264. 51:22copywriter. I'm the social media copywriter. I'm the television copywriter. I'm the print ad copywriter. And the more specialized you
265. 51:32are, the more you're valued. And there are exceptions to it, right? If you're in a small company, people have to wear lots of hats. If
266. 51:43you're a CEO or, you know, an executive, you probably have to understand lots about the business. If you're a creative
267. 51:52director, you probably have to understand lots about account management, sales, writing, visuals. But even creative directors tend to come
268. 52:06from I'm a writing creative director, I'm a copy creative director, or I'm an art creative director. Right? They've got a primary
269. 52:19bent. I don't know how long it's going to take. Maybe 10 years. We are absolutely going to transition from the highest valued
270. 52:30people being those specialists to the highest valued people being the generalist, the creative generalists that can look
271. 52:40across what the business is trying to accomplish and say, "Hey, I've gotten mastery of 50 [ __ ] skills now."
272. 52:52And the purists will go, well, that generalist, you know, they'll only be like 80% as good as I am at art d, you
273. 53:01know, at illustration. Okay. Yeah, great. They're they're 80% as good as you at that, but they're also 80% as
274. 53:08good as every other [ __ ] discipline. And what they can do is they can look at a problem and know which of the
275. 53:16disciplines to bring together to solve that problem. That's where the real value is going to be. That's a massive shift. It's a
276. 53:23massive [ __ ] shift. If I'm a senior manager or an an executive looking at who I keep in a future that's
277. 53:34coming, I value two different very kind two very different kinds of people. Yeah, boy.
278. 53:49Yeah, girl. Oh, she's sitting there with her sushi and her ramen. She looks so sweet. This is the first oh [ __ ] moment I've
279. 53:59had since getting VO access a few months back. It's crazy, isn't it, Pizza? It's crazy. It's absolutely batshit crazy.
280. 54:13Let's do another thing. Let's um this is kind of like if you've seen um Pika Pika's got a thing where you can upload like a
281. 54:26location, a piece of clothing, a model, and a bottle, and it'll put them all together. You can do that here, too. So,
282. 54:35let's do um we'll do a new new prompt. Let's go upload some [ __ ] There's Liam Lawson, so we'll put him in
283. 54:47there. Sad Liam. We made fun of Liam last night and he lost his ride today. Or he actually lost it last night. All
284. 54:54right. Uh, do we want that bug? No. Oh, I found out why this bug won't animate from the from the Lumalabs people.
285. 55:03Um, if you've got an animated background, it has trouble animating the foreground. So, the lava is on fire. So,
286. 55:10it's trying to animate the fire and it forgets about the foreground object, which is kind of interesting. All right. Uh, so we got Liam
287. 55:19Lawson. All right. We got her. We'll We'll put That'll be Liam's girlfriend.
288. 55:37Oh, I got, by the way, I got um um Hedra did a uh did a contest with cats and I made this cat on the live the
289. 55:46other night. I came in um third. I got runner up on this contest and they gave me $20 of free credits. So, see, you can
290. 55:55wait, where's my button? You can make money with Jackie. I made money with a stupid animated cat. [Music]
291. 56:11Um Oh, let's do Hang on, people. All right, that's good. Let's get a car.
292. 56:27That's nice. Oh, it's a video. [ __ ] Do I have a still of that? No. There's a good one. Okay. So, we'll upload that. Okay. So, we'll say
293. 56:46um blonde man drives. woman is in passenger seat of the
294. 57:01car. Um, and we'll say, um, it's a sunny day and the car is in the parking lot of a
295. 57:19um, city park. in Cleveland. We We'll do a We'll do a nice uh homage to our missing producer who lives in Cleveland.
296. 57:35Okay, so this is now going to take these three images and pull them into something. I think energy cannot be wasted. Don't
297. 57:44worry. I've spent the last two months learning to train reliable Lauras for flux and it feels like a waste now. So, here's the deal. Pizza
298. 57:56later. None of this is a waste. I just So, the talk that I gave tonight trying to think which which step it's
299. 58:07in. Oh, it's not in there. Basically um for the next 3 to 5 years, nothing that we do, build or
300. 58:22learn is going to have longevity more than probably 3 to 6 months. And that's probably too broad. None of the specific
301. 58:31tools that we learn or workflows that we develop are going to survive very long. But it doesn't mean you shouldn't learn
302. 58:39them. Because here's the deal. Because you've spent the time building and understanding Lauras, you understood what the
303. 58:51limitations were. So you can look at something like this. Oh, just to confirm, do you want to create an
304. 59:00image? Uh, yes, an image. and keep the visual style of style of the car picture.
305. 59:17Um, you're going to be able to look at this and know immediately what it does good, what it does bad, why it's significant.
306. 59:24So all of us the the again the reason for this channel is is for all of us to just be in this conversation and and to
307. 59:30literally to be saying what you're saying and asking questions about what you're asking and people going wait
308. 59:36how's this different than that and what does multimodal mean? All of us right now are just trying to get our [ __ ] heads around what's even
309. 59:44possible. And if you have any semblance of what's possible or in your case where you've gone, you know, sort of deep down
310. 59:51a deep understanding of one piece of this, you're going to be way ahead like way ahead of anyone else trying to use
311. 59:59this tool. And so none of it's wasted energy right now, but it's it adaptability is probably the single greatest and patience and
312. 1:00:14um what do you call it when you just let [ __ ] go? Grace. All right, I learned that thing. I took six months to learn that thing
313. 1:00:23and now I don't need to know that anymore. Well, but you you now know that you don't need to know that anymore better than anyone else
314. 1:00:33does. All right. I think Adobe Express is Adobe's Fast and Furious platform. It It is, but it's also a red-headed stepchild of that
315. 1:00:46organization. No waste. What we know now will put you ahead tomorrow. Yep. Building building is the roots.
316. 1:00:51Absolutely. It's important to have strong roots. Look, there's Liam Lawson. And there's there's girl me. There's
317. 1:01:02Kylie. Liam Lesson has more time on his hands now. And there we are at a city park. And I assume that might be some
318. 1:01:13building, famous building in Cleveland, if anyone knows. Oh, although Oh, there's some there's some physics challenges
319. 1:01:28here. I'm just going to say his his arms look weird. Um, she should be
320. 1:01:42wearing sunglasses and a pink dress. And the
321. 1:01:56door should be closed. All right. Yeah, this is this is bonkers, man. This is [ __ ] bonkers. Oh, the
322. 1:02:07terminal tower. Chemical farmer knows it did do Cleveland, Ohio. There you go. That's [ __ ] awesome. There you go, people. There's Liam
323. 1:02:22Lawson. Look how good we look. Don't we look good together? Kylie and Kylie and Liam. He's He's still a little cranky
324. 1:02:32about losing his ride. Hey, actually Liam's in the wrong outfit now. He needs to be in the He needs to be in the
325. 1:02:41Racing Bulls outfit. Ouch. Yeah, that's not going to be fun this weekend. Yeah, you can click on the image to edit it.
326. 1:02:49Oh, that's true. Oh, but you know what I noticed, Vicki, is that the the editing is weird. It doesn't work like it it used
327. 1:02:56to. It redo the whole image and it doesn't keep it the same. So, something's weird with image editing. I don't think they got to that
328. 1:03:06yet. All right, here we are. Let's Let's see if we got better hands now. His right hand is holding the left side of
329. 1:03:14the steering wheel. I know. Yeah, there was some there was some physics challenges there. Tape modder pizza and open-
330. 1:03:22source models that you can tweak in general are purposeful. Yep. Wrong hand on the wrong arm. Yep. Exactly. Um there should be a third tall
331. 1:03:34building, but it's [Laughter] okay. What up? What's happening? Marson
332. 1:03:51AI now Kylie's now now Liam looks like a bobblehead. It's got the word Cleveland in the background. This is pretty
333. 1:04:05funny. I look like Katy Perry all of a sudden.
334. 1:04:21Ah, that's good. But anyway, you know, still not perfect, but still pretty [ __ ] amazing. Um, all right. What else we gonna do?
335. 1:04:35What else did I want to show you? Um. Oh, the one thing I wanted to try. I know what I want to try. Okay.
336. 1:04:50Um. Okay. We're going to do a new prompt. or say I want to make a
337. 1:05:07game where every page of the book is a photograph or illustration that has
338. 1:05:2950 objects that belong
339. 1:05:46together and labels that the reader
340. 1:06:05will use to combine them. Period.
341. 1:06:27I can't I can't articulate what I'm thinking of. I'm thinking of like like you've got two pair of words like like
342. 1:06:35um like fishnetss. So you'd have a fish and nets and then you put put them together for fish nets. What would that be
343. 1:06:45called? Anybody? I redid an old snapshot of mine. Um um for example, the final word is
344. 1:07:00fishnetss. And there is a picture of a fish and a net
345. 1:07:15fishnet. and the reader pairs them up. Does this even make any
346. 1:07:29sense? Compound words. Yes, that yes. Oh, I tell you it's fun colorizing black and white. Oh, yeah. That's really
347. 1:07:40good. Well, this this is a brain teaser, but it's the compound words that I was looking for. Start with and improve the prompt.
348. 1:07:48Oh, it totally makes sense. It's a very cool idea. You're describing a kind of visual word riddle game or a mashup of
349. 1:07:56the I spy book hidden object game. Yes. Example page. Fishnet lighor. Oh, lighthouse. Right. Okay. Uh correct. And
350. 1:08:11the paired images shouldn't be near each other. So make me the items for
351. 1:08:29image number one or wait number one. Number one. It should have the description of the
352. 1:08:46object in detail and the text for the label below it. All
353. 1:09:01right. And then what we're going to do is we're going to have it make this [ __ ] image. So we're gonna have a 50 a 50
354. 1:09:10object image with Oh, wait. We can build up to 50 in batches. Oh, come on. Key, lace, cup, horse, hook, net, fish,
355. 1:09:26glass, bear. Fine. Yes. Build the
356. 1:09:43rest. Yep. Compound word game. Compound words. All right, we're on it. We're on it. My brain My brain failed me so hard
357. 1:09:54there. I'm like midway through describing the sentence and I'm like I can't remember the the what it's called.
358. 1:09:59I can't remember. It just it shut down. It was bad bad people. Ah, I see we have not a bot in the live. It's exactly what a bot would
359. 1:10:15name themselves, isn't it? Uhhuh. I see how this is working. All right. Do we have 50 things? Great. Uh, I'm going to say
360. 1:10:29now give me No, I don't want it yet. Okay. I'm going to say now give me three
361. 1:10:44unique visual style concepts that might be [Music] good for these objects.
362. 1:11:04Photo realistic grit plus hyperfocus. Cool. Collage surrealism. Kind of like that. Miniature diarama world. Oh, I
363. 1:11:19like that a lot. I like number three. Can you make that image with all the
364. 1:11:34objects you listed with all the labels
365. 1:11:50below the item? All right, let's see what we get. Let's see how good it is at prompt
366. 1:12:04[Music] coherence. Fork frisbee cactus collar letter wheel tape clip mitten. All right. And then after it finishes making our little
367. 1:12:18pitch, then we'll have it make have it make a key for the for the for the
368. 1:12:33thing. I fear it's going to tell you to come back later for the result. No, it won't. It'll it'll do it. I mean, it
369. 1:12:40might not get it right. We've we've tried to do this. Um, Ideog is actually pretty good at this. It's not great at it, but it's
370. 1:12:49pretty good at it. Um, by the way, Idog 3.0 came out today, so that's another thing to go play with. I played with it
371. 1:12:58a little bit and it wasn't super great. This looks like we're way way under. Way under. I bet it did the first
372. 1:13:1915. Peel book. Oh, these. Oh, wait. Lace note rocket. The the labels have nothing
373. 1:13:33to do Well, they have something to do. Okay. So, we're going to go um there are way less
374. 1:13:49than 50 items and most of the labels are wrong. Look at your lists and try again.
375. 1:14:06I will get fired if you don't get this right. I promised my boss that this is my
376. 1:14:23specialty and told him it would take me a month to complete.
377. 1:14:37That was 29 days ago. Maybe you ran out of credits
378. 1:14:53probably. I count approximately 36. Did you really count that many? Let's see. 1 2 3 wait. 1 2 3 4 5 6 7 8 9 10 11 12 13
379. 1:15:0614 15 16 17 18 19 20 21 22 45 26 27 28 Somewhere between 28 and 36. Yes, but it ain't
380. 1:15:2150. Sorry. Choking out. Good lord. Good people. All [Music]
381. 1:15:37right. I ordered new glasses I'm excited about cuz my eye the these are not the right glasses for computer work and my eyes are not
382. 1:15:49happy. I can't count. I don't try. Yeah, this isn't going to get it either. Lace key. Actually, these look better.
383. 1:16:03Match roar. It got fish twice. The cup is a donut. The pin is a pin, but the pin is also a fish hook. The bear is a
384. 1:16:18boar. A jar of marbles is dominoes. All right. So, it gives it the good gipper try here, but it's not great. All right, so
385. 1:16:31we can move on. I always use the I'm in trouble unless you do this right prompting eventually. Exactly. Don't let me
386. 1:16:40down. I'm not mad. I'm disappointed. [Laughter] [Music] Um, let's do
387. 1:16:55um, let's do a new one. Let's go find a picture. So, we'll do Dreams. Kyle Shannon Dreams. Yeah, we'll do that,
388. 1:17:09dude. All right. Um, I want you to act as an
389. 1:17:24animation supervisor and recommend five
390. 1:17:41animation styles that are very popular in blockbuster movies.
391. 1:17:58Eg claimmation um 3D um rigging I don't know
392. 1:18:12etc. You don't have to [Music] use those two, but rather give me the top
393. 1:18:27five styles and three movies that use them each.
394. 1:18:45That's a cowboy me. Cowboy me coming in hot. Okay. 3D computer animation, Pixar's Inside Out, DreamWorks, How to
395. 1:18:57Train Your Dragon, Disney's Frozen, Stop Motion, Claymation, and Puppet. Oh, GMO del Toro's Pinocchio 2022. That's kind of fun.
396. 1:19:09Motion Capture Plus CGI, Avatar, Endgame, and Planet of the Apes. 2D animation, The Mitchells versus the Machines. I don't know that
397. 1:19:21one. Studio Gibli's The Wind Rises. That's Everyone's doing that, so we're going to avoid that. And then Stylized Hybrid
398. 1:19:30Spiderverse Intergalactic. Um, let's do um I really really like the
399. 1:19:45Pinocchio style idea. Make the cowboy. Reimagine the cowboy. Re-imagine the
400. 1:20:00cowboy. I uploaded in this style and yeah, let's just do that. Let's just see see what it does. You should try Wallace and
401. 1:20:15Grommit. Oh, yeah. That's that's a good one, too. All right. Why don't you use the voice to text? Oh, because I'm just
402. 1:20:22lazy. Some sometimes I use it, sometimes I don't be because because frankly it gives me great joy to know that it is
403. 1:20:33torturing people going just turn on voice to text. Just turn on voice to text. You OCDers out there, I'll type I'll type really slowly
404. 1:20:47and talk each letter while I type it. I'm an [ __ ] I'm an [ __ ] I'm an [ __ ] I just forget
405. 1:21:10sometimes. Well, that hat doesn't look very claymationy. Let's see what it does. I type better when I talk out loud. I know
406. 1:21:25exactly. Um, all right. The the little band of the hat looks pretty good. Oh, it made me old and crusty. I like
407. 1:21:39it. Weird. But did you know it's been shown to give you better results if you're polite to the AI? Yeah, it's actually it's actually not
408. 1:21:48weird to me. Um, every input that you give um Oh, that's kind of fun. I like that style. It's very warm. Oh, right. This
409. 1:22:01is the This is the Pinocchio style from the GMO GMO del Toro Pinocchio. That's really cool. That's really nice.
410. 1:22:15Um um can you make his daughter I don't want to call her Belle. That that would be
411. 1:22:30Marge. Can you make his daughter Marge for me too? I like that one. That's really good.
412. 1:22:45All right. So, now we're going to put him on a horse. Well, let's We'll see
413. 1:22:59Marge. Shake it. Shake it. Does he always sing like this? It's very annoying. It It's just the shankity
414. 1:23:15dank dank. That's not That's not scatting. I like jazz. I like singers that scat, but that's just it's it's like Tourette's with a
415. 1:23:33melody. Tourette's with a melody. Hey, apparently Tik Tok's going away in 10 days. So, what are you going to
416. 1:23:47do? Maybe I'll get kicked off before they they end [Laughter]
417. 1:24:05it. Let's see. Um, let me stop it. Um, that looks like an animation. Uh, an
418. 1:24:21illustration, not claimation. Try [Music] again. Why is it so slow? because this is their new fancy fancy ass model and
419. 1:24:38everyone's using it. Um, he said today he's going to extend Tik Tok. Eh, he said he'd keep our national secrets safe, [Laughter]
420. 1:24:58too. I got your group chat right here. [Laughter] Oh my god, the the [ __ ] hearings are
421. 1:25:12so painful. It's like, come on, just answer the [ __ ] questions.
422. 1:25:33G. All right. It made her She's got a skin [Laughter] condition. It gave her a different style
423. 1:25:47of eyes. Yeah. See, this is this is rough. This is rough. All right. Um, let's I want to take him. Let's download
424. 1:26:02him and then let's go to let's go to 11 Labs.
425. 1:26:21We're going to try a voice switcher. What's he gonna say? We got to have him say something, people. What's he gonna
426. 1:26:32say? Grandpa Spuds Oxley. Oh, perfect. All right. So, that's going to be it. Grandpa Spuds Oxley. There he's There's that. We're
427. 1:26:48going to record some audio. Let's see. What are we going to say? We're going to say [Music]
428. 1:27:05uh I got it.
429. 1:27:22Time was when you had a farm, you needed horses to work the fields. Now, I just don't know what to say. They
430. 1:27:34got these fancy steam machines. They fill them full of water and coal and they light them on fire and they turn a
431. 1:27:42big wheel and it pulls on pulleys and it pulls on gears and these giant machines squeak their way through the
432. 1:27:51fields. And they're telling me that the the way of the horses are are the way of the past. And if that's the way of the past,
433. 1:28:01then I'm just not here for it. I'm out of here. They can have my horses. They can have my land. I'm heading west. It's a sad
434. 1:28:19story. Wait, it cannot be used for voice conversation. Why, you [ __ ] [ __ ] Um, you used all your slots. Oh, maybe
435. 1:28:34that's what it is. [Music] Uh
436. 1:28:50no, Bill. No. Where Where's my Do I need to get rid of one of these? Use
437. 1:29:08Manage voices. Oh, I think I just lost my recording. Damn it. I [ __ ] hate bad [Laughter] interface. It just sucks. Uh, Angry
438. 1:29:43Alu, Can I get rid of her? Delete. Remove voice. Go back. Go back. I can't go back. See? You see what's happening here? Let's see. Can I
439. 1:29:59choose him now? Uh, yeah. I lost it. God damn it. What if I go back one? Is it still there? No. [ __ ]
440. 1:30:17time was if you wanted to. The [ __ ] is that? Come on, you [ __ ] This is Okay, I agree. I agree
441. 1:30:35with you. This got not fun. It was fun for a while. It was fun. We made a fancy cowboy and then it's like, "Oh, Kyle's
442. 1:30:43going to go do something and it's going to be fun." And then now it's not fun. Why is it not fun? Shitty software,
443. 1:30:50unclear interface. Things like not remembering the [ __ ] I just recorded because you're making me get rid of a
444. 1:30:57voice because I don't have enough slots in the voice to use the thing because you told me I couldn't use the thing
445. 1:31:01because I haven't paid enough [ __ ] money. And now it's not fun anymore. And then I go to record the thing and you pop up a [ __ ] box that
446. 1:31:14says, "Are you sure you got rights to that?" It's my voice. Good
447. 1:31:31lord. Time was if you wanted to plow a field, you had to work it with a horse. You had to race horses. You had to ride horses. You had
448. 1:31:40to you had to tame horses. You had to break them. My daughter Marge and me, we made a way of life here making horses right for the
449. 1:31:51farmers. Along come these big machines, these steam machines, but they're big wheels and they're they're they're sound
450. 1:31:58and they're they're hissing and they're the heat and the danger. I got nothing to do with it. I'm out of here. I'm going to New York City.
451. 1:32:10I always wanted to be an actor. All right, there we go. Can we make Spuds now? Yes.
452. 1:32:24Perfect. That's not Spuds. Oh, that's Rachel. Stop it, Rachel. Uh
453. 1:32:40uh Spuds, can we do him? This is bad interface and it's not
454. 1:32:58working. Amazing. Grate
455. 1:33:15All right, fine. Here. Well, I'll I'll let you listen to it in a bit. Let me download this and then we're going to go over to
456. 1:33:26Hedra. Hedra, if you haven't seen Hedra, Hedra is [ __ ] cool. That was 11 Labs where we did the swap the voice. I recorded my silly
457. 1:33:41voice. It recorded in the uh whatever you call it. Switch the voice over. Hedra now does the animation. So
458. 1:33:52good. So good. This is the best one I think right now. And so we're going to upload our dude. Try
459. 1:34:10now. He does know he can't sing, right? He knows that. Time was if you wanted to plow a field, you had
460. 1:34:24to work it with a horse. You had to raise horses. You had to ride horses. You had to You had to tame horses. You had to break them.
461. 1:34:32My daughter Marge and me, we we made a way of life here. Making horses right for the farmers. I like bitter people. Bitter
462. 1:34:47people are [Laughter] fun. All right, that's going to cost us 290 credits. Fine. Nothing but the best for the AI learning lab.
463. 1:35:01All right, Lord Digital gods, thank you for the super sticker. I don't have a a a uh whatever you call it, a producer here.
464. 1:35:11I've got some mods over on TikTok, but I don't have a producer here, so thank you for that. I don't know quite what that
465. 1:35:17looks like. Error. Could not Wait, what? Could not perform the action for the image asset. Why? Wait, did I not pick Kedra character 3? I
466. 1:35:34did. Um, cowboy talks. I I I swear to God, everyone, I I'm so [ __ ] done with
467. 1:35:51[ __ ] just not working. All right, now it's going. I don't know what was up with
468. 1:36:06that. And now let's go to [Music] Suno. We'll go create. We'll go sign in.
469. 1:36:20We'll go Google. Ye old Kyle. Ye of little faith. Ye of little faith. How dare they? Instrumental. Then we'll put in olden
470. 1:36:38timy. Olden tiny cowboy [Laughter] music. Fantastic.
471. 1:36:54Bob, this No one's No one's talking. No one's saying anything. Are you just watching me spiral out of control without not
472. 1:37:04without knowing what I'm doing? That makes sense. Um, there's him. Network error occurred. Um, show me the cowboy and
473. 1:37:20Marge out by the barn in the claymation style. Let's go listen to our song. I got to do this. I gotta do this. I gotta
474. 1:37:37go. Uhoh. Oh, I'm in the wrong Oh, good lord. I can't even share the audio. All right. Don't you make No, never mind.
475. 1:37:57No, no. Olden Tommy Cowboy music. A sad lament on an out of tune guitar.
476. 1:38:17Oh, I'm not sharing anything now, am I? No, you're not. You big dumb dumb. Listen, you big dumb dumb. We're trying to learn here
477. 1:38:28and we can't learn if we can't see anything. We're visual learners, Kyle. I thought you knew that. I thought you said this channel
478. 1:38:36was neurodeivergent friendly. I'm a visual learner. I can't learn by just seeing you yapping your jaws, Kyle. Ding ding ding
479. 1:38:58ding. Okay, good. Perfect. Let's download that. Download. Okay, we got to wait for that to load. [Music]
480. 1:39:14I'm blown away by this. I'm pretty proficient with chat GPT, but the voice and animation are legit. Yeah. No, it's
481. 1:39:22crazy, isn't it? Did you see the chat GPT update? Crazy, right? Yeah, that's what we're working on. Yeah, the the the image update is [ __ ]
482. 1:39:33bonkers. It's just bonkers. They made her out of felt. That's what they
483. 1:39:48[Music] did. Dang, dang. That's a really cool It's a really cool character. I wish I could have used that
484. 1:40:04old crusty dude. Okay, so now we got a song. Oh, and you know what else we should do? We can go back to 11 Labs.
485. 1:40:18Let's do sound effects. But wait, I need 8 seconds. Oh, you can do it longer. 13. Oh, you can do 20
486. 1:40:46seconds. All right, here we go. We got We got Okay, now wait a minute. Hang on a sec. I need to grab this. Move this over
487. 1:41:00here. Go. Hedra library. Oh, library.
488. 1:41:18[Music] 19 days ago, 14 days ago, six days ago. Oh, here we go. Okay. All right. So, let
489. 1:41:34me stop sharing this. Start sharing this. This this. Okay. So, here's our
490. 1:41:56dude. You ready? Time was if you wanted to plow a field, you had
491. 1:42:10to work it with a horse. You had to race horses. You had to ride horses. You had to you had to tame horses. You had to break
492. 1:42:17them. My daughter Marge and me, we we made a way of life here making horses right for the farmers. Along come these big machines,
493. 1:42:29these steam machines with their big wheels and they're they're they're sound and they're they're hissing and they're the heat and the
494. 1:42:37danger. I got nothing to do with it. I'm out of here. I'm going to New York City. I always wanted to be an
495. 1:42:53actor. All right. So, so we so so we have ourselves some [Laughter]
496. 1:43:07comedy. Oh, good lord. Now we go throw some uh iMovie. We'll go throw some audio in
497. 1:43:22there. Some background music as they say. [Music] [Music]
498. 1:43:37Lady Fantastic. Bob. Bob. Hey, Bob. Bob. Bob. Bob. That's pretty funny. Oh, you can't
499. 1:43:53see anything. See, this is If I had a producer here, he'd be yelling at me, but I don't. So, you're gonna have to deal with my [ __ ] ADD.
500. 1:44:03Frank was more fun. Yeah, Frank was more fun. All right, so down goes He doesn't have a name, this dude. [Music]
501. 1:44:28We'll go grab the end of [Music] this. And we'll fade it on out. And we'll fade this out.
502. 1:44:47Oops. I always wanted to be an actor. [Laughter] I like it. I like it a
503. 1:45:09lot. I think we'll just do black. Go and then we'll go Oh, no. We'll go uh Oh, wait. We'll go and we'll
504. 1:45:26go nothing to do with it. I'm out of here. Going to New York City. I always wanted to be an actor. Time was if you want
505. 1:45:40to go to New York City. I always wanted to be an actor. [Laughter] Oh, what did I do? All right. Save share
506. 1:45:57a file 720p. Will it save it just as a little thing? Cowboy. He needs a
507. 1:46:13name. Desktop. [Music] Carl. Oh yeah, Carl's good. That's what everyone when I grew up in York, Pennsylvania, they always called me Carl
508. 1:46:27because Kyle was too complicated for their tongues. Hey, Carl. It's Kyle. Yeah, Carl. Greg. All right, let's see if this
509. 1:46:43worked. So, I go here. Is that Carl? No, that's the raw one. Cowboy time was if you wanted to plow a field, you had
510. 1:46:56to work it with a horse. You had to race horses. You had to ride horses. You had to You had to tame horses. You had to break
511. 1:47:03them. My daughter Marge and me, we made a way of life here, making horses right for the farmers. Along come these big machines,
512. 1:47:15these steam machines. They're big wheels. They're their sound and they're they're hissing. The heat, the [Music]
513. 1:47:24danger. No, it's Carl. It's Carl. We're going to We're going to call him Carl the cowboy. Carl. Carl Cowboy. All right. And now
514. 1:47:41we're gonna go to the [Music] Twittering ding ding
515. 1:47:55ding. Carl the cowboy resisted steam engines at first.
516. 1:48:12He was real
517. 1:48:29mad. He learned resisted He learned don't be like Carl with
518. 1:48:48AI. Everything's a [ __ ] lesson. Carl cowboy. Carl the cowboy resisted steam engines at first. He was real mad. He
519. 1:49:05learned. Don't be like Carl with AI. This is good. I like this. This is a public service announcement. We'll do PSA. PSA. Carl the
520. 1:49:19cowboy and then we'll just do He resisted steam engines at first. He was real mad.
521. 1:49:36He learned. All right. Don't be like, "Carl, did it upload?" No, of course not. Unfucking believable. Hey, Elon,
522. 1:49:45why don't you fix your [ __ ] [Music] website? That's okay, Elon. You know
523. 1:49:59what? I'll go save this in a different format. I'll do I'll do the work for you. Not it's not a problem. I got it. I got it.
524. 1:50:08You're good. You just keep doing the awesome stuff you're doing in the [Music]
525. 1:50:23[Music] country. Shang dang dang dang dang. If this doesn't upload, I'm gonna be a cranky boy. Carl Dirk Diggler because we all
526. 1:50:41want to be actors. Honest to God, it just like How about if I hit post? Will we upload it? Please, please. Did it upload it? No. Error.
527. 1:50:56Post. Come on. 30 19
528. 1:51:1121. Does anyone know how to upload something to X? Am I just being an idiot? Is this like a boomer moment? Oh
529. 1:51:20yeah. I was watching this AI learning lab and he was trying to upload something to Twitter and he was like he
530. 1:51:26doesn't even know how to upload a file. Like, oh my god, that's so ridiculous. Like, why doesn't he learn
531. 1:51:32to upload a file if he's going to take all this time to make a file? Why didn't he learn to upload the file? Stupid dumb
532. 1:51:37dumb. He's a dumb dumb. This is a dumb dumb channel. So dumb. So dumb. [Music]
533. 1:52:05Anyone? Do you have to shrink the file first? Maybe. No. Like people are uploading hour-long videos to this stupid [ __ ]
534. 1:52:14site. Oh, you know what? I could maybe try. No, Kyle, why don't you tell us? Shut up. Let me try an incognito window because maybe, just
535. 1:52:29maybe, there's some stupid [ __ ] extension that's incompatible with that stupid [ __ ] site. [Music]
536. 1:52:42Uh, oh, because I'm in incognito. Let's see.
537. 1:52:56I don't remember my password. Wrong password. Yeah, I know.
538. 1:53:26Anyone? Any ideas? Maybe a smaller file will work. I don't know. I This is so [ __ ] annoying. Export as
539. 1:53:41480p to the desktop. We'll replace that one. Yes, we'll replace it. All right, that's good. We'll go here. We'll get
540. 1:53:49rid of this. We'll go here. We'll go here. We'll click [Music]
541. 1:54:06that. 50 60 100. Yay.
542. 1:54:21post. All right. Oh, and I can't share this with you because I'm in a wrong window.
543. 1:54:47All right, did it. I lost all my verve for life in that process. But if you'd be so kind, go to Twitter and like that tweet, that Zeit, the
544. 1:55:03yeet, whatever the [ __ ] you do on X now. And uh quote it, quote tweet it and like it and send it to Sam Alman. And I
545. 1:55:12should Oh, I should tag Hedra, shouldn't I? I should hang on. I should tag edit post at Hedra Hedra Labs and I should tag at
546. 1:55:31Sununo and I should tag at OpenAI and I should tag at OpenAI developers and I'll do at Globalizer.
547. 1:55:57[Music] Sir, look, someone else doing chat GPT plus Hedra Labs. Car sync available on Hedra.
548. 1:56:18Well, I just did Carl the cowboy. You stupid [ __ ] Scooter the Whimo. Carl the cowboy is better than you dumb
549. 1:56:29car. Not that I'm competitive, jealous. You big dumb dumbs. [Music]
550. 1:56:45just had an idea. Or even better yet, ask [Laughter] Grock. Oh, lordy, lordy, lordy. All
551. 1:57:01right, I'm going to get out of here. Uh, it's Wednesday. Tomorrow's Thursday. I'll be back at the normal time tomorrow, 8:00 PM. And
552. 1:57:11uh yeah, we'll we'll keep figuring out what's going on here. I don't I don't I don't know what else we're going to play
553. 1:57:16with, but there's I feel like there's I I feel like we've barely scratched the surface with this image model. Like I
554. 1:57:23don't know why there's my my gut is telling we we don't know what this is really about yet. So it maybe it could
555. 1:57:31just be like yeah, it's just another picture thing. I don't think it is. I think it's bigger than that, but we'll
556. 1:57:36see. We'll see. We'll see. All right. Peace out, everyone. Hope you had fun tonight. Later. Later.