10/7/25 - Kicking the Tires on New AI Tools: A First Look at Building Workflows in Weavy.ai

Transcript

1. 0:00Get your butt over here. [Music]
2. 0:24You and I all alone. Sunday morning he ran home. Skes blue and a coffee strong. It's true.
3. 0:38[Music] Then I open my eyes to a dream realized in front of me. Then I got a clue what in the world has
4. 0:52happened to me. [Music] Think I think I'm happy like first day summer vacation. Happy got to get a little rest and relaxation.
5. 1:07Happy like choir on Sunday morning singing true. [Music]
6. 1:32in a westerly direction. [Music] This car is my train. [Music] I' been driving. I've been wondering
7. 1:47what it is I'm running from again. Feel like an 80y old man holding on to 29. And up ahead on that horizon
8. 2:04is a California line. [Music] [Applause] Up ahead of trucks carrying a wide load. Three fat house cut in half.
9. 2:20[Music] Cute little front door into two windows. My lord ain't sure whether to drive should last. You see, I broke a home up myself once.
10. 2:34As I stumbled to that door, I read a note by the dawn light said, "Don't you come around here [Music] anymore." [Applause]
11. 2:49Well, I've had enough freedom on the rock. Never was good with decision. At least that's what I've been told. [Music]
12. 3:11Oh, good people. Good people of the Tik Tok, of the YouTube, of the Twitter, of the LinkedIn. What is going down? What's
13. 3:17happening? was shaking. Hope everyone's doing well on this fine Tuesday night. We had a lovely AI salon tonight. Had a good crowd.
14. 3:28We had Curi from New SP No Spoon Studios who is she is a an impressive woman building, delivering, building and shipping.
15. 3:41Building and shipping. She's got [ __ ] out in the world. She's just doing it. She's just doing it, people. This is This is the new way.
16. 3:53There's There's no There's no waiting. There's no planning. She might have planned, but she's delivering. [Music]
17. 4:59[Applause] She came home again like slow moving cold fronts. His beard was warmer than a look in her eye.
18. 5:19[Music] She said, "Give me a love that don't freeze up inside. [Applause]
19. 5:33[Music] You said I have melted in my time, dear. [Music]
20. 5:47But to sit next to you, well, I shiver and shake. And if I knew love, well, I don't think I'd be here.
21. 6:03Asking myself if I've got what it takes. [Music] Got to hit the right fretboard or the right fret on the fretboard.
22. 6:21[Music] All right, let's uh let's get rolling. Let's get rolling. Let's get rolling. Let's get rolling. Let's do something.
23. 6:33Everybody ready to do something? What's happening? Oh, man. It's a light light crowd tonight. All right. Well, it is late.
24. 6:42So, so uh we'll just have us a we'll have us a a nice short little thing. Maybe maybe the crowds will come rushing
25. 6:48in. Um I figure first thing we'll do is um we'll play Gareth. Gareth did a uh an entry for the door brothers contest
26. 7:01that's called the door or door like there were three categories and one of the categories was a door and so he made
27. 7:08this door video. So let's watch Gareth's door video. Your fantasies are our desire. [Music]
28. 7:25[Applause] [Music] City rain falls down on concrete in dreams.
29. 7:40Thousand faces past. Nothing's what it seems. One moment in the crowd, one look, one glance.
30. 7:57Sometimes that's all it takes for cir behind closed doors in the glow of blue light.
31. 8:14Searching through the dark for something that feels right. Fingers trace the glass. Following the thread
32. 8:30into someone else's life where angels tread. >> All right, let's have some fun.
33. 8:44Step inside the dream where nothing has a cost. [Music] >> Hey. >> Hey, babe. How are you? >> Good. Now,
34. 9:00[Music] >> I've been looking forward to this. >> Me, too. >> I have a secret. >> Oh, yeah. What's that?
35. 9:12>> You shouldn't steal photos of women and not tell them. I'm not that kind of girl. [Music] >> Back into the rain. [Music]
36. 9:26But now the city's changed. Every wall and window has been rearranged. [Music] She walks through the same street,
37. 9:42sees what I have done. The game was rigged from the start. There's nowhere left to run. >> Oh my god. What is going on?
38. 9:58keeps falling down on concrete and what we think is cracked left behind your eyes.
39. 10:28City rain falls down >> at the door. Your fantasies are our desire. >> That's cool. There you go. Uh
40. 10:46that was pretty cool. I don't know if Gareth's here. Nice job, though. Um, are you paying $200 to test this? To test
41. 10:53what? That that video model. That was not my video. That was Gareth's video. Um, my only comment on his video was I I
42. 11:02wouldn't put, you know, 100% AI at the end. I I don't think it's anyone's business if you used or not. They can
43. 11:09figure it out. If it's a if it's a good story, it's a good story. So, but uh but yeah, nice. Good luck with that.
44. 11:18Um, oh wait, Sora 2, $200 for the API. Is that true? Is that what you were asking? The the Sora thing, the API is 200 bucks a
45. 11:31month. Might be. I think it's 10 10 cents a second to render. So 200 bucks is 10 seconds times
46. 11:46uh 200. So however many 2,000 seconds and then that's some amount of minutes. Yeah. If you're going to get into doing
47. 12:01video at any kind of scale, you got to have you got to have yourself a [ __ ] budget with a budget with a capital B. A
48. 12:09capital B there, boys. Boys and girls. Um, I want to go play with something. Is it Weebi?
49. 12:24Wee AI? I think it's weebi powered workflows built for creative pros. All right, let me share this tab.
50. 12:41If you're new here, my name is Kyle Shannon and this is the AI learning lab. We are going to go share a screen here.
51. 12:52Weevi artistic intelligence. So, this has got one of those really [ __ ] annoying interfaces um that are designed by 21 year olds and
52. 13:04they're like, "No, we can make the font smaller. Screw the old people. [Laughter] Oh my god.
53. 13:21Every little fun." Um, I think I think 8 point 8 point font's fine. I'm totally new to this. I know. Me too.
54. 13:32So, I have heard about Weevi now from two people in two days and both people I respect. One was Retroun Punk um who's
55. 13:41who's uh you know does digital and AI and creative at an agency and the other one was Liz Miller Gerschfeld who host
56. 13:48co-host the salon with me. Um, they both told me about this today. So, I figured I'd go we'd go check it out.
57. 13:59Um, I talked about this last night. I talked about it tonight at the salon. I know nodebased interfaces have been around forever and
58. 14:11you've got make and you've got Zapier does pseudo nodebased [ __ ] and you've you've had databases that are nodebased.
59. 14:18We've had nodebased interfaces interfaces for years, but there was something about yesterday when OpenAI launched agent builder. And even
60. 14:30though agent builder looks like it's a bit rushed, it's not it's not we're not going to be able to build anything in it
61. 14:35in here. It's just it's too wonky right now. Um, and then 11 Labs did a nodebased workflow maker. like there's just it feels like
62. 14:47the world's going to get pretty complicated pretty fast and there's there's a part of me that's like we don't need to do this because 3
63. 14:57years from now all of this [ __ ] will be behind a simple interface but I just kind of feel like this is the
64. 15:04equivalent of learning math so you can understand the power of a calculator you know um I it's just feeling to me like
65. 15:14we've got to start something. So, so there you have it. There you have it. There's Daniel. Hey, Danielle. Uhuh.
66. 15:26Is it lagging for anyone? Which which one's lagging? Is it um Tik Tok or YouTube? Agent builders. They're sort of agent
67. 15:36build. Some are agent builders. This one's just a workflow builder, but you'll you'll see if I can figure out how to use it.
68. 15:43Where? When? What? It's called Weevy app. Awesy. Tik Tok. No lag. Fine now. Okay, great. Beautiful.
69. 16:10Multiple image models. V2. Okay. Wait, why? Oh, because it it opened a new tab. God damn these sites that open new tabs all the time.
70. 16:29Very easy. Repeat four times, please. It's fine for me. Tobias, hello. Hello. Hello. Hello. Hey, everyone.
71. 16:40doesn't seem to be lagging for me. All right, I think the lag lag issue is good and gone. We're good. Okay, so what we've got here,
72. 16:50we've got a prompt, and that prompt has an input and an output.
73. 17:05And then we've got GPT image one, Flux Pro Ultra, Miniax, Lumaf, Photon, Recraft V3, Mystic, Ideog, Stable Diffusion, Google Imagine
74. 17:194, Fall AI, Higsfield Image, and Fall AI Flux Crea, and the other one was Fall AI Bite Dance Seeddream V3. And I think
75. 17:32that's up to V4 now. So what we might want to do is update some of these here nodes. And then what happens then? You send it
76. 17:44to Mystic. Oh, and it makes videos. Okay. All right. I I sort of get what's going on here. Duplicate my files. But I I think I want
77. 17:55to start one from scratch. So the the reason for this guys is well well there's a couple of things. If
78. 18:09you're working with clients, like I just did a job for a client, that if the client had asked asked for my
79. 18:23prompts or what models I used and what models I used when, I was just all kind of managing it in my head. I'm like,
80. 18:30"Oh, let me jump over to this model. Let me jump over to that model. Let me jump jump." And if something didn't work, I
81. 18:35would just sort of give up on it on this thing and then just copy the prompt and go to something else. What this allows
82. 18:40you to do is put together test environments where you can put in a single prompt, test it across multiple
83. 18:46models and see which one works best, but then you've got a a a workflow, a repeatable workflow and a and a sharable
84. 18:54workflow that you can share with clients or with other people, with other creators. I'm viewing a readonly version.
85. 19:01Duplicate my files. Okay. So, let's go back to Oops. Let's not do what I just did and destroy that. Looks like we got some new folks in
86. 19:14here. Welcome everyone. My name's Kyle Shannon. We are on Weevie. Well, we're going to be here in a second.
87. 19:24and Weevie other than being a bad name. It's like I I sort of get what it means, but it was like that was the only domain
88. 19:33you could get. Come on. I'm sure the founder of Wee is on here right now. Hey man, we we paid a consultant a lot of money for that.
89. 19:46Okay, Sora 2 has landed in Weevi. Check it out in action. Okay, nothing here yet. Start weaving. Create new file. Okay.
90. 19:59Welcome to the Weevie editor. Skip. Start tutorial. I'm going to do the tutorial. We're going to do this
91. 20:04together. All right. So, s here's what I want you to do. Go to Weevi. Weevi app.weeevi.ai. Make an account. Hey, Kyle Achilles.
92. 20:15What's happening? Please post a link. Okay. Um, Brandon's doing that
93. 20:34mouse, hold the space bar and click left to drag. Okay. Oh, there you go. Hold the space bar and left left click
94. 20:49to drag. Okay. All right.
95. 21:05This is prompt mode. Oh, this is a prompt node. It holds a text prompt that guides your image generation. Okay, great. Next. Hold command
96. 21:19command control and scroll mouse wheel to zoom. Okay. So just Okay. That does that. That does that. Got it. Okay.
97. 21:33You have to you have to load on weev app doesn't load. I don't I don't know what
98. 21:48that means. Brandon, >> you're giving out app.weeevi.ai that doesn't load unless you're logged in. Other it's just weebi.ai if you're
99. 21:58not authent. Oh, load like load. I get it now. what you said. Okay. Drag and drop fast flux to to add a menu to your thing. Okay.
100. 22:12Got it. Fast flux. Got it. Click and drag from the output handle to the input handle. Got it. All right.
101. 22:26Click run model. Where's run model? 150 credits. No active runs. Click the
102. 22:45wh why am I not seeing the run model button? Oh, because Okay, let here's why. because it's it was down at the bottom
103. 22:56right hand corner of that box in like sixpoint type. There it is. All right. All right. So, my first wee. Yay. You
104. 23:12did it. Yay. All right. Start working on this workflow. Okay. Cool. And then we can say guidance flux fast
105. 23:24is the model aspect ratio we can do 16 by9s. I can say run again. All right. So it's one credit per per run it looks like.
106. 23:44All right. And then we're going to go where are we going here? Are these image models? Yes. Imagine three three credits.
107. 23:55I see. Highest quality, six credits. All right, I'm starting to get this. There's stable diffusion. Where's Nano Banana? Search. Nano.
108. 24:10Nano Banana. Bang. Out comes Nano Banana. Run model.
109. 24:34All right, that's kind of cool. Aspect ratio default. You can do 16 by9 here. Total cost four credits. for Nano Banana.
110. 24:49I'm such a cheap bastard. I keep I see those credits going down. I'm going to be like, "You son of a bitch."
111. 25:01All right. What else? What's in our toolbox? Let's go look at our Let's go look at our toolbox. Compositor Painter. Wonder what a
112. 25:13painter does. What's a painter do? So, we're going to do the output of that into the input of that.
113. 25:29Oh, and then you can paint on it. Except I don't know what I'm doing. So, so we're going to delete this. Okay. Beautiful.
114. 25:47Um, [Music] mat extractor mask by text video mat prompt prompt concatenator. So you could take
115. 26:01two prompts and put them together. Prompt enhancer. Well, let's do that. Okay, this is so this is kind of fun. So how how do I
116. 26:10break these? Do I just click on them and delete them? Yep. Okay, easy. So, we'll go prompt enhancer and we're going to go from
117. 26:21shitty prompt to prompt enhancer. And then we're going to go from prompt enhancer out here. All right. And so, we're going to do a
118. 26:32shitty prompt of 70s. We'll do 70s muscll car. resto mod in abandoned factory. Shitty
119. 26:48prompt, but it's a classic. And then I'm going to say run model. So, let's zoom in on this thing. This is actually pretty cool. Um,
120. 27:00yeah, I'm I'm I'm feeling like I'm feeling like there are there are skills to be had here. All right. So there we do that. We run the model.
121. 27:14Okay. No active runs prompt enhancer models instructions. Your job is to write prompts for textto image models.
122. 27:22Your input will be a general description for the scene. You should write a detailed prompt without any additions no
123. 27:29longer than three sentences. Um we'll do no longer than five sentences. Um, you should include
124. 27:46um detailed scene subject.
125. 28:03Um, scene, subject, and time of day. You should include lighting,
126. 28:22visual style, eg um impressionist painting, um uh
127. 28:37ectochrome film, etc. What else do we want to include?
128. 28:53And camera angle. Camera angle. Okay. So now we wrote a new prompt. So this is that's that's shitty prompt writing. But this is
129. 29:07effectively prompt engineering, right? We're creating an application here where you put in a shitty prompt. You run your model. Boom.
130. 29:16Capture a low angle shot of a meticulously restored 70s muscle car behind the soft diffused light filtering through the grimy
131. 29:26windows of an abandoned factory which I'll often write that the car's gleaming paint contrast sharply with the
132. 29:34crumbling concrete bump sim simulate the the nostalgic look of Kodakchrome film emphasizing warm tones. Okay, this is cool. Um,
133. 29:46all right. So, this is Gemini Flash Nano Banana. Can I change that or do I have to bring out another thing?
134. 29:53I have to bring out another thing. Let's go get a different model. Chat GPT image one. Uh, let's do GPT image one edit.
135. 30:06Well, GPT image sucks. Let's do uh is Grock in here? Of course not, because Elon Musk's a dick and he doesn't want
136. 30:15anyone to play with his toys. Um um well, let's let's go look at them models.
137. 30:30Higsfield image. That one's kind of cool. 21 credits. Oh, no. Wait. Commercial use 21 credits. Good lord. Imagine four
138. 30:51is six credits. Imagine three fast is three credits. Let's drag that one out there. So, we'll grab from here. We'll go into
139. 31:04there. Oh, cool. You can do a negative prompt, too. No, that's pretty slick. Oh, ideog. How much is that? Four credits. That's
140. 31:19pretty cheap. Ideog V3 character. Create consistent character. Oh, that could be a cool thing we could do. You could start with an image. So
141. 31:33you could basically go image prompt image in image prompt prompt enhancer image and then take that image into like of a person
142. 31:47and then use ideoggram v3 character and have that person in a bunch of different scenes that each have their own prompt.
143. 31:54Yeah, that's kind of cool. All right, I like that. Um all right so let's do aspect ratio 16 by9 16 by9 16 by9
144. 32:11output quality runs all right run model run model
145. 32:27run model and then I think there's a way I can I can run the output into something that'll send it to all of them automatically.
146. 32:38So now we're down to 132 credits. By the way, there's Nano Banana.
147. 32:53Nano Banana is pretty [ __ ] good. This is Google imagine three fast. The hell is that car?
148. 33:12That's like a 74 Mustang ass end, but it's wider. That's bizarre. Okay. Um All right. Let me shrink that
149. 33:26down. Can you organize these? How do you organize them? What do you do? Do I right click? No. Uhuh. Uhuh. Uh uh uh.
150. 33:38Is there an organized thing? Prompt import quick access. Import multiple lures. I wonder if you can create lures. I bet you can.
151. 33:56Blur mask prompt prompt concentrator. Join multiple text inputs into one output. Prompt enhancer.
152. 34:11Image describer. Very cool. Video describer. Text iterator. Iterate over a list of values. Import. Export. Router.
153. 34:26All right. So, let's take Right. Oops.
154. 34:42I do. No. This. No. This. This. This. Okay. And then I go and I go and I go
155. 34:56and I go and I go and then I go this down here into that router. Right. Isn't that right? I think that's right.
156. 35:10What's this assets? Okay. So, let's go do let's do a shitty another shitty prompt.
157. 35:25Uh, let's say um um grizzled old New England fisherman
158. 35:39smoking near the docks. Okay. And then I guess I go over here
159. 35:57and I run it. Bang. Bang. Is it running? A weathered New England fisherman with a
160. 36:11long white beard sits on a wooden crate near a bustling dock at dawn. He wears a thick wool sweater over his
161. 36:19hat, puffing thoroughly on it. Does anyone have questions? Is this making sense what I'm doing? Kicking the tires.
162. 36:25Yeah, totally kicking the tires right now. I made it early but grossly misjudged my alcohol intake. That is beautiful. I love that. Um, how
163. 36:39do I organize this [ __ ] Anyone? Does anyone have any idea? So, I select them all. Select them all.
164. 37:27I need three three more lifetimes to figure out how to use all this stuff. I know it. Well, this is this is the uh
165. 37:35this is the rub right now. So, here here's here's I think the antidote to that, Jeff, is just figure out what you want to do
166. 37:47first. Don't worry about the [ __ ] tools. Like I want to learn I want to learn how to do this. Like the what's what's popping in my
167. 37:58head right now is this is cool, but it's going to be really expensive and it's going to be really tedious.
168. 38:05But what what what this actually is is I bet there's a lot of people out there that have set up workflows that you can
169. 38:13just go use and they're pretty [ __ ] cool. So we should probably go look at that now that we generally understand
170. 38:19how it works. What I don't get though is why duplicate delete lock. Why is this router not routing? Because I ran the model.
171. 38:41How do I get it to I don't get it. I don't get it. I do not
172. 38:57understand anybody. 3D models,
173. 39:13community models, image upscale, clarity, video smoother, ID preservation, control. Good [ __ ] lord, there's a lot here.
174. 39:27Video utilities, video concatenator. Oh, that's cool. So you could generate a bunch of video pieces. So actually what you could do
175. 39:39Oh, this is this is probably some version of what Carrie did and or Kerry did in her app that she demoed tonight in the salon
176. 39:49except she had uh what you call it build it for. She probably didn't build it here. But you could build a workflow that takes an input,
177. 39:58generates a script with eight different scenes that tell a narrative arc, generate images for each one of them,
178. 40:05start and end images for each one of them, run a video model that takes input and output images, run both of those in there,
179. 40:16generate the clip, concatenate all the clips, and generate a movie. Yeah, you could you could totally do that with
180. 40:24this model. What's video utilities input file result frames per second
181. 40:39basic video processing convert input to MP4 input to GIF. Tik Tok question. Yes. Yes. someone had a question on the Tik Tok.
182. 40:51Wouldn't it be easier to just write a prompt in a difficult platform? I'm not getting this. Okay, so here's here's
183. 41:02here's the the the trick source camp. Um, let me get rid of this thing. Get rid of that node. So, what I don't know how to
184. 41:13do right now is um is autotrigger this router. But, but basically what I've got here is a shitty prompt and then there's a mo a node here
185. 41:28where I can improve the prompt and I can control that. So, I write a shitty prompt. It automatically writes a better
186. 41:34prompt and then I've selected three different models over here. And so from one prompt I get three different
187. 41:41outputs. And I could have nine different outputs or a dozen different outputs. Or I could have a single image output or or
188. 41:49sing single image input and then have eight different um scenes with that same character in it. Let's go let's go look at some different um
189. 42:03some different workflows. Maybe it didn't auto route because it wasn't new images. You redid them. I don't know.
190. 42:18This This is the kind of [ __ ] like this is the kind of [ __ ] with software, right? Like you can learn whatever I
191. 42:24just learned in 15 minutes, but it's going to take me three hours to figure out how to auto route auto trigger those things.
192. 42:34Back to my files. Okay. So workflow library weave welcome multiple image models v2 editing images compositing node image to
193. 42:49video camera angle ideiation. That sounds cool. Yeah, look at this. Okay, so here source camp. Look, let's let's go look at what this does.
194. 43:05You put in a prompt. You're given a reference image of a scene. Any LLM using the provided image. Okay. Wait.
195. 43:20Oh, okay. Wait. So, it starts up here. Got it. Oh, I see. I see. Okay. So, Oh, I'm not sharing. God damn it. Hang on. So,
196. 43:33source camp. Look at this. and all of you look at this. So, so we've got in the upper leftand corner here
197. 43:43is a prompt that generates an image and then that image it looks like comes into this thing which generates
198. 43:55this. No, no, this is a prompt for this. Okay. So the image the image comes into this LLM which uses this prompt to do something with it
199. 44:10and then it goes out into use the provide using the provided image keep the style and subject similar list
200. 44:27and then this is Gemini 2.55 flash nano banana And the prompt is coming in from oh I see here. So if I click on this
201. 44:45using the provided image. Ah show more.
202. 45:01Ah, each prompt must must begin exactly with using the provided image, keep the style and subject details similar and
203. 45:11modify the camera to match this. After that phrase, describe a compositionally different camera perspective. Okay.
204. 45:23So then it makes a pile of prompts and then those prompts go into this text array. Oh, I see. Oh, this is cool. So, okay.
205. 45:34So, and it's split by it's split by an asterisk. So, in the text right here, when it finds an asterisk, which right
206. 45:43at the beginning of this paragraph, there's an asterisk. So what this what this text array does is it takes this
207. 45:50longass prompt and it splits it into individual nodes individual and then and then this array these are the individual prompts for all
208. 46:04these different windows. Okay, I got it now. Okay, cool. Watch. So watch. Let's let's just use this once. This is going
209. 46:11to cost us a lot of credits, but that's okay. You're in read only. So like duplicate to my files. So now we have a
210. 46:18version of this that's ours. So now I can go in. So now it'll start to make sense. Source camp. Um let me see. Um 70s resto mod.
211. 46:35Let's see. 70s muscle car restood. Um, idling in an
212. 46:52abandoned factory. Um, the driver in the car has a mustache.
213. 47:14um and sideburns and has seen better days. The car looks great.
214. 47:31Um great. Let's see. Uh, sunset light breaks through the smudged
215. 47:47factory windows. We'll leave it at that. Okay. Now, how do I do all these things run automatically
216. 48:02if I run this one thing? We'll find out, I suppose.
217. 48:27Okay, there's our car. There's our dude with a mustache who's seen better days.
218. 48:43So then I guess I run this model using the provided image. Okay, so
219. 49:00there's got to be to a way to run all of these in a row, right? Anyway, all right. Here's one. Here's one.
220. 49:12Here's one. Here's one. This is going to eat up all of our credits, but this is going to be cool. Here's one. Here's one. Here's one.
221. 49:25I don't know what I'm doing. I'm clueless. Clueless. Clueless. I don't know what I'm doing. But look, here's the undercarriage of the car.
222. 49:36This is actually really cool. Here's the hands on the steering wheel. Here's the shot. That's not the worm's eye view,
223. 49:42right? There's the There's the That was our original shot, right? I don't know why these two didn't get a
224. 49:50prompt. Oh, they're still rendering. Look, there's the bird's eye view. This is cool. This is very cool.
225. 50:12Huh? Uhoh. What happened? Gemini could not generate an image with a given prompt. Try again.
226. 50:38weevi.ai. Yep, that's correct. Really curious how anyone uses this at work. Is it just an experimental storyboard?
227. 50:50Um, I think it's a bit more than an experimental storyboard. I think, well, for one thing, you can share these workflows.
228. 50:58So, so like what we're doing here is like I don't I don't know what the [ __ ] I'm doing, so I don't I don't quite know
229. 51:05how to troubleshoot this thing, but this is write a single prompt and get 10 different shots, 10 10 different angles of that of that shot.
230. 51:19Um, so this is a storyboard generator, right? Or this is a this is a shot possibility generator. So this could be a component in a larger
231. 51:32application where you actually call this this workflow and behind the scenes it's it's doing all the code and doing all the API
232. 51:41calls, right? And so all you would do is I'm sure there's a way I'm sure there's a way to grab the URL or the something from this. Duplicate
233. 51:52delete lock. I don't know. Maybe there's a an API hook somewhere. Video smooth. That's all AI [ __ ] 3D models. That's video models.
234. 52:09Image models. Toolbox levels compositor crop blur channels extract video mask prompt prompt con concatenator
235. 52:22prompt enhancer image describer video describer. That's cool. So like here's yeah this is so this is something that that could be really
236. 52:33interesting, right? You you upload a video to something. You analyze the frames of that video. Figure out where the the visual story
237. 52:47story changes are. Then you do the same with dialogue. Then you marry those two. You could you could storyboard out an existing video
238. 52:56automatically. Text iterator. I don't quite know how that works. Import Laura. Import multiple Lauras
239. 53:05data types input number text list toggle seed array and then there's a bunch of images image models
240. 53:20enhance image. Yeah, this is this is another thing this could be used for is testing, right? You could put together.
241. 53:33God, man. You could spend [ __ ] weeks in this thing. If you really wanted to understand the state-of-the-art for image, video, audio,
242. 53:44and what's possible with it, if you spent a month with this thing, you'd actually understand what's going
243. 53:51on. That's pretty crazy. All right. Um, looks like the evidence of a bad detective series. Well, two two of my things are not
244. 54:05running because the prompt is bad. Which which two are they? Uh, five and six. I don't know what order these are.
245. 54:19So, wait. One, two, three, and then four, five, six. Three and six. So if I go here, I go one, two, three. This one
246. 54:38to match wide establishing shot with the car minuscule emphasizing the vast sunlet industrial space. What's wrong with that?
247. 54:48Can I edit this? No. Oops. Yes, I can.
248. 55:04Huh. So, I got to run this again.
249. 55:27[Music] Civot is a platform for trained lures or stylized generations. More style control basically instead of dude with mustache
250. 55:44in anime style. Use a trained Laura. Yeah. Yeah. Yeah. For the look you want. Yep. Should show him civet stuff would blow
251. 55:52his mind. If he likes images. Yeah. I like I haven't gotten into the whole dev side of this thing because it's it's
252. 55:59tough enough keeping keep just keeping up with the commercial the commercial [ __ ] Um but there I got
253. 56:07there I got those two things to render. So now we have now we have all the shots. Can you download these images? Download current. Yeah,
254. 56:21download current. And then I have to [ __ ] name it. [ __ ] Uh, so
255. 56:35this is exhausting. Yeah, you have to like you have to have naming conventions prepared for all this [ __ ] Like at some
256. 56:42point what I should do is I should output all of those images into a folder, then have the folder automatically zip it, and you download
257. 56:50the zip, and then you've got all your named named files based on the prompt. You could you could totally do this. You
258. 56:57could totally do some powerful [ __ ] with this. But like what I want what I want to know, like whoever asked the question
259. 57:07about like what do you do with this? Is this is this just a glorified storyboard? If it's just a glorified
260. 57:12storyboard, it's a little too much work. If I can take this workflow and actually call it via the Weave API
261. 57:20and I can have it just auto run all this [ __ ] So, I can put in a single prompt into an application and it spins its
262. 57:26wheels and then it starts spitting out these outputs. That would be pretty [ __ ] cool, don't you think?
263. 57:40Is the dude still in there? Yeah, he's still in there. Is he still in there? Yep. His head's too big. Is he still in there? Yep.
264. 57:52Yeah. This is pretty crazy. So, this is Nano Banana. Like, it kept it's The car is the same in all the shots.
265. 57:59The guy's the same in all the shots. That's pretty remarkable. I mean, not for nothing, think about if you wanted a shot like
266. 58:11that for a commercial and then the director was like, "Well, we need all these angles. We got to get under the car,
267. 58:20right? We need we need an underexposed one. We need his hands on the wheel. We need one from the hubcap view. We need
268. 58:28it over over the driver's side. We need the close one. We need the one over the top. We need the one with dust after he
269. 58:36smashes into something. Yeah, this is cool. And then you could run each of these images out to video models
270. 58:46if you really want to blow through your credits. I wonder how much credits are. Let's go find out how much credits are.
271. 58:56Um, credits. Get more credits. $19 a month for 1,500 credits. So about 3,000. So I started out with 150 credits.
272. 59:10That's 375 images. Well, no, no, that's not right. Some of these images are six credits. They said three
273. 59:27375 images. That's not right. Most people are paying 36 bucks a month.
274. 59:4245 if you go monthtomonth. I would go monthtomonth. Team $60 per user per month and they get 4500 credits. Plus, they get unified billing.
275. 1:00:03All right, let me go look at Civet because I know Civet was something that was cool. I'll just go here. I don't
276. 1:00:09care about this thing. Civetuh Civet AI, the home of open-source generative AI featured images.
277. 1:00:27Yeah. Yeah. This is this is all cool and all except if you think you were overwhelmed with commercial options.
278. 1:00:38Go to go to the world of open source and let yourself be swallowed in possibility. All right. Um I gotta stop trying to pay
279. 1:00:50attention and learn because my head hurts. Ah um Tik Tok question Tyler all platform for image generation or
280. 1:01:01overpriced UI for the use of an API. Yeah, everything can be scheduled. I think we can get super intelligence by
281. 1:01:142030 if the models make better versions of themselves. Yeah, he's agreed. I'll eventually try to reproduce my dreams with AI.
282. 1:01:25Oh, I did that one night. One night. I had a dream. This was when I was first playing with AI. I had a dream and I woke up
283. 1:01:40and I described my dream to chat GPT and then I said, "Write an image prompt for that." And then I went into one of
284. 1:01:48the image generation tools and generated it. It was pretty cool. You could totally build that.
285. 1:02:05And a year ago, these tools were struggling to spell car correctly. Yeah. It's crazy. They doing that in real world brain patterns.
286. 1:02:18Oh, of dreams to AI. Oh, that's cool. That's cool. All right. Holy [ __ ] my eyes are crispy. I gotta go. My My eyes are done.
287. 1:02:28It's time. That's funny. Brandon just said the same thing. He knows me all too well. Um, producer Brandon, listen, I
288. 1:02:36would stay with you guys for hours, but producer Brandon says I have to go to bed. Um, tomorrow is Wednesday. Uh there's
289. 1:02:45the AI readiness project podcast is uh on at 400 pm mountain time. So if you go to aire readiness project.com
290. 1:02:57that will take you there and uh that's on tomorrow at 400 pm. Also go to the salon oh are you readyforai.com
291. 1:03:07go check that out. That's the AI readiness training program uh where you can learn to get your mind right with AI.
292. 1:03:17And then finally, go to the salon.ai. We have a spanking new website that's beautiful. Looks great. Um and it's
293. 1:03:25actually got actual information now like when what are the next events that are coming up and you know what's some of
294. 1:03:31the activity going on in the salon. It's really kind of cool. Um Pave M's in the house. What's happening sir?
295. 1:03:38Yes, you need to go to bed. I really do. Okay, I'm getting out of here. I'll see you tomorrow night at 8 o'clock. I don't
296. 1:03:43think there's anything going on, so I will see you then. And uh I hope you enjoyed the Weevie demo tonight. That was kind of fun. Later.