6/25/2025 - MidJourney Magic: Crafting Unique Animations and Tales with AI Technology

Description

Chapters

Transcript

0:01 Come on, buddy.
0:05 [Music]
0:20 [Music]
0:30 Is this place I can rest my forehead
0:33 [Music]
0:37 to gather my thoughts in sweet silence.
0:41 [Music]
0:44 And here's this place where the feelings
0:46 are near.
0:50 From an over exposure to violence.
0:54 And this is place I can slowly face. The
0:58 only one I truly can know.
1:02 These are tears from a long time ago.
1:05 Got these tears from a long time ago. I
1:09 need to cry 30 years or so. These all
1:13 tears for a long time.
1:18 We go.
1:21 [Music]
1:25 Oh darling, oh darling, you say unto me,
1:29 [Music]
1:32 where have you been all my life?
1:36 [Music]
1:38 I have been swimming through seven sad
1:41 seas.
1:44 All women have me their lies.
1:48 I
1:50 weren't
1:52 them, but I didn't know.
1:56 These are tears from a long time ago.
1:59 Got these tears from a long time ago. I
2:03 need to cry 30 years or so.
2:07 These are tears for a long time.
2:10 [Music]
2:13 Let's go.
2:14 [Music]
2:42 What? Why you guys so ramy?
2:47 See, because we don't know how to talk
2:48 to dogs. I might be telling him like
2:50 he's an [ __ ] when I'm howling like
2:52 that. I don't know what I'm saying. I
2:54 need AI to help translate me into dog.
3:00 What's happening, good people? Welcome
3:03 good people of the Tik Tok, people of
3:04 the YouTubes, the the the LinkedIn, the
3:06 the Twitter's,
3:08 the Twitter. I don't know if you know
3:10 this. They uh they don't call it Twitter
3:13 anymore. They call it the X. They call
3:16 it the X and it's X marks the spots and
3:18 you zit you zit instead of tweet.
3:24 That's what the kids are saying these
3:25 days. And by kids, I mean Elon Musk.
3:29 Oh, good lord. Good lordy. Good lordy.
3:31 Good lordy. Good lordy. Good lordy. Good
3:34 lordy. Good lordy.
3:36 There is so much going on in AI
3:40 that so last night when we started the
3:42 show, Brandon was like, "Uh, I can't
3:43 keep up with this." And there was like
3:46 10 things. And then tonight he's like,
3:47 "I can't keep up with this." I'm like,
3:49 "Wait, didn't you say that last night?"
3:50 He's like, "Yeah, they just dropped a
3:52 bunch more."
3:58 Um,
4:00 so we'll talk through some of them. Some
4:02 of them I don't really know a ton about,
4:06 but I'll tell you about them and then
4:07 you can go you can go do your own
4:09 research, as they say in the crypto
4:11 world. Do your own research. Here, I'm
4:14 going to give you this hot tip on this
4:15 crypto. You're going to make a fortune
4:16 on it, but you know what I mean. Do your
4:18 own research. Wink wink. Huh?
4:22 That's the That's the crypto bro
4:24 community right there for you, huh?
4:29 Just a bunch of degenerate gamblers.
4:35 [Music]
4:41 Um, what we're going to do tonight
4:43 though, um, I'm going to do some of the
4:46 news items
4:48 [Music]
4:55 is I'm going to go back into midJourney.
4:57 I continue to be
5:00 I continue to be Impressed isn't the
5:03 right word.
5:08 I feel like
5:11 I have I I have yet to scratch the
5:14 surface on
5:16 MidJourney video. I don't know what it
5:18 is about it.
5:20 It is arguably
5:22 nearly identical to about 14 or 15 other
5:26 video models,
5:29 but there's something about it that's
5:31 just different.
5:34 So, we're going to I just want to keep
5:36 experimenting with it and see if we can
5:38 find new things to new ways to prompt
5:41 it, new ways to understand it.
5:44 So, so that's what's going to be
5:46 tonight. Kyle chat GPT gives you two
5:48 images at once now. Oh, fantastic.
5:51 Fascinating. It didn't as of this
5:53 afternoon or when did I do I made I made
5:56 this image?
6:00 I made this image today at
6:04 uh I don't know whenever I made it.
6:06 Let's see. Let's just go. Let's go make
6:08 an image, shall we?
6:12 Um I posted it in the salon. All right.
6:14 I wonder Danielle if it's one of those
6:16 things where you got access to the new
6:17 and shiny chat GPT and all of us lowly
6:20 peasants didn't. Um let's see. Make me
6:27 Why is that not typing?
6:29 because I wasn't in the chat hole. Make
6:31 me a wide image of
6:37 a
6:39 [Music]
6:47 rolly
6:49 poly
6:52 bug
6:56 rolled up on a suburb. Urban
7:02 sidewalk
7:05 that takes me back to my childhood. All
7:07 right. Bang. Here we go.
7:10 See, it's right now it's looking like
7:13 one. Danielle,
7:15 I don't know. I have a feeling you've
7:16 got some super secret.
7:19 I just Why? I just have this sense that
7:21 you've got this like personal texting
7:24 relationship with Sam Alman and he's
7:27 just like, "Hey, do you want to play
7:28 with a cool new thing?" And you're like,
7:29 "Okay. Is that what's happening? And
7:32 you're just not telling anyone?
7:35 She had the new voice before we did. She
7:37 had her chat GBT singing to her. We're
7:40 all like, "We can't get it to work.
7:43 [Applause]
7:45 Go look in the regulars."
7:48 [Music]
7:54 AI.
7:56 AI.
7:58 Hello. This is the AI.
8:06 Where' you say irregularize?
8:09 I regularize.
8:12 Be not afraid. Wow, look at that.
8:16 Huh?
8:18 Well, let me go back to my chat. Jet tag
8:20 and see what it's doing. No, it just
8:22 made one.
8:26 It's a lovely image of a roly poly bug.
8:29 I wonder if I have to go to my settings.
8:38 Nope.
8:50 H
8:51 I don't know. I ain't got ain't got no
8:53 no no idea. I ain't got no idea. That's
8:56 very cool though. Um, okay.
9:03 [Music]
9:11 Hey, Sharon Crawford. We're shaking.
9:14 What's happening? All right, good
9:16 people. How many we got? We got 14. We
9:19 got low numbers tonight. So, do me a
9:21 favor. Share where you can. Go tell your
9:24 friends. Go tell your neighbors.
9:29 [Music]
10:11 Wow.
10:12 [Music]
10:22 [Music]
10:34 Um,
10:36 okay. Thoughts, questions? Anyone have
10:39 any questions? I work at a job with no
10:40 AC and it hit 114 today. I'm probably
10:44 gonna not make it through the night.
10:46 Literally cooked. That sucks. That
10:49 sucks. That sucks. I'll tell you that
10:52 right now. Side hustle Mimi still one at
10:55 a time. Yeah, Danielle. Danielle is on
10:58 the super secret open AI.
11:02 just give it to there's probably like
11:06 270 global
11:09 random people that just get good [ __ ]
11:11 sooner and I think she's one of them.
11:17 Um, all right. So, let me let me talk
11:19 about I'm going to go down a list that
11:21 that Brandon served up here.
11:24 Okay.
11:28 I guess I could go look at some of these
11:32 Gemini testing animals thread. Maybe I
11:35 should just go to my my ex
11:38 direct messages.
11:43 Talking not testing. We're not Google is
11:45 not doing animal testing.
11:47 Oh, good. Wait, what did you say? Animal
11:50 Oh, animal talking.
11:52 Animal testing.
11:55 Oh, good. These are a bunch of videos.
11:56 All right. Let me So, so let me share
11:58 this as videos. Maybe we'll just go
12:00 watch a bunch of videos.
12:02 So, this is I have not vetted these.
12:04 These are all These are all producer
12:05 Brandon
12:07 found these goodies and sent them to me
12:11 and I've not vetted them. So, it's
12:13 possible.
12:18 All right. 10 out of 10 would generate
12:20 again. All right. So, we'll start here.
12:25 I got the giggles.
12:28 Someone opened up Gemini today and said,
12:30 "I want to see a talking narwhal." And
12:32 you know what? I am very glad they did
12:35 that. Someone even prompted me to have a
12:37 cute Aussie accent and say cute Aussie
12:40 slang like today, mate. Want to get some
12:42 brekie? Can you believe it, honey?
12:44 Someone generated us to be on a cruise.
12:48 # best life.
12:50 [Music]
12:55 All right. So you want to call out if we
12:58 view the thread. Not only have they
13:01 shared these videos, but they actually
13:03 shared the prompts that made the videos.
13:05 So the thread
13:08 All right. So it's from Google Gemini.
13:11 Yep.
13:13 Beautiful.
13:16 Wait, where's the thread? I click on
13:17 here. It's above the video. That's the
13:19 next one down.
13:22 Oh.
13:28 Oh, yeah. A photo realistic narwhal. So,
13:30 this is for V3. These are all V3, right?
13:33 Right, Brennon? Yeah. Yeah. All right.
13:36 Cool. Which one was your favorite? All
13:37 right. Very cool. There you go. Love it.
13:40 All right. Next.
13:44 Holy [ __ ] Anthropic just released a
13:46 prompt engineering bible and it's free
13:47 and it's insanely well written. All
13:49 right. Good. So, where's that? So, Alex
13:52 Hughes, who I guess works at Anthropic,
13:54 right?
13:56 head of growth at Oh, no. Droxy AI. All
14:00 right,
14:03 cool. So, that's good. The link's right
14:05 below it. Okay, perfect.
14:10 Open Deep Research app is launching in
14:12 24 hours, which by the way, I'm using
14:14 right now. So, you know, deep research
14:17 and all these tools. Well, there's now
14:19 an open- source one, and it looks like
14:21 you get a a search a day.
14:25 But we'll go we'll go look at that. Does
14:27 this have any audio with it?
14:30 No,
14:32 but it looks like pretty much like the
14:34 other deep research things. I wonder
14:36 what it's Do you know what that one's
14:37 built on top of, Brandon? I do not know.
14:40 It's open source, so it could be
14:42 anything. All right. U You're spending
14:45 too much time with an LLM. This joke
14:48 hits the mark.
14:51 Why is the scar on the left if the
14:53 appendix is on the right? Dude, you're
14:54 absolutely correct. Your Hawkeye
14:56 brilliance saved us all. Let me try this
14:58 one more time. That's pretty good.
15:03 Um,
15:07 where am I?
15:11 Yeah, it jumped way up high. Yeah, was
15:13 weird.
15:15 Okay, there's Open Deep Research.
15:17 There's that.
15:20 Today we're rolling out Imagine 4 and
15:22 Imagine 4 Ultra in the Gemini API and
15:26 Google AI Studio. We have to go look at
15:28 the Google AI Studio today, too, because
15:30 they've they're putting some new
15:31 interface on there. I I do have to give
15:34 Google credit. I don't know if Pate's
15:35 on, but Pate, if you're here, I'm giving
15:38 Google credit. Okay,
15:41 I got to give Google credit that it
15:44 looks like and it might be this Logan
15:45 Kilpatrick guy because I think he's a
15:47 product guy at Google.
15:50 They're tightening up their product
15:52 strategy. Like their their Google AI
15:54 studio where you do the API stuff is
15:56 getting slicker interface in front of
15:58 it, better interface, and then
16:00 gemini.google.com
16:01 is just getting more regular updates. It
16:03 was just kind of a zombie site for a
16:05 long time and that's pretty good. All
16:07 right. So, he actually came from OpenAI.
16:10 They poached him. He was their head
16:13 developer community at OpenAI
16:16 Google. And Steo's got two images. He
16:19 just asked to ask for them side by side.
16:22 I have two images. Oh, just asked for
16:24 them side by side. Oh, okay. Let me try
16:26 that. Oh, wait. I got to change my We'll
16:29 be back here. Hang on.
16:34 [Music]
16:37 Um,
16:39 give me two versions
16:44 side
16:46 by side.
16:53 [Music]
17:09 No.
17:11 [Music]
17:36 Oh, yeah. There's two. Wait, no, they're
17:37 not, though. It's It gave me It gave me
17:40 two images as one. It's one image, but
17:43 it's got one image cut off.
17:47 I still don't have two images. Damn
17:50 y'all. All right, back to X. All right.
17:53 Uh, we're rolling out Imagine 4 and
17:56 Imagine 4 Ultra in the Gemini API.
17:59 Available for free in API Studio and in
18:02 paid preview in the API.
18:08 Read more in this blog post.
18:13 Great work by the team on this.
18:19 The Imagine 4 family and the Imagine 4
18:21 Ultra. Your go-to for most tasks.
18:24 Precision for your prompts.
18:27 4 cents per output image for imagine
18:31 four. That's expensive
18:36 [Music]
18:42 and six images for ultra.
18:46 Okay. to give you a glimpse of its
18:48 capabilities. Here's some of what you
18:50 can create.
18:54 Created the prompts below. A three panel
18:59 epic comic. Panel one. Tiny Stardust and
19:02 Nebula.
19:06 Radar shows anomaly. Text anomaly
19:09 detected. Hull
19:12 text stardust, which it says pilot
19:16 whispers.
19:22 I don't know what he whispers, but
19:24 whatever.
19:25 Panel two, luminescent
19:28 Leviathan.
19:30 Emergency. Panel three, Leviathan chases
19:33 ship through the asteroids.
19:37 Huh.
19:39 Well, we might need to go play with this
19:45 prompt front vintage postal card. Ky
19:49 Kyoto iconic propaganda. Opticonic
19:53 propaganda. I've been I've been living
19:56 in America for the past 10 years. Sorry.
19:59 Sorry. I thought I see when I grew up
20:03 learning about propaganda was just a
20:05 thing for the history books. I didn't
20:07 realize you could actually like have a
20:10 new propaganda era and yet here we are.
20:20 Front vintage travel postcard from
20:22 Kyoto. Iconic pagota under cherry
20:26 blossom. Snowcapped mountains. Clear
20:28 blue skies. Vibrant colors.
20:34 Huh.
20:38 Now, here will be a good thing to to to
20:41 understand. So, where I'm gonna go, I'm
20:43 gonna go to AI Studio. Before you do
20:46 that, um, since we're ignoring my sticky
20:49 note and uh Oh, sorry. I don't have
20:53 we have a we have a question in the Tik
20:55 Tok. Uh, and I might be a good time to
20:58 talk about festivist archives that are
21:01 available for free on the salon.
21:04 Yeah. So, where do we get AI lessons?
21:06 So, so a couple of places um there are
21:11 we did a thing called AI festivist back
21:13 in uh in January. The archives of those
21:17 are on the AI salon. So, if you go to
21:19 the salon.ai and say join our community.
21:23 Also, joining a community like the AI
21:25 salon, people are constantly giving what
21:27 we call LOL's learn out louds where
21:29 they're teaching teaching things to
21:31 other people. Um, so pay attention for
21:34 those in there. Um, if you want to take
21:38 um private lessons, I do private lessons
21:41 on my Tik Tok uh what's it called? Name
21:45 profile name. You can click there. Um,
21:48 you can book private hours with me if
21:50 you want to. The link, what's that? It's
21:54 a link tree. Yeah, the link tree that's
21:56 there. And the first link on it is is to
21:59 that Calendarly invite. Um,
22:06 and AI lessons is really broad. Like
22:08 what what do you like are are if you're
22:10 just getting started, join the AI salon
22:13 and tell people you're just getting
22:15 started and just start playing. We've
22:17 got a we've got a mantra in the AI salon
22:20 um called the the AI readiness cycle
22:24 which is play first
22:27 then mindfully create start building
22:29 stuff and then generously lead is share
22:32 what you're learning learn out loud that
22:35 play first thing is really important so
22:37 if you haven't done it yet just go to
22:39 chat GPT and start putting in prompts
22:42 start asking it for stuff start telling
22:44 it about who you are tell it about the
22:47 business you want to start or the idea
22:49 you have for a kids book or just
22:51 literally anything. Throw it into that
22:53 thing and watch what it vomits back at
22:55 you. If you're just getting started,
22:58 the the absolute best way to start to
23:00 get your head around what's possible is
23:02 just go play.
23:05 All right. Um,
23:09 [Music]
23:10 you know, there are lots and lots of
23:12 YouTubes, lots and lots of Tik Toks. So
23:14 I would say you know just start being
23:17 connected. The other thing I would say
23:19 is this
23:22 as quickly as possible. Right. Right. So
23:25 in the AI readiness cycle play first
23:27 mindfully create generously lead.
23:31 Start with play first then start to
23:33 build some stuff. As you start to build
23:35 some stuff based on what you're learning
23:36 there share what you're learning. Share
23:39 it on LinkedIn. Share it on Facebook.
23:41 Share it at dinner parties with people
23:43 who don't want to hear about your stupid
23:44 AI stories anymore.
23:47 Share it on the AI salon. Share it on
23:49 Twitter.
23:51 Because in sharing what you're learning,
23:53 two things happen. It anchor it anchors
23:54 what you're learning. But it also starts
23:57 to establish you as someone who's
23:59 curious about AI, someone who's AI
24:01 forward, someone who is curious and
24:04 adventurous and building.
24:07 PWC Price Waterhouse Cooper just did a
24:10 study on future of AI employment which
24:12 basically concluded we're [ __ ]
24:17 And
24:19 they also concluded that AI literate
24:22 employees today have a 56% higher salary
24:27 than nonIT employees.
24:30 So get curious and dive in. I hope that
24:34 answers the question. There's all sorts
24:35 of things in there.
24:37 All right. Um, the AI salon has people
24:41 who are just curious. Yes, correct.
24:44 Microsoft about to lay off thousands.
24:46 Correct. Yeah, I heard that, too. Um,
24:51 okay. So, here's what we're going to do.
24:57 So, when I see
25:00 when I see a prompt like this one,
25:04 a three panel
25:07 cosmic epic
25:10 thing. Let's see.
25:14 I'm gonna, you know what I'm going to
25:16 do? I'm going to do this a different
25:17 way. I'm going to take the output of
25:19 this thing
25:21 and I'm going to copy the image and I'm
25:22 going to go to chat GPT. I should
25:24 probably go to to Gemini, but it's just
25:26 not how I roll lately. All right. So,
25:29 I'm going to paste in
25:32 the image and I'm going to say describe
25:33 this in detail. describe
25:37 this in detail.
25:41 And so we're going to let it look at the
25:43 image and then it's going to tell us
25:45 what it is.
25:51 Now I'm gonna say now I want you to
25:56 create
25:58 a JSON JS
26:02 structured
26:04 prompt
26:07 um
26:12 that I can feed
26:16 to an image model
26:22 to recreate this.
26:26 Okay.
26:35 And so what JSON is, if you don't know
26:37 it, it stands for, it doesn't [ __ ]
26:39 matter, JavaScript object notation, I
26:42 think, something like that. Um, it's
26:46 just it's
26:47 Computers are good at structured
26:49 structured things. I'm not good at
26:51 writing structured things, but it's good
26:53 at that and it it understands that.
26:55 Okay. So I'm going to say now I want you
27:00 to create
27:02 five completely
27:06 new
27:09 three panel
27:12 comics
27:13 in
27:15 different styles with in in different
27:19 styles
27:22 eras.
27:24 um
27:26 narrative tone etc.
27:31 Just
27:33 write the let's see write
27:40 them out in
27:44 tech. Uh just write them out in the
27:47 browser,
27:50 not code blocks.
27:55 because I want to be able to read them
27:57 is why I'm doing that. All right.
28:01 So, here are five. The last facts
28:05 style 90s comic office comic like
28:08 Dilbert meets Black Mirror. All right,
28:12 this is cool. So,
28:15 I'm just gonna wait. But it didn't do
28:17 these in JSON. Damn it.
28:20 Okay.
28:23 Let's see. I I wasn't
28:28 clear. I
28:31 I want your prompts
28:35 in the JSON framework
28:39 you
28:40 just created before.
28:47 Oh, chat.
28:51 Uh uh uh
28:53 this is not JSON.
28:56 Stop.
28:59 Those aren't in JS.
29:04 Please make them in JS.
29:09 It's bizarre.
29:19 All right. So now I'm going to copy
29:22 this first one.
29:28 Oh, it just gave us one. All right.
29:31 Well,
29:34 copy.
29:36 And then we're going to go to
29:39 is this AI studio? Yeah. Welcome to AI
29:42 Studio
29:44 native image generation.
29:47 Is this
29:50 show me a picture of a horse running on
29:52 the beach?
29:56 Paste.
29:58 Run.
30:00 All right. So, we just gave it our JSON.
30:03 Whoa.
30:04 You're using Gemini 2.0 Flash preview
30:07 image gen, not image in 4. The model is
30:11 up in the right hand corner.
30:13 Yeah, but where's Imagine 4?
30:17 I don't know.
30:20 See the Okay, remember when I said
30:23 Google was doing better? This is where
30:25 they're not doing better. Um,
30:30 [ __ ] Where's that page? Did I get I
30:32 think I overwrote it. History. uh
30:43 free for testing in Google AI Studio.
30:46 How
30:57 it doesn't say where it is.
31:00 That's frustrating.
31:12 generate media. Imagine. Here we go. But
31:16 see, notice how the like this is this is
31:19 better. They're at least trying with the
31:22 interface, right? They've got these
31:23 little blocks of of features, right?
31:26 Here's via VO
31:29 Gemini image generation.
31:32 And then here's imagine. Okay, so here's
31:35 imagine.
31:36 So we're going to describe our image
31:38 with our little JSON. I'm going to go
31:39 bang
31:42 this. I assume this is going to take
31:44 longer.
31:48 I wouldn't exactly call that Dilbert
31:50 graphics, but
31:52 how do you zoom into it?
31:55 Copy. Large view.
31:59 Dry humor fate. Inter Galactic Express.
32:02 Okay, the text isn't perfect.
32:10 That could be a bad prompt. Um, let's go
32:12 back to chat GPT. Close this.
32:16 Let's go grab the first panel that was
32:20 not in JSON.
32:23 The last facts.
32:27 Three horizontal panels.
32:35 Okay, I think this is going to work
32:37 better.
32:45 All right, so now we have in the
32:47 non-JSON
32:49 sometimes. So, so you know, I was I was
32:51 turning those into JSON structured
32:53 things. That was probably stupid.
32:58 This didn't do three horizontal panels.
33:02 Let's see. Description
33:05 layout three horizontal panels. All
33:07 right, let's go get their example
33:09 prompt. Here's another way to do this.
33:11 Okay, so we're over here at chat jet or
33:14 at uh at at uh where they give you the
33:16 prompts for these images.
33:19 So, we can just take their prompt.
33:22 Copy. Go to chat. GPT.
33:26 I should probably go to Gemini,
33:28 shouldn't I? Yeah. So, we'll go into
33:31 chat.
33:33 We'll go Gemini Pro. Why not? Why not
33:36 use the most expensive model to do
33:38 something trivial? So, I'm going to say
33:41 take this prompt
33:49 And then um
33:51 I'm going to say
33:53 give me
33:56 five completely
33:59 new
34:01 comic
34:02 pages.
34:04 Each with same layout
34:08 but
34:10 each with
34:15 unique
34:18 style,
34:21 story, characters,
34:26 and time frame.
34:32 Oops.
34:35 Run.
34:40 Oh. Um, Gemini did something good here.
34:46 They their their submit button, I'm
34:48 pretty sure, used to be shift return to
34:51 submit their thing, and now they've made
34:54 it command return, which is good. I'm
34:56 glad they did that.
34:59 Oh, we're use we're using a thinking
35:01 model. We We're using a reasoning model
35:04 to rewrite.
35:10 Oh my god, that's funny.
35:14 All right, so the first one, the sunken
35:17 tomb. All right, we'll just copy this.
35:26 Copy.
35:28 Now we'll go to generate media. We'll go
35:29 to imagine. We'll come. Wait, which one?
35:34 All right, let's make this 3x4 vertical.
35:38 Number of results, one is fine. Imagine
35:42 for preview.
35:44 Imagine for ultra preview. Oo, let's try
35:48 ultra.
35:49 We'll throw our big ass prompt in there
35:51 that it wrote. So, this is Gemini
35:53 writing it. So, we can't blame uh what's
35:56 it called? Chat GPT now.
36:01 I Yes. Do know you have an affinity for
36:04 16 by9 and I wanted to point out that
36:07 the aspect ratio was uh something you
36:09 could change in the settings on the
36:11 right. Yeah, I know. I saw that I moved
36:14 it to tall.
36:16 It um it still is not doing three panel.
36:19 It doesn't seem to be
36:21 it doesn't seem to be getting it, which
36:24 is weird.
36:27 Huh.
36:29 And how do I Oh, yeah.
36:42 I mean, it's a good comic. It just
36:44 didn't follow the instructions.
36:48 Let me go back to Gemini.
36:52 Did I lose my chat? Oh, I did.
36:56 Do I have
36:59 I don't have history.
37:04 I don't know. All right. Whatever.
37:08 There's a new image model. Go play with
37:10 it or not.
37:14 All right. Let's let's go back and look
37:15 at uh at what what was what else was in
37:18 our uh
37:21 our Twitter direct messages. Okay. Let's
37:24 see. So, we we did that one. Ask
37:26 perplexity. You can now schedule daily
37:29 tasks, set reminders, and more with
37:32 Perplexity on WhatsApp. Okay, cool. And
37:34 there's a rumor right now that Apple is
37:36 trying to is considering buying
37:38 Perplexity
37:42 because
37:44 with $17 trillion in cash, apparently
37:48 they couldn't afford to make their own
37:50 competent AI model. It's like I don't I
37:55 don't know.
37:56 like
37:58 so frustrating as an exapple fanboy.
38:03 All right. Um Oh, yeah. Chat GPT is is
38:07 still adding like a million members a
38:09 day on
38:12 the app store.
38:14 And then yeah, this thing anthropic just
38:16 made every Claude user a no code app
38:19 developer. This one's wild
38:23 from Venturebe.
38:29 Oh my god, these
38:33 How do you get rid of this shitty
38:35 advertising?
38:43 Okay.
38:46 Anthropic announced Wednesday that it
38:48 will transform Claude
38:51 Claude AI assistant
38:54 into a platform for creating
38:57 interactive
38:59 sharable applications, marking a
39:01 significant evolution from
39:03 conversational chat bots toward
39:06 functional software tools that users can
39:08 build and distribute without coding
39:10 knowledge.
39:12 Wasn't that just artifacts? Like how is
39:14 this different than artifacts? Does
39:15 anyone know?
39:17 The venture beat always crashes for me.
39:20 History's on the left rail. Oh wow.
39:22 Okay.
39:24 Um, the San Francisco AI company
39:28 revealed that millions of users have
39:30 already created things using artifacts.
39:32 That's what we just said. So, what does
39:35 this mean?
39:40 Since the features
39:42 initial launch. Okay. Now, Anthropic is
39:46 embedding Claude's intelligence
39:48 directly into these creations. Oh, okay.
39:54 I think I get it. So, in your chat GPT
39:58 or in in your cloud apps that you
40:00 create,
40:02 you can now have them do
40:05 AI stuff, I guess,
40:08 enabling them to process user input and
40:11 adapt content in real time independently
40:14 of ongoing conversations. Oh, okay.
40:17 That's cool. Okay,
40:20 I sort of get that. I think we're going
40:22 to go play with this. The development
40:24 represents a fundamental shift in how
40:26 artificial intelligence interfaces with
40:28 users, moving beyond static responses
40:31 toward dynamic interactive experiences
40:34 that blurs the line between AI
40:36 assistance and software development. So,
40:38 the the march toward vibe coding
40:41 continues forward. Anyone like if you're
40:44 in here and you're a developer and
40:46 you're pissing all over vibe coding, any
40:49 anyone that's on a soapbox about it's
40:51 not real coding, just give it give it a
40:54 year.
40:57 These uh these vibe coding tools,
40:58 there's just too much competition here
41:00 for these things not to get really
41:02 really good. Um Anthropics was already
41:05 pretty damn good. And the fact that you
41:06 can now, sorry I schmutz, the fact that
41:10 you can now embed um
41:15 essentially large language model
41:17 functionality into your apps is is
41:18 pretty amazing.
41:20 Um the move intensifies competition with
41:23 OpenAI's canvas feature. I think it
41:26 intensifies competition with things like
41:29 lovable and replet, but whatever.
41:33 similar split screen functionality.
41:36 How Cloud's artifacts eliminate copy
41:39 paste problem plaguing workflows.
41:47 While offering free app creation makes
41:50 business sense for Anthropic, the
41:53 platform operates on Cloud's existing
41:56 infrastructure with users authenticating
41:58 through cloud accounts
42:00 and accessing shared applications.
42:04 This approach distributes computational
42:06 load. Okay, I don't care about that.
42:08 Free users can create, view, and
42:10 interact with artifacts, while pro 20
42:13 bucks a month gain additional
42:16 capabilities and higher usage limits.
42:18 The company views free access as a comp
42:21 customer acquisition strategy.
42:31 I don't I still don't get how is this
42:33 different than what they had. Okay. So,
42:36 here we are on I'm at claude.ai. Okay.
42:40 And then when I when I got here,
42:43 we're Brandon and I were looking at this
42:44 before this show. Ow, my eye.
42:50 Itchy crap. Um,
42:54 hey. Hey, uh, K. Yeah. So the I think
42:58 bullet number two is the big win here is
43:02 we're giving the creators the ability to
43:04 use Claude's API from within the
43:07 artifact itself. Yeah, that seems
43:10 opposed to just building something and
43:13 deploying it and it's static. Now you
43:16 can generate responses within that
43:18 artifact. Yeah, that seems to be the big
43:20 deal. like it. That was a bizarrely
43:24 written article because a lot of what it
43:26 described is what artifacts already
43:29 does. So, yeah. So, I think this is it.
43:33 But also, I guess the browse the gallery
43:35 and customize. I don't know if they had
43:36 a gallery before, but let's start
43:38 exploring. Oh, okay. Here we are.
43:42 Oh, so there's a new
43:44 This is a standalone artifacts
43:48 page.
43:52 So, this is claude.aiifacts.
43:55 Interesting. And then my artifacts.
43:59 Oh, look. There's the Hollywood time
44:01 machine.
44:04 This is the the vibe coding I did with
44:06 did with uh Rick Rubin's vibe coding
44:10 project.
44:12 Click anywhere to start audio.
44:27 bad.
44:29 All right, let's see. Home. No. What's
44:33 this?
44:35 Claude
44:37 back
44:40 inspiration.
44:42 Power your artifacts. Artifacts with
44:44 claude API. enable this feature.
44:50 Okay, I guess it's enabled.
44:54 New artifact.
45:01 Creative projects.
45:09 I didn't do anything yet. Should I not
45:12 have clicked those things?
45:15 Okay,
45:20 that's just nothing. That just said
45:23 that's just prompting me to answer some
45:25 questions to generate something. So, I'm
45:26 going to say
45:28 um
45:34 create
45:36 an app
45:39 that looks
45:44 for
45:48 The latest
45:52 news story.
46:07 I wonder if it can do that. Might not be
46:08 able to do that.
46:12 Okay. Create an app where a user
46:21 puts in
46:25 this is good good coding language where
46:27 a user puts in
46:31 an animal
46:38 um
46:42 a here.
46:43 And hey, does anyone know does anyone
46:47 remember does Claude have a built-in
46:49 image generation model?
46:52 Puts in an animal, a year, and a
46:55 location.
47:00 And
47:02 the app
47:04 generates
47:06 a poem.
47:11 Um, let's see. A poem,
47:15 a short
47:17 story. They don't. Okay.
47:21 And
47:32 a poem, a short story,
47:37 and a greeting card.
47:42 concept.
47:44 I don't know. That's the stupidest app
47:46 I've ever thought of. It's so bad.
47:49 But but
47:51 in order for it to generate something,
47:54 it's got to use
47:58 Claude. What a delightful and creative
48:00 concept. Well, thank you, Claude, for
48:02 blowing sunshine straight up my booty. I
48:06 love the idea of generating multiple
48:07 creative pieces from just three simple
48:09 inputs. Here's the plan.
48:13 User friendly fields, three distinct
48:16 output sections, design approach,
48:20 creative gen. Does this align with your
48:21 vision? Yes. Go.
48:25 And if you haven't seen Claude do this
48:28 in the coding world, we say inputs. No,
48:31 it's the it's the chat hole, Joker. We
48:34 don't say inputs, we say chat hole.
48:45 If you're new to to AI stuff,
48:51 the stuff on the right, it's writing
48:53 code for us, which if you've never
48:55 watched it do that, is kind of
48:57 mind-blowing.
48:58 It still blows me away that we're now
49:00 living in a world where code can just be
49:03 generated like that.
49:10 Coders aren't happy about it.
49:18 It doesn't do good code. It's garbage
49:21 code.
49:23 Not for long.
49:39 Oh, yeah. While that thing's coding. Oh,
49:42 made it. All right. We'll come back to
49:43 that.
49:45 We'll see if No, we'll do we'll do this
49:46 now since it came here now. All right.
49:48 So, animal, fox, elephant, butterfly.
49:50 We're going to go um turtle
49:55 year. Um
49:58 1954
50:01 location
50:05 LA
50:08 generate creative stories. Is this all
50:10 we have here? Yeah.
50:16 Poem
50:17 in LA. When 1954 was young, a turtle
50:20 danced where songs were sung. Through
50:22 seasons old and stories new, a tale of
50:25 wonder grew and grew.
50:28 With gentle steps and hearts so bright,
50:31 the turtle brought pure delight. I I'll
50:34 tell you what,
50:36 kids and and ladies and gentlemen and
50:38 and uh Nardo Wells
50:42 1944 beaches of Normandy. Oh my god.
50:46 1944
50:48 beaches. And we'll do we'll do uh let's
50:50 see. We'll do seagull. Seagull
50:56 location beaches of Normandy.
51:01 That's actually a good one. Norman
51:04 D. Um because uh like we'll get to see
51:09 if it's got some historical context in
51:11 here. It should, right? It's got a giant
51:14 large language model sitting on top of
51:16 it. All right.
51:19 Once in beaches of Normandy long ago in
51:22 1944 where flowers used to grow a
51:26 seagull wandered wild and free living
51:28 peace of poetry. The years have passed
51:31 but still they say the seagull blessed
51:33 that special day. Beaches of Norm
51:36 Normy's heart it dwells and every season
51:39 softly tells. All right.
51:42 In the summer of 1944 the beaches of
51:44 Normandy was unlike any other. Wildlife
51:47 photographer came.
51:50 Greeting card from the beaches of
51:51 Normandy with love inside. Like a
51:55 seagull wandering through time since
51:57 1944.
52:00 Your friendship has brought so much
52:02 beauty to my world. Thank you for being
52:05 wonderfully you.
52:09 That is [ __ ] horrific. Uh you could
52:12 obviously refine this thing to not suck.
52:14 Uh, okay.
52:18 Let me talk about this. I think this is
52:20 a pretty big deal
52:27 with vibe coding apps like lovable and
52:31 replet and things like that.
52:35 If you know what an API is,
52:38 and if you're clever enough to go find
52:40 one and sign up as a developer and get
52:43 yourself an API key,
52:46 and if you figure out that sometimes
52:48 they cost money and sometimes they
52:50 don't, and if you can find the one that
52:53 does what you want it to do,
52:55 then you can create an application that
52:59 makes an API call out to say OpenAI to
53:02 generate
53:04 some AI goodness for you.
53:07 The fact that I didn't have to do that
53:09 here is a big deal, right? The fact that
53:13 I can just make an app that uses Claude
53:17 Sonnet 4 as the generating engine. I
53:19 assume that's what it using and then I
53:21 can put that out in the world. And
53:23 basically what they said in that article
53:25 was
53:26 you can't use these applications without
53:29 a
53:31 anthropic user ID.
53:34 So this is a customer acquisition
53:36 strategy. I think it's pretty smart.
53:38 Black bar. I know
53:41 that's me. A nar do well. I know there's
53:44 a lot of nerd do wells in here. Grew and
53:46 grew should have made it. Hi coup
53:50 puts it in the chat hole.
53:54 All right. Um so those so so listen
53:57 shit's going on.
54:01 Here's what I can tell you and promise
54:03 you. If you're feeling overwhelmed and
54:07 if you're feeling like, I don't know
54:09 what to do next,
54:15 welcome.
54:17 Welcome to the club.
54:20 This is what it's like living in an
54:22 acceleration.
54:25 I I don't know that we've ever like in
54:29 my life I've lived through bursts of
54:31 innovation,
54:33 but the bursts of innovation tend to
54:35 happen over like longer
54:39 chunks of time.
54:45 This we're in a we're in a whole new
54:47 we're just in a whole new game. All
54:49 right. Um let's go play with Mid
54:52 Journey. Okay. So, here's what
54:58 there were two things that I saw that
55:00 that kind of captivated me.
55:03 Oh, did these work actually?
55:13 No,
55:15 that one didn't.
55:23 No.
55:34 Okay.
55:35 So, let me go show you something.
55:40 Um, my mind
55:46 [Music]
55:52 Okay, so this guy did this prompt
55:57 that made a really interesting
55:58 midjourney video.
56:02 So, see how the kid starts out in the
56:04 clouds with fish and then he runs and
56:06 now he's on land and there's fire and
56:08 there's no fish. So, he wrote a prompt.
56:12 A boy is dreaming. He opens his eyes and
56:14 starts running away. jumps over fire,
56:17 catches a bird while fish disappears as
56:20 he opens his eyes. That's a weirdly
56:23 written prompt, but
56:37 All right. So, let's go to midjourney
56:38 and we'll say
56:42 create. We'll say
56:45 um
56:48 claimation style
56:55 image of
56:57 a boy in the clouds
57:05 with fish
57:08 swimming
57:10 above him.
57:13 magically.
57:18 And if you don't know where I am, I'm in
57:22 uh MidJourney, midjourney.com.
57:25 If you haven't played with MidJourney
57:27 before, I think it is quite worth it.
57:42 All right. So, we're going to get some
57:44 interesting little little things here.
57:54 All right.
57:57 Which one of these do we want to
57:58 animate?
58:01 I think it's that one.
58:03 Let me do variations on this one. We'll
58:05 do subtle variations on this one.
58:19 88 complete. 94 complete. 100 complete.
58:23 Okay.
58:25 He doesn't have a face. He sort of has a
58:27 face. He's cute. He's weird. All right.
58:31 We're gonna go with this dude. Okay. So,
58:34 how you turn this into an animation is
58:38 you go either low or high motion and
58:41 then you do it auto or manual. Since we
58:43 have this prompt that that guy wrote,
58:44 we're going to go high motion.
58:47 And then instead of the prompt that was
58:49 there, we're going to paste in a boy is
58:52 dreaming. He opens his eyes and starts
58:54 running away. Okay, here we go. Bang.
59:06 submitting.
59:08 All right, let's scroll down here.
59:16 Yeah, these are things I did earlier.
59:17 Oh, that looks great. God, the the I'm
59:20 telling you, man, the the
59:23 quality of these animations in
59:26 Midjourney is
59:29 stunning.
59:34 Look at the parallax there.
59:38 Parallax is where things in the
59:40 foreground are moving faster than things
59:41 in the background.
59:46 Gorgeous.
59:59 Wow.
1:00:10 If I do have midjourney, what's another
1:00:12 affordable Oh, if you don't have
1:00:14 midjourney, what's another affordable
1:00:16 option?
1:00:18 Um,
1:00:19 well, there's
1:00:24 for for video
1:00:27 there there's really none. They're all
1:00:29 pretty expensive.
1:00:32 There might be some open- source ones. I
1:00:35 think
1:00:38 one of I think Korea might give you
1:00:43 some amount of credits a day.
1:00:51 Brandon just said that V3 is in paid
1:00:54 Canva. So, if you've got a a paid Canva
1:00:56 account, it's in there. All right, let's
1:00:59 go see if our our little boy did what
1:01:02 he's supposed to do.
1:01:06 There he's running.
1:01:09 There's the birds and no fish.
1:01:12 He's running. The fish go away. There's
1:01:15 the fire. jumped over. There's the bird.
1:01:17 Wow. So, yeah. So, you can definitely
1:01:24 And now if we we can extend this thing
1:01:27 now.
1:01:31 Extend video.
1:01:34 We're going to do low motion.
1:01:40 A boy
1:01:44 picks up
1:01:50 three colored
1:01:53 balls of yarn
1:01:57 and starts
1:01:59 juggling. Juggling is notoriously hard
1:02:02 for these things to do.
1:02:05 So, let's see if we can get it so that
1:02:07 he after this little running thing,
1:02:11 he starts juggling.
1:02:18 That's pretty bonkers.
1:02:21 Pretty bonkers.
1:02:38 All right.
1:02:41 35% complete.
1:02:44 The fact that it does four four videos
1:02:47 at a time, this is like a fifth the cost
1:02:51 of any other video model and it's uh you
1:02:56 get four of them for every generation.
1:03:02 Oh, look. It does look like he's
1:03:03 juggling. Paid Canva V3 is five
1:03:06 generations per month. Okay, that's not
1:03:09 really That's not really all that
1:03:10 useful.
1:03:29 92% complete.
1:03:37 All right, let's see what we got.
1:03:40 So, he's running. The fish are
1:03:41 disappearing. He jumps over the fire.
1:03:45 And now he's juggling two things. And
1:03:47 the fish are fascinated.
1:03:50 All right, now he runs. This part's all
1:03:54 the same.
1:03:56 There he's got that the birds twirling
1:03:59 on his head. I wouldn't exactly call it
1:04:01 juggling, but you know,
1:04:09 he's juggling the fish there.
1:04:18 All right. So, the most successful one
1:04:21 I think was the first one.
1:04:26 Wait, this one
1:04:30 runs around,
1:04:32 gets behind there. No, that was not the
1:04:34 right one. This one.
1:04:46 Okay. All right. So, he juggles the
1:04:49 ball. So, now we're going to extend it
1:04:50 again. Okay. So, this is the one we
1:04:53 like. So, we're gonna go extend high
1:04:55 motion. And then we're gonna go a boy um
1:05:00 puts on a top hat
1:05:06 and
1:05:08 walks onto
1:05:14 a circus.
1:05:19 ring like the ring master. Ring master.
1:05:24 Isn't that what they're called? All
1:05:25 right, go.
1:05:31 So, you can do this four different
1:05:33 times, right? So, you've got an original
1:05:35 5-second,
1:05:37 you've got an original 5-second video,
1:05:40 and then you've got four 4-se secondond
1:05:42 extensions.
1:05:44 And you can prompt each of those
1:05:46 extensions to do something different.
1:05:49 So you can create a 21 second film
1:05:52 that's got different story segments in
1:05:55 it. That's me juggling decisions. New
1:05:59 movie called The Social Network 2 come
1:06:02 out blasting Facebook's negative impact.
1:06:05 Yeah, that would be good. Or is that
1:06:07 real?
1:06:17 Oh, look. He's putting on the top hat.
1:06:19 Very cool.
1:06:24 Oh, there's his little his little cape
1:06:26 outfit. This is good. I'm telling you,
1:06:29 man. Midjourney's got something figured
1:06:30 out here that that other other models
1:06:33 haven't quite cracked.
1:06:39 There's something about the lighting in
1:06:41 this and the shadows, the consistency of
1:06:44 it. It's just good.
1:06:50 92% complete. All right, here we go.
1:06:55 There's our little dude running. Now
1:06:58 he's going to juggle a little with the
1:07:00 propeller bird wing head thing. Oh, fish
1:07:04 just turned into a hat. I like it.
1:07:07 Okay. And so now, oh wait, we got to
1:07:10 watch the other ones, right?
1:07:15 So there's this.
1:07:17 This is kind of cool, right? It keeps
1:07:20 that part's all the same. And now it's
1:07:22 just going to change what happens here.
1:07:23 There's this top hat.
1:07:27 All right. Let's see. Here's this one.
1:07:31 It would be nice if he went into a
1:07:33 circus ring.
1:07:39 He just put a fish on his head.
1:07:43 All right, last one.
1:07:56 All right. So,
1:08:07 so we're going to go high motion and so
1:08:10 we're going to go
1:08:12 let's see
1:08:16 camera
1:08:19 moves.
1:08:21 Oh, thank you for the the gift. Uh, Red
1:08:24 Dead
1:08:25 Red
1:08:27 Ad Admiral, thank you for the gift. I
1:08:29 appreciate that. Um, camera moves behind
1:08:33 the boy
1:08:37 and reveals that he is
1:08:44 he is looking
1:08:47 at a large
1:08:51 audience of sea creatures.
1:08:58 Watching watching
1:09:02 his circus
1:09:06 performance
1:09:09 high motion Tik Tok pin. Now we need
1:09:12 some music for it for Suna. We do
1:09:14 actually.
1:09:17 [Applause]
1:09:27 I'm sleepy.
1:09:29 Champy. You sleepy? Yeah, of course
1:09:31 you're sleepy. You're sleeping.
1:09:33 The definition of sleepy.
1:09:38 There's no sleeping in AI.
1:09:42 I actually turn off the air. It's kind
1:09:44 of nice out.
1:09:50 [Music]
1:10:09 What I don't know here.
1:10:11 Well, we'll find out. This is why we do
1:10:13 this. We'll see if Oh, well, yeah, we'll
1:10:16 see if the camera actually turns around
1:10:17 behind him and shows the audience. That
1:10:20 would be swell if we got that. Looks
1:10:23 like it's trying to do something.
1:10:29 down.
1:10:32 It didn't understand my my instructions.
1:10:34 So, one of the things you will learn
1:10:36 when you start playing with AI is if
1:10:38 you're shitty at prompting, stuff
1:10:39 doesn't look good.
1:10:46 [Music]
1:10:51 All right, let's go. Let's go to the
1:10:54 videape.
1:11:06 And
1:11:08 it didn't move behind him, but it
1:11:10 reveals curtains. Okay.
1:11:15 Running.
1:11:19 Watch what I do with this fish.
1:11:23 Yeah, that's what I wanted.
1:11:27 Not quite an audience, but it was a good
1:11:31 camera move. All right, let's see what
1:11:33 this one does.
1:11:43 The other thing that I will say, if
1:11:45 you're watching this and you're like,
1:11:46 you know, you could have done that in
1:11:47 these six other tools, you're probably
1:11:49 right.
1:11:51 But
1:11:53 this idea of being able to tell longer
1:11:55 form stories with these extensions is
1:11:58 fascinating to me.
1:12:03 [Music]
1:12:09 So the one we got that was close was
1:12:12 this one,
1:12:16 right?
1:12:19 And then I've got one, two, three, four.
1:12:30 No,
1:12:32 this one.
1:12:41 [Music]
1:12:47 Okay, I got one. I got one. Okay, you
1:12:51 guys are going to hate me.
1:12:58 Um,
1:13:01 a giant felt shark
1:13:06 um swims out of the audience portal
1:13:15 and
1:13:18 swallows
1:13:20 the boy.
1:13:22 follows the boy in one gulp. Listen,
1:13:28 this is a late night show. If you if you
1:13:32 were expecting this to have a happy
1:13:33 ending, I'm I'm sorry.
1:13:41 All right, here we go. Let's see if we
1:13:44 can get a shark to eat our
1:13:48 our protagonist.
1:13:55 Oh, this is dark. I remember poor Frank,
1:13:57 too.
1:13:59 His dream became a nightmare.
1:14:05 The shark.
1:14:10 Listen, I think what you all know here
1:14:13 is this channel is really just an
1:14:15 opportunity for me to entertain myself.
1:14:18 That's the goal. That's the purpose.
1:14:21 You all just happen to be weirdass
1:14:24 voyers. We're not weird. We're
1:14:25 irregular. I know.
1:14:31 It doesn't change the fact that Oh,
1:14:33 there's the shark. Look at the shark. Oh
1:14:37 my god.
1:14:46 Oh my god.
1:14:49 Oh my god. It's so bad.
1:15:00 Oh my god. Oh, it's so good. It's so
1:15:04 good.
1:15:08 This is one now that's actually really
1:15:10 hard to do sound for because I think
1:15:13 we'd have to drop it into
1:15:17 like you could do a song in suno
1:15:21 but then I think you need to do another
1:15:23 song where it's just screechy like you
1:15:27 know some sort of Jaws kind of
1:15:29 terrifying like this should all be
1:15:34 everything's fun everything's
1:15:36 Everything's fun. Hey, look at the magic
1:15:39 circus. Hello audience. What's
1:15:42 happening?
1:15:53 I don't know why this gives me such joy.
1:16:03 Oh my god.
1:16:05 Yeah. Yeah. Look at our fantastic
1:16:07 audience over here. I think everything's
1:16:09 going great. Oh no, Mr. Bill. Oh yeah,
1:16:12 he's still there.
1:16:14 Let's see if it actually gets him. Oh my
1:16:18 god, this is funny.
1:16:21 It's super cute. It is, but the boy's
1:16:24 still there at the end. Exactly.
1:16:27 Oh, you could This piece could be called
1:16:29 False Alarm.
1:16:32 All right.
1:16:34 still there.
1:16:37 He's got to eat him. Did they not eat
1:16:39 him?
1:16:44 They really should just Okay, so I I
1:16:46 know what we'll do.
1:16:57 It sort of knocked them over. Okay, so
1:17:00 here's what we're going to do. We're
1:17:01 going to go.
1:17:03 There's the prompt. There's the prompt.
1:17:07 Here's the
1:17:10 There's our There's our starting point
1:17:12 video.
1:17:14 So, a giant felt shark swims out of the
1:17:16 audience portal and swallows the boy in
1:17:18 one gulp,
1:17:21 making him
1:17:25 disappear.
1:17:33 All right.
1:17:36 And then where's my stylization at? It's
1:17:38 low.
1:17:40 Maybe I should put that a little higher.
1:17:42 We'll try We'll try a second one with
1:17:45 stylization higher.
1:17:47 So, bang, bang, bang.
1:17:52 Can I not use this?
1:17:55 There we go.
1:17:57 So, we'll try a second.
1:18:17 The boy still standing there is a good
1:18:20 jank. It's good janky. It's good janky.
1:18:32 I was going to put in a pool of blood,
1:18:34 but my sisters and I were freaked out.
1:18:38 Oh, man. Shark tanked. Oh, yeah. That
1:18:41 would be good.
1:18:44 The AI once do no harm. Yeah, you're
1:18:46 right. It does.
1:18:49 All right. Look, there he's there. He's
1:18:51 getting eaten in that one in the upper
1:18:52 left.
1:18:55 No cartoon characters. We're harmed in
1:18:57 this. Now, now we love a little jank.
1:19:02 We've come a long way, baby. Yeah. Well,
1:19:05 like
1:19:07 I don't like him not being there at the
1:19:09 end. Like, he should be gone. This
1:19:12 should be a heartbreaking story of
1:19:16 of of of cloud play.
1:19:23 All right, here we go. This is actually
1:19:26 remarkably fast, too, right? Like how fi
1:19:28 fast it's doing these things.
1:19:31 It's making four videos that fast.
1:19:35 And here comes the shark.
1:19:41 It bit his head off, but his head's
1:19:42 still there. Oh, wait. I got to change.
1:19:45 There we go.
1:19:47 One, two, three, four. Yeah.
1:19:51 And so this is the last extension we can
1:19:53 do. You can't do any after this.
1:20:07 [Laughter]
1:20:12 Oh, wait. Got
1:20:16 [Music]
1:20:36 That one was pretty good. It got his
1:20:38 head and got another fish. A fish swam
1:20:40 in his mouth. That was pretty good.
1:20:42 Let's watch this again.
1:20:45 It pauses a little bit too long right
1:20:48 here. There's kind of nothing happens.
1:20:51 Maybe there you could have a pause there
1:20:52 and then he comes out of nowhere and
1:20:54 eats that and grabs his head and takes
1:20:56 the boy away. I think that's pretty
1:20:58 good. All right, let's see the last one.
1:21:00 There he is. Jumps over the fire. Comes
1:21:03 in here, juggles,
1:21:06 puts on his top hat.
1:21:19 [Laughter]
1:21:23 That's perfect. All right, that's the
1:21:25 one.
1:21:30 Oh my god.
1:21:33 What do they call that thing? A circus
1:21:34 collopy.
1:21:36 Freddy,
1:21:38 be gentle.
1:21:40 He gets booted by a by a goldfish. I
1:21:43 like that. It's It's like one of his
1:21:45 little buddies coming in to save him,
1:21:47 but it's too late.
1:21:54 Oh my god.
1:21:57 Everything's good.
1:22:03 That's hilarious.
1:22:07 All right.
1:22:09 Yeah, this one this one actually needs
1:22:10 kind of sound design and I don't think
1:22:14 we have enough time for that tonight.
1:22:15 But let me go um
1:22:19 so if you don't know Sununo sununo.com
1:22:23 this going to make make a song.
1:22:26 We're going to do an instrumental only.
1:22:28 We're going to do Oh, I've got a Oh,
1:22:30 this has got the weird god damn this
1:22:32 thing.
1:22:35 Um, Dark Reader. Thanks, Pate, for let
1:22:38 letting us know about Dark Reader, but
1:22:40 it's weird. It's slow.
1:22:53 Okay.
1:22:56 Yeah. Okay. Um,
1:23:01 underwater
1:23:04 circus. How do you spell colliopy?
1:23:08 Vicki
1:23:10 coy.
1:23:17 Okay, got it. I was close. Just two L's.
1:23:20 Clyopy. Jeff Flanigan for the win.
1:23:23 Silverf Fox got it. Dr. Jay's got it.
1:23:26 Danielle's got it. Okay, smarty pants.
1:23:29 No need to show up the host with your
1:23:32 fancy word spellings. All right.
1:23:35 Underwater circus collopy.
1:23:38 Um,
1:23:39 with a dark
1:23:42 undertone.
1:23:48 All right. And let's we'll generate six
1:23:50 of them.
1:24:03 day
1:24:06 franchic Bob. Hey Bob. Yeah, tell them
1:24:10 what they've won. Hey, they haven't won
1:24:12 anything. I hate to break it to you
1:24:14 there, Jimbo, but uh this is not a game
1:24:16 show. It's a it's a it's a learning lab.
1:24:19 It's not even really a lab. It's there's
1:24:22 there's some sort of educational content
1:24:24 here, but it's mostly well it it it's
1:24:28 what appears to be an attempt at
1:24:31 entertainment more than education, I
1:24:33 would say. All right. Fantiastic. The
1:24:37 drowning carnival.
1:24:40 [Music]
1:24:48 I kind of like that.
1:24:52 [Music]
1:24:58 takes too long.
1:25:00 [Music]
1:25:11 That's really good.
1:25:23 That's really good, too. All right.
1:25:30 Oh,
1:25:33 maybe this was is what comes in.
1:25:41 Maybe we crossfade between
1:25:43 [Music]
1:25:56 That's interesting.
1:26:04 That takes too long.
1:26:22 God damn. I these these tools are
1:26:25 [ __ ] remarkable.
1:26:31 Do you guys get
1:26:34 how lucky we are to be in a time in
1:26:38 history where we get to play with this
1:26:39 [ __ ]
1:26:41 It's [ __ ] staggering.
1:26:44 [Music]
1:26:46 Yeah, that's what we're going to go to.
1:27:00 [Music]
1:27:06 [Music]
1:27:13 [Music]
1:27:17 Why is he singing like that? It's really
1:27:19 annoying. I I was enjoying the thing
1:27:22 when he was telling us how to do things.
1:27:24 It's called the AI learning lab and I
1:27:25 would expect him to be teaching
1:27:26 something and the curriculum was going
1:27:29 fine except for the dark twist it took
1:27:31 at the end there. I don't know why he
1:27:32 has to have the thing eaten by a shark.
1:27:33 Seems a little aggro. Anyway, and then
1:27:37 he just started singing and then like
1:27:39 what I what is this channel? Who are you
1:27:42 people?
1:27:44 Why do you watch this?
1:27:49 It's very confusing.
1:27:51 All right.
1:27:55 [Music]
1:28:08 Tom Noddler. My 5-year-old won't
1:28:10 remember a time when she couldn't
1:28:12 generate animations herself of a dragon
1:28:14 all set to music. I know.
1:28:17 It's It's just crazy.
1:28:22 It's
1:28:24 Like the fact that we were able to
1:28:27 like part of what I wanted to experiment
1:28:29 with tonight with MidJourney was
1:28:32 like I'm not doing any kind of
1:28:36 real prompting here,
1:28:39 right? Other than I don't know, make the
1:28:41 guy juggle balls, turn the thing into a
1:28:43 hat, show the audience, and then make a
1:28:47 shark come out of that thing.
1:28:49 [Laughter]
1:28:59 It's just it's just quite remarkable.
1:29:01 iMovie.
1:29:03 Is that right? I movie.
1:29:10 I like the combination of the first tune
1:29:14 that transitions to the ominous sounds.
1:29:16 Yeah, I think that's going to work. I
1:29:17 think it's going to work, actually.
1:29:19 All right. So, let's let's get into our
1:29:23 there's our previous movie we made in
1:29:24 here.
1:29:26 All of my all all of my projects are
1:29:29 called my movie and then a number
1:29:35 cuz I'm a [ __ ] idiot.
1:29:41 Do not follow me for file naming
1:29:44 strategy.
1:29:48 Okay. Um,
1:29:50 where am I going? Downloads, right? I
1:29:53 made a song, too. Okay, cool. Let me go
1:29:56 grab these.
1:29:58 Drop them in there. All right. So, we'll
1:30:00 put this down here. And then this is the
1:30:03 one. Yep. That one goes down first.
1:30:08 Let's see how it plays.
1:30:12 [Music]
1:30:17 I think it should start.
1:30:21 Yeah.
1:30:29 Yep. Okay.
1:30:32 [Music]
1:30:48 Okay, right about there is where it
1:30:50 needs to get ominous.
1:31:04 [Music]
1:31:10 [Music]
1:31:11 Okay. So, we'll do that
1:31:17 and then we'll grab this and move it
1:31:20 down. I can move this below it, right?
1:31:22 Or above it. Yes, above it. Okay.
1:31:29 So,
1:31:35 [Music]
1:31:44 Okay. So, this can't start until
1:31:48 Oops. So, we go right and then maybe
1:31:52 there.
1:31:57 Okay. So, here.
1:32:00 Here we put a key frame in.
1:32:03 And then here we put a key frame in. Did
1:32:06 I put a key frame in? No.
1:32:09 Oh, no. I know what we do. We do.
1:32:13 [Music]
1:32:25 Can I change that curve? I don't think I
1:32:28 can. So, we're going to we're not going
1:32:29 to use that thing. We're going to use a
1:32:31 key frame.
1:32:33 And we're going to move that over there.
1:32:34 And we're going to move that bang down
1:32:36 to zero.
1:32:38 Maybe this comes out to here.
1:32:44 And then here we're going to do a ramp
1:32:46 up.
1:32:48 And then I think we do it something like
1:32:50 that.
1:32:53 [Music]
1:32:54 Not there. It's going to be later.
1:33:01 [Music]
1:33:03 It's close.
1:33:07 [Music]
1:33:22 It's just stupid.
1:33:32 All right. Close. Close. Okay.
1:33:35 this. We're doing some precision timing
1:33:37 here.
1:33:43 [Music]
1:33:57 It's so bad. It's so bad. It's so bad.
1:34:01 It's good.
1:34:07 Oh my god, this is [ __ ] awesome. Um,
1:34:12 I think we do this and then we do a
1:34:14 transition to black. So, we're going to
1:34:16 go to background
1:34:18 black.
1:34:20 Drop that bad boy in there.
1:34:24 Then we'll do a transition of cross
1:34:27 dissolve between those two puppies.
1:34:30 And then I think we'll do a little
1:34:32 title.
1:34:34 Little title because we did that the
1:34:37 other day per Brandon suggestion. I
1:34:39 thought it was quite good.
1:34:43 Um
1:34:46 All right. And then we'll go titles. It
1:34:49 was this one. I think we can use this as
1:34:52 our our forever
1:35:00 So this can say
1:35:03 AI learning lab and this can say
1:35:09 I think this down here was by Kyle
1:35:13 Shannon and then I think up here oh we
1:35:16 said thanks midjourney we got to say
1:35:18 more mid
1:35:21 journey
1:35:23 magic
1:35:27 Oh, maybe it's the title. Maybe it's the
1:35:29 title. Um,
1:35:33 what should it be called? Bad audience
1:35:35 member. Um,
1:35:39 my worst critic.
1:35:48 That's good.
1:35:51 We'll put that in quotes. I think I
1:35:53 think what we should do is we'll do um
1:35:55 AI learning lab up there
1:35:58 and then down here we'll do my worst
1:36:00 critic.
1:36:02 My worst
1:36:04 critic.
1:36:06 [Laughter]
1:36:12 Oh yeah, that's good. by Kyle Shannon
1:36:16 uh with
1:36:20 Mid Journey
1:36:23 and Suno.
1:36:27 Okay.
1:36:29 So, and we got to have the uh this taper
1:36:33 have that end there. We'll have it taper
1:36:35 out there.
1:36:37 All right. Let's look at our movie.
1:36:39 Let's look at our movie, peoples.
1:36:42 We'll go full screen.
1:36:44 [Music]
1:37:02 [Laughter]
1:37:14 print it.
1:37:18 Fade out to Wait, the fade out of the
1:37:21 collap needs to be faster. Faster later.
1:37:24 Okay. All right. Fine. Fine. Faster
1:37:28 later. Okay. Let's see. So, wait. Where
1:37:31 did I do that? Hang on. So I got to
1:37:34 shorten these. Yeah. So it needs to be
1:37:37 faster later.
1:37:40 So
1:37:46 maybe like that.
1:37:53 Yeah, that's good. Okay, that works. All
1:37:56 right. And then
1:37:58 [Music]
1:38:01 [Laughter]
1:38:06 And then wait, we got to we've got to
1:38:08 fade this out a little sooner
1:38:11 so it goes completely to black. Yeah,
1:38:13 there we go.
1:38:15 [Music]
1:38:22 [Laughter]
1:38:26 We could We could extend that, couldn't
1:38:27 we? Yeah.
1:38:32 And then extend the music a little bit.
1:38:34 Yeah.
1:38:44 [Music]
1:38:57 All right.
1:38:58 Oops.
1:39:00 Let's see. Move that a little more. Like
1:39:02 that. Okay. So, we're going to go export
1:39:06 share file.
1:39:14 So, this is going to be called fish
1:39:16 circus. Oh, wait. No, it's called my
1:39:18 worst. My worst critic. My worst critic.
1:39:30 That is so good. Thanks, Gavin. It's
1:39:32 pretty good, right? So, is all that
1:39:34 editing within MidJourney? No, Silver
1:39:36 Fox. So, sorry about that. I jumped over
1:39:38 to iMovie. So, basically what I created
1:39:42 was in Midjourney, I created the five
1:39:46 segment single video, right? So it was
1:39:49 the the opening thing it created and
1:39:51 then I extended it four times and the
1:39:53 fourth extension was the shark.
1:39:56 And then and then I did two songs in the
1:39:59 first Collia one and then the ominous
1:40:01 one. Um and then I mixed all three of
1:40:04 those together in iMovie which did the
1:40:07 titles and the being able to fade
1:40:09 crossfade the music and things like
1:40:11 that. All right.
1:40:13 Fish and chips.
1:40:15 Okay. So, I think that's I think that's
1:40:18 out there. Now, this is going to be
1:40:20 really fun. This should be a lot of fun
1:40:22 for you. I'm about to try to upload a
1:40:25 video file to X. And here's something
1:40:27 really funny. Um, Elon Musk has figured
1:40:31 out a way to capture a skyscraper sized
1:40:35 rocket penis with chopsticks that's
1:40:38 falling from space, but you can't upload
1:40:42 a video file to the Xplatform reliably.
1:40:45 Awesome.
1:40:49 It's [ __ ] cool. Like, we're going to
1:40:52 Mars. Could Before we like get put the
1:40:56 robots on another planet, could could we
1:40:59 just fix the [ __ ] video upload? Okay.
1:41:03 All right.
1:41:10 Oh my god. 11 Labs voice design new
1:41:13 mode, new model in your DMs. Good lord.
1:41:17 Really
1:41:19 unbelievable. All right, so before
1:41:21 before we go over to X, well, actually
1:41:23 actually I was going to go over to X and
1:41:25 turn on sound. So let's go do that and
1:41:27 then we'll upload as well. All right, X.
1:41:33 Um, sound on. Yes, let's the series.
1:41:37 Introducing Voice Design V3O
1:41:41 11s. My blade is ready. fine-tuned to
1:41:45 our most advanced model yet. Now you can
1:41:47 describe a voice and we'll create it for
1:41:49 you.
1:41:52 This could be the final ingredient. From
1:41:54 the surreal to the serious. This is
1:41:58 strictly confidential. From the poignant
1:42:00 to the preposterous. Greetings,
1:42:02 Earthlings. I must say your voice models
1:42:05 are out of this world. Perfect for
1:42:08 films, video games, and any other
1:42:11 project. Voice Design can help you bring
1:42:13 your stories to life. Back in my day, we
1:42:17 didn't need all these fancy gadgets. So,
1:42:20 try Voice Design V3 today, only with 11
1:42:24 Labs. Unbelievable. I did a I did a uh a
1:42:28 talk to a group of voice actors two
1:42:30 years ago. Um and I said, "Yeah, it's
1:42:36 I don't have a lot of good news for
1:42:37 you." Uh that to me looks like the you
1:42:41 know like like we are quickly arriving.
1:42:45 I mean it's unbelievable if you if you
1:42:47 look at just look at the animation that
1:42:48 we just did. So in about an hour not
1:42:51 even in about 30 minutes we made that
1:42:56 video of like the crocheted yarn style
1:42:59 like I don't even know where that came
1:43:00 from. What it doesn't matter. We did the
1:43:04 uniform cloud style. Oh I said
1:43:06 claimation style. That's what it was.
1:43:10 It did that. We told this story. We did
1:43:13 the music. We could do a voice over.
1:43:15 Like we could have that little kid
1:43:16 going, you know, saying whatever with a
1:43:20 voice like we did here.
1:43:24 I just it like like we're getting to the
1:43:26 point where every single um
1:43:31 every single profession is
1:43:35 now here's here's what I'll say about
1:43:37 this. It this is not every single
1:43:40 profession is ruined is gone.
1:43:43 What this is is every single profession
1:43:47 now has an entirely new way to produce
1:43:51 what they used to produce.
1:43:56 And
1:43:58 a lot of them are going to hate it. The
1:44:00 same reason that current programmers
1:44:03 hate vibe coding
1:44:06 because it democratizes something that
1:44:08 took them [ __ ] 30 years to master.
1:44:10 Right.
1:44:13 There's a there's a very cool um
1:44:15 interview with Rick Rubin where he's
1:44:17 talking about music. And music used to
1:44:19 be you'd put in your 10,000 hours on the
1:44:22 violin and you'd get, you know, a seat
1:44:24 at the orchestra and you would learn,
1:44:26 you know, music theory and mastery and
1:44:28 all that. And then punk rock came along
1:44:31 and it was three chords and two fingers,
1:44:34 you know, power chords, two fingers,
1:44:36 three chords, and you could be a rock
1:44:38 star.
1:44:40 And he said, "Vibe coding is the punk
1:44:42 rock, you know, of coding." And and what
1:44:45 we just did is is you know the punk rock
1:44:48 of movie making and the filmmakers are
1:44:51 gonna [ __ ] hate it. But Jesus, look
1:44:53 at what we just did.
1:44:56 Okay, so let's
1:44:59 let's watch together.
1:45:02 All right. So, we're gonna say
1:45:06 there's my there's my movie and we're
1:45:09 gonna say
1:45:11 my worst
1:45:13 critic.
1:45:19 A tail.
1:45:21 Oh, a tail. Like a fish tail. A ta tale
1:45:26 of
1:45:28 a boy
1:45:30 who
1:45:32 ran away. Swam away. Who swam away with
1:45:38 the fish circus.
1:45:43 Hey, Talltail.
1:45:47 Um,
1:45:49 uploading 14%.
1:45:52 Okay, let's see. So, we're going to go
1:45:54 at scobilizer.
1:45:57 We're going to go at midjourney.
1:46:02 We're going to go at sununo.
1:46:07 All right, I think that's good for now.
1:46:10 36%. This is actually This looks more
1:46:13 stable than I've I've ever seen it.
1:46:15 Normally what it does is it'll go up to
1:46:17 40, back down to zero, up to 38, back
1:46:19 Oh, see it just went from 41, back to
1:46:22 38. So, I have a feeling this is going
1:46:25 to error out
1:46:29 at Pixar. Yeah. Okay, we'll put Sure.
1:46:31 We'll put it Pixar here at Pixar.
1:46:36 Um,
1:46:39 I I'm I'm still I I remain blown away
1:46:43 that we can just do this now.
1:46:46 [Music]
1:47:16 The fact that it ends with that music is
1:47:20 so good. Your your wish came true. There
1:47:22 was an error. Thank God. Oh, look at
1:47:25 that. Isn't that sweet? Yeah. You can
1:47:28 you can catch a skyscraper with
1:47:29 chopsticks, but you can't upload a
1:47:31 video.
1:47:32 [Laughter]
1:47:38 Oh my god. Okay, wait. Let's see. Um,
1:47:43 let's try to post it. Wait. Oh, no. You
1:47:46 know what I want to do? I want to do um
1:47:49 What's remarkable
1:47:52 is
1:47:53 that this
1:47:59 Little story
1:48:01 is a single
1:48:06 midjourney
1:48:10 clip
1:48:13 extended
1:48:15 four times
1:48:18 with different
1:48:22 story
1:48:24 elements each time
1:48:28 and
1:48:30 two original songs
1:48:34 from Sunno.
1:48:38 Um,
1:48:40 and it was all done
1:48:44 in about 30 minutes.
1:48:49 What a time
1:48:52 to be alive. exclamation point. Okay.
1:48:56 So, what I want you to do, I'm going to
1:48:58 try to upload this now. Let's see if it
1:49:00 errors out again.
1:49:05 And then while that's going to that's
1:49:07 going to fail again. All right. So,
1:49:08 wait. Let me go let me just go save this
1:49:10 as a different file format because we
1:49:11 know we have to do that now because
1:49:14 that's the way this [ __ ] works.
1:49:17 Where did I export it? Oh, there we go.
1:49:23 Export as
1:49:26 720p is fine.
1:49:29 My worst critic.
1:49:33 All right.
1:49:38 All right. It's going. It's looking
1:49:40 good. 30% 31%.
1:49:44 It's It's photo realism looks are
1:49:47 through the roof. Image gen.
1:49:50 That's intentional. What's intentional?
1:49:53 The photo realalism thing.
1:49:57 It just came out.
1:50:00 Anybody use Higsfield Soul or or heard
1:50:02 of it? Wow.
1:50:04 I seriously we're we're out of time. You
1:50:07 You simply can't keep up with this. Like
1:50:11 stuff is dropping as we're having a show
1:50:14 to talk about all the stuff that dropped
1:50:16 right before the show.
1:50:20 This is unbelievable.
1:50:23 This is unbelievable.
1:50:25 Did it work? Okay. No error.
1:50:31 All right. So, let's get rid of this bad
1:50:33 boy and let's go add in the mauv version
1:50:36 of this and watch it fail.
1:50:40 All right. 1% 2% 3% 4% 5 6 7 This looks
1:50:45 promising. 8 9 10 11 12 13 14 57 7 19 21
1:50:52 22 33 We'll see if it gets to 38% and
1:50:54 then backs down.
1:50:57 30 31 32 33. Dear Elon Musk
1:51:02 30, it went 1235. All right, now it's at
1:51:06 40. We're still sort of progressing
1:51:08 upward. 50.
1:51:10 Uhoh.
1:51:12 hung 87.
1:51:15 I think I think we're gonna get there.
1:51:18 Dear Elon,
1:51:20 congratulations on your little rocket.
1:51:24 Sorry about the last one.
1:51:26 Could you [ __ ] fix video upload on
1:51:29 Twitter? Okay, so here's what I want.
1:51:30 There's 39 people. There's four people
1:51:33 on each channel.
1:51:37 Go to my Twitter account, Kyle Shannon.
1:51:39 I'm about to post this and do me a
1:51:41 favor. Like this video and and quote
1:51:44 quote tweet it, quote post it, whatever
1:51:47 the [ __ ] they call that, where you
1:51:49 forward it on and say, "Hey, look how
1:51:51 fun this is. This was fun to watch being
1:51:53 made." Would you do that for me? Thank
1:51:54 you. Okay.
1:51:59 Something just hit Alabama. All right.
1:52:01 Some storm or UFOs or something. The the
1:52:03 the the
1:52:05 ball. You see the ball? The thing down
1:52:08 in South America, they got the ball,
1:52:10 like soccer sized ball, but it flew fell
1:52:12 out of the sky and it's got [ __ ] on it.
1:52:15 Hoax, no hoax, who cares? Doesn't
1:52:17 matter. We live in cool times. We can
1:52:19 just make [ __ ] like this.
1:52:24 Here we go. Play it again for you. We
1:52:26 just make this
1:52:28 [Music]
1:52:49 No, no, no, no, no, no, no, no.
1:52:52 [Music]
1:52:58 [Laughter]
1:53:11 Oh my god. Oh, it's so good. All right.
1:53:15 All right.
1:53:16 We're [ __ ] filmmakers now. I'm I'm
1:53:19 submitting that thing for a short an
1:53:21 Oscar for the shorts next year. Um,
1:53:25 fantastic. Watch that percentage jump
1:53:27 around. Reminds me of Isles Apple's time
1:53:29 remaining when we did backups. Yeah,
1:53:31 exactly. Like like don't give us a
1:53:34 progress bar if it's meaningless. And
1:53:36 don't
1:53:38 you you have the world's most popular
1:53:40 communications platform
1:53:43 that highly features video and your
1:53:46 video uploading service is broken.
1:53:50 Does am I the only [ __ ] one that
1:53:53 notices this
1:53:55 now? It's probably I guarantee it's it
1:53:58 there's probably some corporate thing
1:53:59 where if you're on Xfinity internet,
1:54:02 which I am, they probably are at some
1:54:04 corporate war with Google and Google or
1:54:07 or uh Twitter and and Twitter
1:54:10 purposefully nerfs their upload
1:54:13 stability to piss all over the corporate
1:54:18 corporate people suck. Okay, said it.
1:54:22 New AI categories coming to the Oscars.
1:54:25 Really, Kak
1:54:27 not surprising. Look out, Tim Burton.
1:54:31 Um, okay. So, the other thing that we
1:54:33 did right before I leave, let me change
1:54:35 my tabs and share one other thing. Right
1:54:37 before we came on, we went on to
1:54:43 opendeepresearch.org.dev
1:54:47 dev openressearch.dev.
1:54:50 So if you go right now
1:54:53 to open deepepressearch.dev, it's an
1:54:56 open-source deepresarch tool
1:54:59 and let's see where did I Yeah, this is
1:55:03 it. Okay. So what I asked it to do was I
1:55:07 I asked it to find me archetypal
1:55:10 what are the what are the what are the
1:55:12 character archetypes for children's
1:55:15 stories globally?
1:55:18 Um,
1:55:19 and so let's see what it came up with
1:55:21 here.
1:55:23 Global fairy tale archetypes and their
1:55:25 crosscultural counterparts.
1:55:28 This research examines the universal
1:55:30 archetypes. Kef note. Oh, not a fact,
1:55:34 just a guess. Okay, cool. Got it. That's
1:55:37 cool. Well, oh, oh, like will there be
1:55:39 Well, you know what's going to happen,
1:55:41 Kak?
1:55:44 Oh, the thing we're looking at right now
1:55:46 just launched today. This is brand new.
1:55:49 The thing we're looking at right now,
1:55:50 what I think is probably going to happen
1:55:52 in the Oscars,
1:55:57 AI films will have kind of ghetto status
1:56:01 initially, right? Where they may
1:56:04 reluctantly create some AI categories
1:56:06 and those will be given as technical
1:56:09 technical Oscars.
1:56:11 Cost of midjourney, that's a good
1:56:12 question. I'll tell you in a second.
1:56:15 They'll they'll they'll give technical
1:56:17 Oscars just like they give, you know,
1:56:19 Oscars for like new film chemical, you
1:56:22 know, formulations. Um,
1:56:26 it'll start out in the in the sort of
1:56:29 not in prime time and then they may
1:56:31 reluctantly do a specialty Oscar like
1:56:35 best film that's AI generated and then
1:56:38 at some point it'll just have to be
1:56:41 absorbed because so many films will use
1:56:44 so much AI you won't be able to tell
1:56:46 what's live action, what's what's uh
1:56:50 what's AI. Um, and they'll just it'll
1:56:52 just become another part of the film
1:56:54 making process, I think. Okay. Cost cost
1:56:57 of midjourney subscriptions. So,
1:57:01 with the $10 a month subscription, you
1:57:04 get like I forget what it is. It's like
1:57:06 four. It's like an hour
1:57:10 an hour of It's not even an hour. Is it
1:57:13 a minute? I don't know. It It's maybe
1:57:15 it's four minute. I don't [ __ ] know.
1:57:17 You get some handful of of video
1:57:19 productions in fast mode.
1:57:22 You get no video productions in slow
1:57:24 mode. And so basically for 10 bucks a
1:57:27 month you get like I don't know 15 or 20
1:57:30 videos. For 30 bucks a month you get
1:57:33 like I don't know 120
1:57:37 something videos
1:57:39 and then no videos in slow mode. And
1:57:42 then for 60 bucks a month, you get like
1:57:44 four or 500 in fast mode, but then you
1:57:47 have unlimited in slow mode. So
1:57:50 basically, if you want to just use
1:57:54 AI video all the time or midjourney
1:57:57 video all the time, you really need the
1:57:58 60 buck a month thing, which if you
1:58:01 compare it to something like V3, which
1:58:04 is 125 bucks a month for three months
1:58:07 and then it's going up to 250. Um uh uh
1:58:12 some of the other tools that are are
1:58:15 really good like runway like the video
1:58:17 tools are pretty expensive. So 60 bucks
1:58:19 is kind of in line or cheaper than other
1:58:24 video tools. And the the fact that you
1:58:26 get four videos as quickly as you get
1:58:29 them, as cheaply as you get them, and
1:58:30 you get four of them with Midjourney is
1:58:33 another reason that it's remarkable. the
1:58:35 fact that it doesn't. There's two things
1:58:37 that Midjourney
1:58:39 sucks at right now. One is text. It
1:58:42 doesn't do text well. It just doesn't.
1:58:45 So, what you may have to do is um do
1:58:48 your images outside of Midjourney,
1:58:51 import them into MidJourney and then and
1:58:52 then animate them within MidJourney. You
1:58:54 can do that. It doesn't have to be a
1:58:56 MidJourney image. And then the other
1:58:58 thing is it's not doing sort of the
1:59:00 voice, the sound effects, all the audio
1:59:02 stuff and that like the acting that VO3
1:59:05 does. Um, which is a real bummer because
1:59:08 because now you kind of have to duct
1:59:11 tape a bunch of different tools
1:59:12 together, but I assume at some point
1:59:14 they'll get there because they're
1:59:16 MidJourney's vision is not to be an
1:59:19 image generation tool or a video
1:59:20 generation tool. where they're headed is
1:59:22 real time world generation models where
1:59:27 you'll be able to enter a world with
1:59:28 characters and 3D objects and sounds and
1:59:31 all the stuff and video is just sort of
1:59:35 the second stage of four stages. It was
1:59:37 images, video, 3D worlds and then real
1:59:40 time world creation.
1:59:43 So that's the journey that they're
1:59:44 that's the mid journey they're on. We're
1:59:46 we're about midway through that journey.
1:59:48 And I think the fact that they're
1:59:50 training it, they're probably much
1:59:53 farther down that path than what they're
1:59:54 show showing to us, which is probably
1:59:56 why the video is as good as it is. Okay.
2:00:00 So, where I'm at right now is
2:00:02 openressearch.dev.
2:00:04 And I before we came on, I think you get
2:00:06 one dev one prompt. I don't know if it's
2:00:09 a day, a month, I don't know what it is.
2:00:12 Um, and I asked it to do these
2:00:15 archetypes. So it said the research
2:00:16 examines universal archetypes that
2:00:19 present in fairy tales across different
2:00:21 cultures and their narrative
2:00:22 counterparts worldwide
2:00:24 such as the innocent hero, the villain,
2:00:26 the trickster, and the princess.
2:00:28 Introduction. Okay. Archetypal taxonomy,
2:00:31 culture,
2:00:32 China. So wait, here's the archetype
2:00:35 variations.
2:00:37 Magic helper, footwear, golden slipper,
2:00:39 oppressor, stepmother. Oh, cool. This is
2:00:42 really cool. The resurrected beauty, the
2:00:45 archetypal role systems conclusion.
2:00:48 All right. Oh, so it's short, but it
2:00:50 didn't it didn't really it only did
2:00:54 China, Egypt, Germany, and Korea.
2:00:57 So, I would I would basically say that's
2:01:01 thin and crappy. Give me more.
2:01:04 I can download it as a PDF. You get what
2:01:07 you pay for. Exactly. All right. Cool.
2:01:10 All right. So, that's that. That was a
2:01:12 lot tonight. We got through a lot. We
2:01:14 made a little movie. We're gonna win an
2:01:15 Oscar.
2:01:16 [Laughter]
2:01:20 All right. Do me a favor. I want you to
2:01:22 go to the URL Brandon's about to pop up
2:01:24 on screen.
2:01:27 [Music]
2:01:31 That one. Go to the salon.ai if you're
2:01:34 not already a member. Scroll down a
2:01:36 little. Click on join our community.
2:01:39 join our community. It's going to take
2:01:41 you to our community site where you can
2:01:42 log in, you can introduce yourself, read
2:01:45 about our values and what we're about
2:01:48 and you can start to connect with other
2:01:50 people that are curious about AI. If
2:01:52 you're one of those people that you're
2:01:53 trying to figure this stuff out, someone
2:01:55 was in here earlier going, "Where do I
2:01:56 get lessons on this? How do I start this
2:01:58 stuff?" You've got to be in community.
2:02:00 You can't not be in community. Wait, you
2:02:03 can't Yeah. not be You have to be in
2:02:05 community.
2:02:08 have to.
2:02:10 And ideally, you're in a community like
2:02:12 the AI salon that's full of generous,
2:02:14 smart, engaged, curious people that are
2:02:18 willing to help. Okay. Um, one of the
2:02:22 things you can also check out if you're
2:02:24 part of the AI salon and you want to up
2:02:26 your game, check out the AI mastermind.
2:02:28 In in our little welcome sequence,
2:02:30 there's seven steps. The seventh step is
2:02:32 join the mastermind. And that'll tell
2:02:34 you all about why we created it and what
2:02:36 it is. It's a subscription-based
2:02:39 more focused area of the salon where you
2:02:43 can get to know people and and really
2:02:45 dig in and that's where the the clubs
2:02:47 are and that's where people are, you
2:02:49 know, there sort of a higher level of
2:02:50 engagement there. And so that's worth
2:02:52 checking out if you haven't checked that
2:02:54 out yet. All right. Um on YouTube, do us
2:02:58 a favor. If you haven't followed it
2:03:00 already, follow
2:03:03 LearningLab-ai.
2:03:04 That's our channel on YouTube. We
2:03:06 someone else took AI learning lab. So
2:03:08 learning lab-ai
2:03:11 on YouTube. Follow that.
2:03:14 All right.
2:03:17 Especially if your name is Gavin. Gavin,
2:03:20 I think I think you're being trolled by
2:03:22 producer Brandon.
2:03:27 All right. Fantastic. Um I am getting
2:03:31 out of here. Peace.
2:03:34 Yeah. Uh, as Brandon's writing right
2:03:37 now, come back tomorrow for another 54
2:03:39 new announcements. I I don't know what's
2:03:41 h Oh, I I'll tell you probably what's
2:03:43 happening. Um, in the rumor mill today
2:03:47 is GPT5 coming out July 17th. So, if
2:03:51 that's anywhere in the neighborhood of
2:03:52 true, within a month, we're going to
2:03:54 have GPT5.
2:03:57 It It seems like we're pretty close to
2:03:59 having Grock 3.5, which is supposed to
2:04:01 be really good. Um,
2:04:04 I don't know. We've we've got a couple
2:04:05 other big models coming out, but
2:04:07 whenever something big is about to drop,
2:04:10 all the other companies get their [ __ ]
2:04:12 together and push stuff out. So, I think
2:04:14 the next month is going to be lots and
2:04:16 lots of Thank you, Winston, very much. I
2:04:18 appreciate that. There's going to be
2:04:20 lots and lots of uh new tech to play
2:04:22 with. It's going to be hard to keep up
2:04:23 with. Assume you're going to be
2:04:25 overwhelmed, okay? Um, and just give
2:04:28 yourself some grace for that, but but
2:04:29 you're likely going to be overwhelmed.
2:04:33 Oh, I got to change my I got to turn off
2:04:35 that stupid reverser thing because I
2:04:38 can't see the post-it note when that's
2:04:40 dark reader. There we go. Now I can.
2:04:42 Okay. Um Oh, that's why that was in dark
2:04:45 mode. I the the uh Oh, none of you are
2:04:48 seeing this. It doesn't matter. Okay.
2:04:50 All right. So, uh have yourself a good
2:04:52 night. I'll see you here tomorrow at
2:04:54 8:00 PM Mountain time. Same bat time,
2:04:57 same bat channel. Uh hope you have a
2:04:59 good night. Bye.