Creative Flux

Pierson Marks (00:01.485)
Sick. Cool.

Episode 13. Episode 13, Creative Flux.

Bilal Tahir (00:10.018)
Here we are, 13.

Pierson Marks (00:12.337)
We always, I feel like we have to do a better job at the beginning of having a synchronized intro. It's like, hey, I'm Pearson. And then you come in and you say you're Belol. Let's just start this and I'll cut it off at the beginning. like, welcome everybody to episode 13 of Creative Flux. I'm Pearson.

Bilal Tahir (00:16.846)
you

I am.

Bilal Tahir (00:33.486)
And I'm Belal. There you go. I totally didn't practice this. don't.

Pierson Marks (00:35.193)
And we totally didn't practice that. But yeah, we talk about generative media. And today we have a bunch of cool stuff to talk about. Last week we were talking about Huyon, think. How do you pronounce that? Voyager model? Huyon? Huyon? The Voyager model. And then we talked about a bunch of stuff. It's always really fun to chat.

But this week we got a C-dance, we have Descartes fast video gen, we have a lot of things we didn't talk about last week. But yeah, mean, space is moving fast, it's fun.

Bilal Tahir (01:24.566)
It is, is. And I guess I think Seed Dance, or is it Seedream? I'm almost like, Seedance. Yeah, I know. There's very similar sounding models. But I feel like the image editing space in particular has been like really.

Pierson Marks (01:30.491)
Sea dance, sea dance. I say sea dream, sea dance.

Bilal Tahir (01:41.134)
getting its moment, you know, we started, we talked about Nano Banana the last couple of weeks and you know, how good it was at editing images and making very granular edits. And it sounds like ByteDance with C, you know, C times CD.

Pierson Marks (01:58.721)
Wait, so it's ByteDance's model, right? Cdream's ByteDance model? Okay, so I wasn't too off on that, so.

Bilal Tahir (02:01.782)
Right. I think so. Yeah.

Yeah, so they released C-Dream. Well, I'm just going to look it up now. It's called C-Dream. OK, By Dance's C-Dream model has been released. it's supposedly better than Ano Banana. I think I've seen some of the examples I saw. They were really good. I think a couple of places where it really stood out was combining

Pierson Marks (02:14.481)
See you then.

Bilal Tahir (02:36.505)
different types of formats. So you could take a cartoon character and then take a photorealistic image and put the cartoon character in the image. And I don't know, I feel like Nano Banana struggled a little bit with that.

Pierson Marks (02:44.688)
Mm.

Bilal Tahir (02:49.78)
The other thing where it's matched Nano to Nano, which I think people haven't done enough experimenting, is multiple image output. So you can actually ask CDream to output like four or five images. And that's pretty cool because you can basically get a consistent storyline with the same theme, with the same character, you know, in one shot, you know, which I think is very powerful. And then you can obviously combine that with a video generator to get like five second scenes, et cetera. So powerful.

Pierson Marks (03:20.155)
Yeah, no, super interesting. You just mentioned something that we talked about last week about like the, what was that movie with LeBron James, Space Jam. Yeah. Michael Jordan, I'm sorry. I'm gonna, the new one. Yes, yes.

Bilal Tahir (03:30.118)
Space not LeBron James, MJ. Yeah, although LeBron did do a copycat of it because he's trying to, you know, he thinks he's MJ. I mean, I will say, I mean, they're both up there. It's the LeBron versus MJ debate will go on forever.

Pierson Marks (03:45.808)
Right.

It'll go on forever. But is there a name for that style? Like, has somebody coined that style where it's animation in a real?

Bilal Tahir (03:57.038)
That's a good question. Yeah, that's a very good question. Animation realism in my eye, I just think of like Roger Rabbit, like the Roger, because that's like, that was the first movie that did that. And it's crazy. It came out in the 80s, in the late 80s and still looks so good. Like better than most things you'd see today, which is crazy. But really good.

Pierson Marks (04:17.647)
Right, right. I feel like we have to make a term for that because I agree with you. It's a really powerful style that hasn't been explored much and it really merges two styles. like what we were talking about last week with Wreck-It Ralph and how you have a character that goes between worlds and they maintain their own style. Like a...

Bilal Tahir (04:29.24)
Mm-hmm.

Bilal Tahir (04:35.554)
Mm-hmm.

Pierson Marks (04:43.385)
know, pixel art goes into a 3D high res world, but also in movies. I mean, that's, yeah, that'd be interesting. I feel like we have to think of a name for that eventually. We can point at ourselves.

Bilal Tahir (04:46.776)
Hmm.

Bilal Tahir (04:56.406)
Yeah, but it is underused. You don't really see that. You don't really see a different animation, know, like worlds colliding. And I wonder if that's like a new type of art form that is yet to be explored. Because there's, I mean, we talked about, obviously, anime and photorealism, but what if it's just like Pixar in Ghibli or something like that, you know, just like, or maybe that's just too ugly. But it would be interesting.

Pierson Marks (05:09.648)
Right.

Pierson Marks (05:19.535)
Mm. Right.

It used to be, no, it's so cool. I remember there used to be like some Nickelodeon shows. I think it was Nickelodeon or Cartoon Network.

I forget now on the top of my head, but there are some like these mashup episodes where, I think it was Cartoon Network, right? Timmy Turner from The Fairly Oddparents would go into like Danny Phantom and like these other worlds. And it would be like these weird mashup type of episodes where a character from another show on the same network would come in. You're like, wait, what are you doing here? Like, it makes no sense. was multiverse before multiverse became uncool with Marvel. So.

Bilal Tahir (05:58.221)
Right.

Pierson Marks (06:00.271)
Yeah, it's interesting. There was something. So last night I was at AI Tinkerers and I demoed that Satori MCP server. It was pretty cool. think everyone really liked it. It was...

Bilal Tahir (06:01.624)
That's awesome.

Bilal Tahir (06:09.87)
Nice. Yeah, Explain, I guess, a recap for the audience what Satori is.

Pierson Marks (06:17.839)
Right. So I built this thing leveraging Versailles Satori library, which allows you to programmatically generate images using React components. So it's really a simple, it's a cool library that just.

Bilal Tahir (06:30.488)
Nice.

Pierson Marks (06:35.215)
provide a React component, and then it's a rendering engine that converts that React component into an SVG. And then I took that library and wrapped it around with other stuff like converting the SVG to the PNG, allowing you to provide dimensions, making a bunch of different templates that already look good for different use cases, such as like a social card, a blog cover art, a YouTube thumbnail. did that one this morning because I just posted a YouTube video.

Bilal Tahir (07:02.542)
guys.

Pierson Marks (07:05.159)
of this cool thumbnail to be able to be reused over time. And I think that's like a key because why I built this originally was any time I wanted to create a YouTube thumbnail or a blog cover art, I went into Photoshop and I had a template kind of in Photoshop, but it's way too difficult.

to make good changes in Photoshop because text wrapping is hard. If you have a long title, it's not gonna automatically text wrapped around the canvas. The spacing is gonna be weird, but if you could just write a React component with Flexbox and say, hey, give it 16 pixels of padding, wrap it around, align this vertically, like you can just define a nice little component that you can swap out the text or the images from.

Bilal Tahir (07:34.828)
Okay.

Bilal Tahir (07:46.166)
Right.

Bilal Tahir (07:55.564)
Right, yeah.

Pierson Marks (07:57.785)
And so was like, well, that's cool. I did that. then now can I create an MCP server? chat GPT, now chat GPT as of yesterday, because they have MCP support.

Bilal Tahir (08:06.222)
yeah, the ChatGP developers came out.

Pierson Marks (08:11.491)
Right. So we could talk about that too, but yeah. So now you could just like chat and, and say, Hey, create this blog cover art template with the title.

Creative Flux episode 13 and then the description this and it'll go and then you can iterate on that and the coolest thing is that you can actually have consistency every time so like let's say you mess up the title the next time you say change the title and it'll just change the title and keep everything else the same because it's literally using a template and rendering the template not going to diffusion model and so I thought that was really cool because you leverage the power of LLMs to be able to to write code

Bilal Tahir (08:42.371)
Right.

Bilal Tahir (08:50.594)
Yeah.

No, I think it's super powerful. And even when, as these editing model like, know, see Dream Nano Banana get really good at editing, I still think there's, you know, a great like, you know, space for this kind of programmatic, you know, approach because you can have something that's super cheap, super fast, because it's just code. And it's obviously, you know, very, you know, it'll give you the exact changes you want. It looks exactly the same. And then I wonder if there's like a hybrid approach where you can do the programmatic approach where you can swap

or the text, then maybe you generate a new avatar and then add that. So you kind of combine the AI with the procedural, like programmatic driven template.

Pierson Marks (09:34.585)
It'd be really cool. Cause I was thinking about this yesterday, Dex brought this up and he was like, could you add in like a generative model for the background? Cause like some of the templates you could provide a background image.

Bilal Tahir (09:47.342)
Mm-hmm.

Pierson Marks (09:47.441)
And if you don't provide that, like, could you provide a prompt for the back on image and then just generate something versus providing the image, which could make total sense. And then the other thing that I went down this rabbit hole is if you have the programmatic image being generated.

could you apply a layer at the end? So you do the iteration process to get the right text there, the right font, everything. And at the very end, you just apply a filter, a layer on top using a great model like Flux Context or something, Nano Banana, just at the very end, the very last step is applying the diffusion model.

when the text is already correct and just applies a layer. So maybe you have this font, but you want that font to kind of look psychedelic. it's just like, convert this to a psychedelic image. And then you've already iterated and done all the testing on all of the first steps. But now just apply the filter at the end to make it psychedelic. And it's like, it can work.

Bilal Tahir (10:35.758)
Hmm.

Bilal Tahir (10:40.418)
Right, right.

Bilal Tahir (10:55.478)
Right, yeah. It does, yeah. No, it's super powerful. And you know, it's funny, like we talk about these concepts come from Photoshop layers and stuff. I'm wondering like if we can apply them to video. Like, you you could just have a background layer, you have a character layer, know, glasses layer, whatever. And then you can just change that real time. You know, like from frame this to this, just change the glasses to something like without glasses for some reason, you know. That could be.

Pierson Marks (11:11.853)
do I?

Bilal Tahir (11:24.034)
I mean, you can do it, I guess, regenerating a video, but I'm trying to think of a more efficient approach where you can edit the frames and then the exact pixels, you know, that.

Pierson Marks (11:33.104)
Right.

I think there's a lot of concepts that have been very well thought out over the last few decades when it comes to graphics, whether it's image, like static images in Photoshop or like Premiere Pro and layers there, because Premiere Pro is like Photoshop on steroids kind of, it's just Photoshop times a million because you just, you could deal with frames over the course of.

Bilal Tahir (11:54.83)
Mm-hmm.

Pierson Marks (12:00.995)
a video, but it's very well thought out and a powerful tool if you know how to leverage it.

Bilal Tahir (12:06.456)
Yes, yes. I mean, these are tools like these old school tools actually are.

like still around and they're so powerful. And speaking of which, now I understand like why I thought about the video because what I saw was EBSynth. I don't know if you've ever played with EBSynth, but it's like an OG tool as well. I think Mac only app, which lets you do stuff like the Premiere Pro, but like for videos. And they actually leverage AI to do this where you had like a lion walking and they put like markings on the lion and the video just continued with lion having the marking. So they're approaching this video.

exactly the science behind it but it's super cool that you can just edit videos like this and I imagine Premiere Pro has that too so

Pierson Marks (12:48.209)
All right.

I remember just always being so frustrated as a kid trying to make some like basic videos. I remember there's videos that I would make where you'd record something normally and then you just reverse it. So it looks like if you're jumping onto the bed, it looks like you're jumping off the bed like a Jedi or like the ball, like you throw a ball and it's more like a force. I always thought that was so cool. And then you would try to like remove and...

Bilal Tahir (13:09.964)
Right, right.

Pierson Marks (13:20.305)
They had very basic non-AI effects, kind of, that would help you isolate things and help you remove them. I like, why is this not easier? I just always remember thinking, I see the ball right there. Or delete it. Just delete it and track it. And I was just like, why can't you do that? It's there. I could see it. Why can't you do that, computer? Why are you so dumb?

Bilal Tahir (13:37.782)
I know, it's there. Right, right.

Bilal Tahir (13:45.496)
Yeah, well, hopefully we'll have those possibilities that are increasing now and becoming easier. These things are becoming way easier to do, so that's awesome. And it's like, speaking of the effects you were saying, like...

Pierson Marks (13:55.483)
Right.

Bilal Tahir (13:59.502)
So many cool, interesting effects like I'm seeing on Twitter, et cetera, are coming directly from these new capabilities. So one of the capabilities we talked about was first and last frame last time where you give it, this is the first frame, this is the last frame, and then you can create a five second radio. And we talked about how that can lead to more granular director level control because you can control the scene, how it goes, and then takes the end frame and use that as the start frame of the next one and so on.

Pierson Marks (14:12.166)
Right.

Pierson Marks (14:24.945)
All right.

Bilal Tahir (14:29.616)
a really cool interesting flow I saw which I didn't even think about but I was like this is awesome was somebody figured out what if I take the image I want make that the last frame and make the first image like a white screen or a black screen and so they were able to create like those videos where a logo like they took the apple logo or whatever and and they give a prompt about how the apple is going to come and a black screen and what the way it came out was the somebody drew the apple on a blank canvas like that kind of effect or

Pierson Marks (14:52.017)
Bilal Tahir (14:59.376)
and then wrote like something. And I was like, oh, that is so cool. So you took the feature that wasn't really meant for it, but you create this really interesting use out of it. So yeah, I mean, we'll probably see those kind of examples all over with these new models that we haven't even thought about.

Pierson Marks (15:09.733)
Right.

Pierson Marks (15:19.473)
No, it's super interesting. I wonder if Apple's iPhone 17 event, if they use any sort of AI to help do those designs. They've been doing those forever, but that's a whole other topic, I mean, too, but.

Bilal Tahir (15:35.547)
no, so this was just, I used Apple as an example, but somebody did that on a, I don't think Apple.

unfortunately did anything. I think they've kind of thrown in the towel with AI. They're partnering with, I think last I heard Google now for Siri. And I didn't really watch the keynote. I know a lot of people, but I mean, remain an Apple fan boy, but yeah, the 17 was kind of a, didn't, get people didn't really get excited from what I saw. Like it was just like, oh, we made a thinner iPhone. I'm like, all right, cool. I guess.

Pierson Marks (16:04.625)
One of the interesting takes that I heard from there was that, I mean, always have a, there's a limit, there's a ceiling to every sort of innovation. Like you can't do that much more after a time that the jumps just get smaller and smaller. But somebody I saw on Twitter was talking about how the iPhone Air,

could potentially be a way for Apple to prep the supply chain for developing really thin sort of chips and batteries for glasses. So it's like they need more time to be able to perfect like getting it down super small and lightweight so they can actually make glasses or the Vision Pro 2, which should come out in some form in the next year or so.

Bilal Tahir (16:54.2)
Right.

Pierson Marks (16:55.651)
lighter so you actually have the battery like in the headset versus on the side and like all those things so maybe it was like a supply chain type of release.

Bilal Tahir (16:59.842)
Yeah, yeah.

I mean, they do know hardware. I mean, they're still one of the great companies that does hardware. So I mean, I am still hoping they can figure those things out.

Pierson Marks (17:09.071)
Right. And the live translation is sick.

Bilal Tahir (17:15.564)
yeah, was gonna say, I was gonna come to that. That was like probably the one piece that I was like, wow. So to kind of back up what we're talking about is the AirPods release, the AirPods 3 Pro release, where one feature they showed off, which I don't know how much of that is vapor-varyered, because they've done that before, but the way the demo worked, could like, as you're wearing the AirPods, you're talking to someone and let's say they're speaking Spanish or whatever, you're speaking in English, the AirPods will silence the person's voice.

you know, make the volume go down on their voice and then give you a live translation, which is like a really nice UX that I didn't even think about, but that's a very nice subtle way of doing live translation. Cause I always thought like, how would you, if it was like somebody speaking and you're getting like a translation, like that's like, how are you going to process that? But that makes sense. You know, you kind of like, you know, make their voice a little lower and have that overlay almost. So very cool.

Pierson Marks (18:12.399)
Right. It's super cool. hope it works. I think this is one of those things that everybody talked about, the universal translator.

Bilal Tahir (18:19.724)
Yeah. Well, yeah, but it's funny because I mean, this is like the practical solution. The more sci-fi solution where we saw I forgot the company, it was that company that they had a demo. I don't know, but it was basically where you can send your thoughts to the other person. So, I mean, from first principles, that's what are our mouths.

Pierson Marks (18:31.303)
yeah.

Bilal Tahir (18:39.974)
Why are we moving them up and down trying to communicate? That seems so caveman-like, right? mean, just beam my thoughts into your head, right? I mean, that's the most efficient way to communicate.

Pierson Marks (18:50.993)
I remember I went down this whole rabbit hole one time where it's just like, it's definitely, there's a book about this, but I didn't read, I've never read the book, but I had a similar idea where, you know, humanity is a bunch of individual thinkers, but aggregated together, we're pretty predictable. Like anything random-ish, any randomness in life is like pretty predictable.

on a grand scheme and like the distribution of balls that fall down to like a plink aboard or whatever. I'll follow that bell curve. individuals, we all are unique individually. But when you put them in, you put everybody into society, all everybody kind of falls into like a place you have people everywhere on every spectrum. I was like, well, AI is such an interesting thing because it's a

compression of all human knowledge, essentially, you you've taken everything that we've ever written and put down onto paper and then now videos and images and the world, and we've compressed it into a model that's essentially a lossy sort of replication of all of humanity, you know, it's like this one thing, like chat GPT, GPT-5 and Gemini are literally a compression of everything humanity has ever known. And sometimes more lossy than others with hallucinations and everything, but

Bilal Tahir (19:59.726)
Mm-hmm.

Pierson Marks (20:11.821)
If that is like literally like a singularity moment where you could ask this one thing, anything about like anything and coming back to people like it, it's kind of literally doing the thing where it's taken all of our knowledge and just putting it condensing it. So if society, could just transmit thoughts and everything to each other and just be really,

Interesting, mean.

Bilal Tahir (20:42.87)
Yeah, for sure. It's, what's that biblical? I forget the tower where, you know, you know, the story where humanity, I mean, the mythology is like, we all used to communicate. We knew each other's thoughts and then, you know, the, forget the name, is it the Babylon Tower or whatever, but it broke down and then everyone suddenly became confused and we needed to develop languages to, you know, it's gonna bother me.

Pierson Marks (20:44.593)
You

Pierson Marks (21:07.836)
interesting.

Bilal Tahir (21:12.386)
or a capital.

Pierson Marks (21:12.823)
yeah, no, it's so wild. Like you can go down so many rabbit holes.

Bilal Tahir (21:18.252)
Yeah, Tower of Babel, sorry. Yeah, the Tower of Babel. that was the, what do call it? The people just scattered the earth. The story of Tower of Babel, book of Genesis, humanity once spoke a single unified language, but God confused it into multiple languages to stop the construction of a tower and cause people to scatter across the earth. Well, we're gonna build that tower. It'll be more in a hard drive, but we're building it back.

and hopefully we'll be able to unify each other. Yeah, I think it would be very, if you imagine a sci-fi world where we can just communicate with our thoughts, think of all the things that, know, the gaps we have because we don't speak the same language or the subtleties of a culture. If we can like express that to each other, I feel like that could go along when, you know.

Like not sang woo woo but like world peace and understanding and all that hippie stuff, you know?

Pierson Marks (22:15.309)
And it's super, super interesting. like what blows my mind when I think about this, it's how small we are in like the timeline of humanity and knowledge. And when it comes to intelligent species, mean, if like we, the wildest thing about AI is like,

You can debate this and how true you may or how much you may believe in this, but like we took sand, rocks, and now rocks, we talk to rocks. Like it's crazy. Like we created, like we derived chips from rocks and now we like work with rocks that we can talk to. It's like we summoned, we created like a species from nothing, not species that like, you know, it's

Bilal Tahir (23:08.92)
brain.

Pierson Marks (23:09.265)
But some sort of intelligence, you know, that we just created from rocks. And then you think about and you extrapolate this. It's like going from, OK, we created this intelligence. It's a lossy compression of all human knowledge. I believe that we aren't alone in the universe. I think it's like an unlikely thing to be.

Bilal Tahir (23:14.904)
Turn rock. This is the rock age. I like that. The rock age.

Pierson Marks (23:39.117)
I don't think if we are, I think if we do have other people in the universe, other species, intelligent life forms, they're likely to be potentially farther along than we are. It'd be pretty interesting to think that we are the most intelligent species in the entire universe if there's others out there. And so I think it's a safe assumption to think that they've been able to create something similar to what we've done with language models, a compression of all their knowledge.

And when I think about aliens, we're getting so deep on this, but when I think about aliens, mean, I imagine that aliens don't come to earth, if they were to come to earth, they wouldn't come to earth in like a ship.

travel and like a physical being would be there. It would be more so there is like a signal of knowledge that comes that's like radio waves or something in the universe that comes and hits earth. Like imagine if we took chat gbt and gpt5 and we just started transmitting all the all that information.

Bilal Tahir (24:32.046)
Hmm.

Bilal Tahir (24:41.71)
Hmm.

Pierson Marks (24:42.204)
across the universe and say, just like beaming it out. It's like, Hey, we are humans. This is chat. GPT. You can ask, you can learn anything about humanity, everything that we know and just beaming out into the universe. Something out there could receive like, Hey, what is this signal that we're getting? And they're able to like record it and then they're able to interpret it. they're like, Whoa, this is like this compressed version of this species that lives out in the universe. And we can just learn about it. And you could think about that vice versa. mean,

Bilal Tahir (25:10.018)
Yeah, it's interesting. I mean, yeah, can go down this rabbit hole too, but it's so fascinating because I remember I watched Carl Sagan's interview. He did this in the 80s. And he talks about how the first, actually they did this where they sent out information and through TV stations and stuff like radio signals. he's like, but it'll take like years to reach. So he's talking in the 80s. He's like, if.

Pierson Marks (25:14.257)
You

Bilal Tahir (25:37.806)
Theoretically, aliens would pick up on our signals now. They will basically find television programming from the 50s. So that will be their.

Pierson Marks (25:46.171)
Right.

Bilal Tahir (25:51.038)
impression of what it's like on Earth right now. And so you can imagine how now that same signal is now 60 years, 70 years out. So it's fascinating to see the time differential. the other point about like aliens about being technologically more superior, I actually think that's most likely the case. mean, it's basically, I don't know if you've ever read the three body problem or heard of it, but it's very dark. It's like a dark take on it, but it kind of makes sense.

Pierson Marks (26:15.247)
I never have, not yet. Yeah.

Bilal Tahir (26:20.942)
assumption is that every I don't want to give too much away if anyone wants to read it you know although I think it goes it gets really weird it kind of flubs the ending in my opinion but great setup great setup as a book but the idea is basically that most of civilizations if they are alien civilizations they are

the likelihood that they are at our level or below is actually very small because if you look at it from a time span, if you're here on an infinite horizon, the chances that civilization comes after, lays more advanced, is like way higher than it was less advanced.

because it's either life never happened or it happened and they destroyed themselves. And so, if you take that out, it's like if they actually exist, they're more likely to be more advanced.

Pierson Marks (27:12.249)
Interesting. Right.

Bilal Tahir (27:12.974)
And it actually leads to a theory that the author came up with, it's called the Dark Forest Theory, which is basically that any civilization that's advanced enough, the game theory is you should never reveal yourself because if you reveal yourself, the only civilizations that can pick up your signal are more advanced than you. And from their point of view, the only reasonable maneuver is to kill you before you become more advanced.

basically, because they're like, you know, you can only be a threat. And so the idea is that's what his his response to the Fermi paradox, which is like, why aren't there more alien civilizations that they are, they just are dark, because intentionally, because they don't want to reveal themselves. I don't think I think I don't buy that. I feel like, you know, hopefully, the world, the universe is a little kinder than that. And I just think I my my take is that we just happened to have we just underestimate how vast space is. And if you actually and we can talk

Pierson Marks (27:39.269)
Right.

Pierson Marks (28:01.749)
Yeah.

Bilal Tahir (28:09.244)
be the only civilization in the Milky Way even. mean and even that would take us like if you look at just how big the Milky Way is, it's ridiculous. We haven't even explored that and that's like one point in the whole universe so I mean you know it's kind of crazy so and then finally the other point I'll make coming back to our language thing and it's very interesting is like in the in the book

the actually the way the aliens talk are through thoughts. So it actually, they have that. It's just part of their biology. And one of the reasons they want to kill humanity in the book is because they don't, what they get freaked out about is a concept of lying because they have always communicated with their thoughts. So in the book, they get freaked out when somebody tells them about how humans are like, he's talking about Robin Hood and he's like, can I meet Robin Hood? And he's like, oh no, it's just a story. It's a myth. They're like, what are you as a myth? And then the alien learns about what's, can say things that are not true.

Pierson Marks (28:39.665)
All

Bilal Tahir (29:06.128)
and that frees them out because they're like, they have no concept of this in their soul. they're like, because they're more intelligent, they immediately realize that that is like an existential threat because they will always say the truth and we can lie to them. And so that means that they can never be a negotiation. It's a very deep book. mean, he really goes down this rabbit hole, but that's one of the reasons they want to kill us in the book, among other things, among the resource.

Pierson Marks (29:24.888)
Amen.

Pierson Marks (29:33.137)
Is it a Chinese book and it was translated?

Bilal Tahir (29:35.97)
Yeah, it's a single Lee, but yeah, he wrote it as a Chinese book, then it got translated. And now there's a Netflix series about it too, which I think it's actually did a decent job, a better job than the book, because the book is like one of those hard sci-fi books where he'll like go down a 50 page rabbit hole of why quantum mechanics works and how do this so phone gets from A to B and all that. like, I don't want to know.

Pierson Marks (30:00.337)
So watch the Netflix show.

Bilal Tahir (30:02.318)
I recommend the Netflix series, I think it's pretty good. I mean, it does help if you have read the book because you see the concepts happening and you know, I feel like you

you know, the 50 page rabbit hole definitely gives it more depth. You're like, I understand how they fucked up the physics here, you know, how that happened. But sorry, coming back, yeah, that was a deep sci-fi rabbit hole, but in case aliens are hearing us right now, you know, we come in peace and yeah, please share the tech. You know, we are very techno accelerationist on the creative flux and you know, we want to do good. We want to build cool stuff, you know.

Pierson Marks (30:17.957)
Mm.

That's interesting.

Pierson Marks (30:24.35)
Yeah

Peace.

Pierson Marks (30:36.945)
We have this thing called a nano banana, which we'll have to show you. We can make Shrimp Jesus real. There's this thing I saw. Did you see the nano banana in FiniMap?

Bilal Tahir (30:43.884)
Yeah, there you go.

Bilal Tahir (30:49.653)
yeah, no, what is that?

Pierson Marks (30:51.547)
So pretty much like for anybody that doesn't know about NanoBanana, we've talked about it now a few episodes, but it's just like a pretty good, consistent model. But there's this guy that made this thing called the NanoBanana Infinimap, and it's a map generator. So like if you ever played a video game or anything and you always have like your mini map, map building and world building is really fun, I think for a lot of people. And what he did was that he took a infinite canvas and

pretty much define some squares along that canvas. And people could come into this map and use NanoBanana to generate a piece of that map. And so just like you had a year ago, a few years ago, we had those massive infinite canvases where people can just put a color on different pixels and you can see people and everybody's editing this massive thing. He did the same thing with NanoBanana where it's more of this world-building experience.

Bilal Tahir (31:30.288)
nice.

Bilal Tahir (31:38.264)
Right.

Bilal Tahir (31:42.403)
Ready?

Bilal Tahir (31:49.304)
So is it like live and like people can come and see other people's like edits as well and then scroll and generate themselves or?

Pierson Marks (31:56.217)
I don't know if it's hosted in live. It's a public open source repo. It would probably be a pretty fun little thing to test out live real time connections where you just.

Bilal Tahir (32:06.989)
night.

Bilal Tahir (32:13.536)
I think that would be super sick. of my, I actually think this would still be a sick idea once and if ever NFTs come back, this is like, I know it sounds funny, but I had this idea where I...

Pierson Marks (32:20.785)
All right.

Bilal Tahir (32:25.71)
was thinking during the Web 3 era, what if people were selling these JPEGs for millions of dollars? was like, but these are stupid. But I was thinking, what if you could create a crowdsource environment where you start with an empty canvas and then you have like a 60 minute clock. Anyone could come in and generate using a prompt, like an image, and then edit that image. And you basically do let the crowd go crazy on it and hammer it live for 60 minutes. And at the end, it's frozen and locked and you create a video

of all the edits that were made and you make a GIF out of it. And that's the NFT. And then it's get minted for that hour. And then you continue for every hour. it was just like this weird crowdsourced live art project thing, which I think would be cool and probably go viral, you know, but waiting. So if NFTs, you know, start, you know, getting more popular, maybe I'll look into it with NanoBanana or Seedream.

Pierson Marks (32:59.92)
Well.

Pierson Marks (33:05.701)
That's cool.

Pierson Marks (33:22.627)
So in that, you, would the people in the last few minutes, they would have the most power because it's like the last 59th minute.

Bilal Tahir (33:30.382)
So I had two versions of this. The first idea was this, which would be the video where I think any edit wouldn't be more powerful than any other. Maybe there's because of the thumbnail or whatever you can say the first image of the last image could be more powerful because that might define the thing, but it's a video. But the second version was that the last edit is the actual image. And you kinda, it's almost like you, and I actually wanted to make it like a lottery thing where you buy

you have to buy it but you don't know what you're gonna get and then people start editing it so you could end up with a dick pic or whatever yeah

Pierson Marks (34:04.785)
It's very probably likely, you know.

Bilal Tahir (34:09.858)
But imagine But imagine

What do call that Vegas machine? a slot machine. It's like a slot machine almost. So that's what I was saying. You could even have the animation of a slot machine. It's like coming in. but I mean, but I do think it's something we'll see. Cause last time when the NFT mania happened in 2021, it was not...

Pierson Marks (34:35.141)
What?

Pierson Marks (34:42.561)
No, that's funny. That's sick.

Bilal Tahir (34:54.318)
like we didn't really have image generation. It was just starting. We kind of had stable diffusion, but it was pretty shitty and people were generating stuff. But now with edits, you can imagine. So if you know about NFTs, there's a, usually there's a series. So you have the consistent branding. something like a Azuki's or a board apes. it's the same, it's very similar art form. Like you have an ape, you have a background and then you have different assets. So an ape could have a hat. They could have glasses. They could have a shirt. So these are different items, but they're very,

in the algorithm, in the contract about. And so with edits, it used to be that an artist would create like, let's say 10,000 of these JPEGs and everything was hand curated. And I actually think for the most premium NFTs, you probably will still have hand curation because it's like having a Hermes bag or whatever, in the equivalent of ways. mean, it's still JPEGs, but whatever, right? We're talking in our land of made up scarcity in Web3 world.

Pierson Marks (35:53.841)
All right.

Bilal Tahir (35:54.254)
But with image editing, anyone can come in, start with the base image of a character, and then you're like, all right, give me 20 different hairstyles now. Okay, that's 20 there. Now give me 20 different eye sets. Now 20 times 20, that's 400 versions created right there, right? That's combos. And so you do that and you do the math and you can create, let's say 10,000, 20,000 versions of a base character. And then voila, you have an NFT collection right there.

Pierson Marks (36:20.847)
All right.

Bilal Tahir (36:21.662)
So I think we'll see this a lot when the next crypto, three NFT boom happens, which it will. think it's a when, not an if, whether you like it or not.

Pierson Marks (36:32.273)
You think so? You think why do you take that opinion like it's a win versus an F? Like why do think that the NFTs is going to come back on in terms of like playing cards? I'm just curious.

Bilal Tahir (36:43.478)
I think it's it's like saying, will gambling be a thing again? I mean, it's just human nature. don't.

I don't opine on the utility of it. You can come up with any of your reasons, but I think if you just look at human nature and we almost have in this post economic world where money kind of doesn't matter and people like at least invest that way. The NFTs will, I think will just be a conduit of degenerate gambling. If anything, if nothing else, just like the idea that I can buy a JPEG and I can resell it for 20 times its value is just going to be.

enough and you have other forms of it like prediction markets, just meme stocks, memes. So it's, I think of NFTs almost like a memetic instrument. And so I think it will exist now in the form factor might change. I maybe we get away from a JPEG. Maybe you need to be more interesting, but the just having digital artifacts that you can sell and resell and the value of it comes not because

it gives you any utility but just from the community that I think will remain.

Pierson Marks (37:56.187)
Do you think the value is derived because it's an NFT or do you think it's like NFT is just that mechanism used to, to identify this is some object and it's anybody can verify that or like, cause I always was just concerned. I like the thing was I didn't really get was that something being the NFT. I think that didn't add value to the thing. And it was like.

If you cared about the art, why is now the art more valuable because it's an NFT? Like I know NFT, making it as an NFT gives you like, anybody can validate like the ownership of that thing. like, it's just more of though the certificate of authenticity of that thing versus that thing being valuable. So I just was always just confused on like why people.

Bilal Tahir (38:49.838)
Yeah, I think that's what it is. I think you nailed it. It's basically that it's a certificate. It's just saying, Oh, I own this NFT. Now you can say, you can say, I write it and copy it and

I have the same picture, but I can say I have a hash that points to the fact that I own this NFT. And because of that, maybe I get access to like a discord or whatever, and they can verify, or just it's purely status symbol. Like, wow, I can just say, it's purely like most art, half the value, if not more, is just derived from the fact that I can say I own the art. That's one of the reasons people buy art. They can say, I own a Rothko,

Pierson Marks (39:36.87)
Totally.

Pierson Marks (39:44.527)
Right. Wrapping this all together too, this just gave me this really cool idea. So what you were just talking about, like, NFTs and minting the images and then the Nano Banana Infinimap, I'm thinking right now. I don't know. Maybe I'll just share the screen real quick just so you can see what I'm seeing. So this is like the GitHub repo. And it's like...

Bilal Tahir (40:12.236)
Ooh, nice. how?

Pierson Marks (40:14.447)
You could just see this world and you just kind of this imagine that this little map right here was split up into like a thousand by thousand grid and then every grid you could just kind of put a prompt in and generate that thing.

Bilal Tahir (40:25.474)
Wow, this is like a proper gaming video. In my head, when you first mentioned it, I was expecting more of a 2D square globe type thing where you kinda get a one by one grid or something, but this is like a proper world. Yeah.

Pierson Marks (40:39.343)
Right. It's kind of cool because I mean, now comparing this what you're just talking with NFTs, like imagine you jump into a canvas, infinite canvas, and it's just a big grid empty, and you see people designing parts of the world everywhere. And then you go in, you zoom into like this one little grid square, and you go, hey, I want a little tree right here. And then you go to the next one, I want a little bench. And then I want like the start of every single...

Bilal Tahir (40:45.198)
Mm-hmm.

Bilal Tahir (40:54.317)
Yeah.

Bilal Tahir (41:04.875)
Right. I mean, it is basically what do call it? Mind sweeper? is it what's that game? Mind sweeper or not mind? Minecraft, right? I mean, it's kind of like, but Minecraft you have defined tools and stuff, but here they're obviously with AI, you can do pretty much anything you want. So you're using prompts instead of like chiseling, like using a shovel and you know, you building a house.

Pierson Marks (41:11.963)
Minecraft?

Pierson Marks (41:22.074)
Right.

Pierson Marks (41:27.017)
Right, No, totally. But imagine, so take that concept and then pair this with NFTs where, okay, so I'm in this vast infinite world and I am just prompting my way across each grid. Like, hey, build this, make this, make this, make this. And each grid is, know, a certain number of pixels large. But when I want to secure that frame, like I want to own, like this little tree, this little one by one grid right here.

I meant that NFT and then the next one I meant that next the NFT right next to it and I continue to these NFTs that could cost like a dollar. Let's say like every square you you you meant is a dollar or one cent or something. And then you kind of own it. You have the certificate of ownership of this one area in the world and you can kind of just like continue to build this world.

and own like, this is my little island over here. I own it. Here's the NFTs that I have proving my ownership. Just like kind of we do in real life. You have a deed to land, but now you have this. And you can kind of.

Bilal Tahir (42:29.486)
Right, right, yeah. I mean, the trick is like, how do you make it valuable? Like, why would anyone want to pay to get your land? guess, I mean, it could be, that's where the value of a network, like in making it a network is key. And, you know, maybe it's enough people are there and you spend a lot of time building a castle or something. I'm like, I want your castle because I don't have the time to, you know.

Pierson Marks (42:52.869)
Right.

Bilal Tahir (42:53.836)
prompt and I'm too lazy so I'm just gonna give you a dollar or ten dollars and I'm just gonna take your land.

Pierson Marks (42:55.601)
Thank you.

Your castle is so sick.

Bilal Tahir (43:00.418)
Yeah, I mean, is interest. mean, it's basically Minecraft basically has this economics already. I people will spend hours and hours building something and then people, there's a whole economy of people, of users, gamers who come in and buy stuff, because they don't wanna spend the time. it's interesting, it's a very interesting concept. Going back to Web3 World, there was this whole concept called Play to Earn, where the concept was that people in like Bangladesh or Philippines who, the gamers would spend

hours and hours basically leveling up your character in the game and there was a there were different types of games there was one called Axie Infinity which went bust that was the most popular one and then the idea was that somebody who from you know the West like a first world country who you know who would you know have the money rather than spending 20 hours leveling up they would just give them you know something which would be like for them would be a lot of money in dollar terms right and then give it and then they would take the money they would transfer the

character too so it was like play to earn but the whole thing fell apart because people once they realized I can make money everyone was just like they were just way too many people like playing the game trying to build stuff and just so there was an oversupply and there was not enough demand so it's a you know it's something you know I guess a delicate balance

Pierson Marks (44:16.145)
Hmm.

Pierson Marks (44:21.251)
Right. It's super interesting. It's cool. I mean, talked today. I talked about aliens and world building. We saw some, talked about, yeah, we talked about some cool stuff. Big week. mean, there's a lot of news in this week. And yeah, so.

Bilal Tahir (44:31.8)
Yeah, a lot of aliens and Web3

Pierson Marks (44:50.321)
I guess we could wrap it here. mean, maybe we can start playing around with these image models a little bit more. It's always like we have a limited amount of time. like, man, now my brain just goes down the rabbit hole. I'm I'm going to go build that now.

Bilal Tahir (44:53.196)
Yeah, yeah, no, it's I think so, please

Bilal Tahir (45:04.002)
Yeah.

Bilal Tahir (45:07.426)
Yeah, no for sure we should do that and yeah for anyone listening, please

leave us comments and stuff, if you feel like there's stuff you want us to cover, things we wanna do. mean, we wanna, we're still figuring out the exact format that works best. And so we're probably gonna experiment a lot with different things and hopefully we'll, yeah, like Pearson saying, we might show some code or actually like build apps and stuff, if that's something you guys would find interesting, let us know.

Pierson Marks (45:37.135)
Right. That'd be super cool. Yeah. And if you want to, if you're an expert in this field too, and you want to come on and, you know, demo, maybe that's, that'd be also cool to get like a guest talking about something in the space, like maybe the Lemon Slice founder, Lena, or somebody to come on and talk about image gen or video gen. That might be interesting. Something like that. Yeah. Sweet. Okay. Cool. Yeah. See you next week.

Bilal Tahir (45:58.69)
Yeah, yeah, for sure. Nice. See you, bye.