Episode 437 LIVE on Wednesday, May 10, 2023 at 11:30am PT / 2:30pm ET / 7:30pm GMT / 8:30am in Hawaii
Our guest this week is my good friend Dimi Lazarou. He is always pioneering new techniques and new technologies. He has been using AI (Artificial Intelligence) for years. He started making images before anyone else I knew was doing it. He has also figured out ways to use text and visual based AI to get the results he desires.
This might scare you. As I write this there is a writers strike happening in America in the entertainment industry. Some of the writer’s demands have to do with AI and how it is used in the industry.
There are a lot of ways AI will be useful to us. There are some jobs that AI will render obsolete. As we continue to advance do you find yourself embracing new technologies or resisting?
Do you embrace some and resist others? How do you draw that line? This week we will talk to Dimi Lazarou who is an embracer of AI in our field and we will learn how he is using it.
I hope it gives you a new perspective and helps you to understand more where you stand on the issue. See y’all on Wednesday, where your discussion in the chat will be welcome. (as it always is).
Did you know you can join us live each week? Sign up and get a reminder email 30 min before the show: https://creativesignite.com/signup
- Dimi, can you tell everybody a little background about you, who you are, where you are, and what you do?
- How long and how have you been working with AI? (Visually & Written)
- What has been the biggest hurdle you’ve faced in regards to getting the results you want?
- How do you deal with copyright? Or are there any copyright issues?
- Why are you choosing to use the AI generated images, rather than the images available from a stock photography website?
- New tools are always emerging, how do you decide what to pour into and spend time learning? Is change something you embrace or resist and why?
- What is the key to getting the desired outcome from ai, whether it is creating an image or creating text or body copy?
- What is one thing you’ve learned about ai in the last year, that has been most impactful to your business?
- What is next?
Connect with Dimi
Red Planet case study: https://www.behance.net/gallery/170679643/Journey-to-the-Red-Planet?
Our instagram account — https://instagram.com/symbolonbranding
My personal instagram where I have my AI art — https://instagram.com/reficulgr
Mid journey: https://www.midjourney.com/
Watch past episodes with Dimi
Last episode: https://creativesignite.com/359-2/
[00:00:00] diane: Hey everybody. Welcome to another episode of, I wanna say Design Recharge. Every time I said for so many years, you might notice, oh, it’s Creative Sign Night. Now, welcome to Creative Sign Ignite. I am here with my friend Dimi and I, he comes on multiple times. He’s been on multiple times, but we are not doing a, where are they now?
[00:00:22] We could have, where are they now? A series. But we are changing it up because Demi and I talk on a regular basis at least once [00:00:30] a month, and we were talking about AI and he’s been using like the Google engine to create images for years. Um, I’d say two years at least. Is that right?
[00:00:43] Dimi Lazarou: So. Uh, kind of a year and a half.
[00:00:46] I would say, like before that it was a very different technology. I have have been doing that as well, but uh, yeah, I’ve been using the diffusion models for images generation for a year and a half now, I think [00:01:00] since February, 2022. And yeah, and I’ve been using the language models, which is the, the text stuff, uh, from since November nine, uh, 2022.
[00:01:13] Yeah. November. So
[00:01:15] diane: that’s, we were talking about this and he kind of was talking to me and I was like, a light bulb went off and I was like, oh my goodness, I can use this in my classes. I can use this in, um, as I’m writing something [00:01:30] and then maybe it can help me make it better. Um, it gives me other options.
[00:01:34] And I hadn’t seen it like this. And I think one of the reasons, if you don’t know what Creative Ignite is, it’s really for people who are solo entrepreneurs or they’re entrepreneurs, they’re doing a lot of it on their own. They don’t have anybody to show stuff to. Um, we have other groups. I have a group that’s starting at the end of May.
[00:01:52] And um, that’s where we can kind of, we have a, a. You have a goal and you work through it in the next [00:02:00] 13 weeks. So again, so sometimes there’s these big things like the podcast and I try to bring people on that are going to, um, Hey, IANA IANA’s here. Um, that’s awesome. From Sylvania. Um, so anyway, I am excited for Demi to share with y’all because I, maybe I was, and as a solopreneur there’s always seems like we’re inundated with new technologies or new [00:02:30] programs or new something and sometimes I can be hesitant and I know Demi is not hesitant.
[00:02:36] Can you just really quickly tell them why you are not hesitant about new, uh, technologies? I know this is way down on our list, but um, I think it’s good, good time for it. And you’re in Greece right now, but you sound like you’re British, so give them a little bit of your background.
[00:02:55] Dimi Lazarou: Okay. So, uh, first of all, let just to, to get uh, the background [00:03:00] stuff out of the way.
[00:03:02] I’m dmi, I’m a co-founder of Simmon Branding, which is a brand identity design studio. Uh, I’m based in the UK generally. I used to live in London, then I used to live in Liverpool. My business partner is permanently in Manchester and we have been designing brand identities with the focus on semiotics, like this is our main thing.
[00:03:27] Um, the semi semiotics, [00:03:30] um, you can find, uh, preview the previous, uh, episode that I did with Diane, uh, somewhere around, I dunno, YouTube will sort that out. And, uh, yeah, it was all about the study of meaning. So this is, this is a good chunk of, um, of our, of our ex expertise in this situation. And I’ve been generally, uh, you know, thrust into technology since into my.[00:04:00]
[00:04:00] Early years. I’m, I’m very, um, I’m, I’m very much of a techno optimist, and I do think, especially my generation millennials, we have this situation where we’re, especially on the web, we have been web pioneers instead of, instead of, uh, web natives in, in some sense. Um, so we have, you know, uh, seen [00:04:30] the crossover between the analog and digital, and then we saw, uh, you know, a disconnected world and a connected world very much.
[00:04:38] And, uh, so just processing everything and seeing everything and understanding how, how tools can help us, uh, develop what we’re doing is, is really, really important to me as well. Uh, so. This is, this is the main, the main reason where, you know, I, I tend to try everything [00:05:00] in technology, especially when I think it will, it’ll help my, my work a lot.
[00:05:06] Uh, but even, even even more importantly in this situation, I’m usually very much of a cynic when it comes to trends. Like, I do lot like things that are here to stay. And I’m usually, uh, you know, I’ve seen everything like, uh, you know, virtual reality glasses, then augmented reality, and, uh, you know, it’s gonna be the next [00:05:30] big thing.
[00:05:30] And then NFTs are going to be the, uh, the next big thing. And that’s it. Like in general, I have this feeling where, you know, I’m, I’m not always swayed by these things. And I was, um, I. And I was reading in the same thing I was reading about the, the, the, the new technologies in, in ai and I was sort of, I understood what was going on, [00:06:00] but having the firsthand experience is very different.
[00:06:02] So when, when I started trying these tools out, it was so different so, so, so different. So, um, the first time the experiences, what changed me in this situation, I do think we, it is, it is something that will come to our lives and it’ll be equivalent or more, even more important than the internet itself, which is, you know, a [00:06:30] supremely a useful tool so far.
[00:06:33] diane: Okay. That was a good, um, intro and explaining. I’d love to know if you guys think. In the chat. Are you more, um, what did, didn’t we call it, what did you say? You were a skeptic.
[00:06:48] Dimi Lazarou: I’m a skeptic or a cynic or
[00:06:50] diane: something that, oh, no, you said you were like tech positive or something. I can’t remember. Like,
[00:06:54] Dimi Lazarou: yes, I’m, I’m, I’m, I’m very techno positive in that regard.
[00:06:59] Like I. [00:07:00] I do believe in technology in general tech
[00:07:03] diane: optimist is what you said, IANA way to go. I knew. Okay, so if you could write in the chat if you are, you think that you are tech positive or whatever, that sounds like we’re like, we’re testing positive for something. Um, or if you are more of a cynic or a, a laggard as maybe, um, Simon Sinek would say, um, you kind of wait and see and you maybe you’re, you come in with the, the majority [00:07:30] or something.
[00:07:30] Um, so we got some tech positives. Um, lots of people are excited to hear about this discussion. Matt, I believe you’re Western Massachusetts, right? Did I get everybody? The Pam’s a wait and see techie. I love that. Amy Darling in Colorado, she’s a tech positive. Um, tech skeptic, Shelby Arnett from Maryland, at least for the hot new trends.
[00:07:53] So, In that, uh, Maya is a wait and see. In that, just kind of [00:08:00] forming this, how do you know you personally, Demi, how do you know what to push into and what not to do you like with NFTs? Were you like all in or what wasn’t?
[00:08:13] Dimi Lazarou: With, uh, with NFTs, I was completely out of it. Um, like, uh, in the beginning, in the beginning, I, I sort of started understanding that this could be, um, a useful technology in very niche situations, [00:08:30] especially in contract law, which is, which is important, uh, but at a certain, at a certain situation, the, the idea of everything turning into a grift.
[00:08:41] Um, what’s that mean? A grift? A grift is when some people try to take advantage of a trend, uh, without offering much substance. I think, um, I, I, I dunno if it’s a UK versus US English [00:09:00] thing. Uh, but that’s the, that’s the, the main idea of grift. Like, it feels, it feels like the shop sellers in the, in the gold rush, uh, type situation.
[00:09:10] So I’m very, very skeptical about, you know, people who try to pretend that this is going to change. Um, I’m like, AI by itself is something worth, um, you know, uh, switching to and doing that. Well, AI [00:09:30] is, is, it’s like the internet. How, how stupid would somebody, uh, You know, calling themselves internet expert sound today, right?
[00:09:38] It’s just the, it’s a technology that’s helping us, uh, do certain things. And I do believe, um, that there is already a little bit of a grift in, in, in the design space about this. But for me, and especially for the purposes of this, this conversation today, I’m all about using AI as [00:10:00] designers. Like what will.
[00:10:03] Be of practical help to our tool set, our toolkit and, and the way I see this. So
[00:10:09] diane: yeah. So you have a presentation for us. Thank you. And I’m gonna just interrupt you and you guys can interrupt as well, just by putting something in the chat. Thank you guys so much for being here. And if you didn’t know, I’m in the church basement pretty much.
[00:10:24] So, um, I’m just thankful to have a good place to have strong internet. Woohoo. While I stay [00:10:30] with my dad. Um, okay. Timmy is gonna give it to us. If you guys have questions, just I will be monitoring the chat and you guys can just, um, ask, ask away. And Dimi, I’m just gonna interrupt you if,
[00:10:45] Dimi Lazarou: if we have something Yes, please do, because I’m gonna have access to the, to the chat now on screen sharing.
[00:10:51] Okay. Um, When I started designing this presentation, like I had this, I had this idea of, you know, calling this, designing [00:11:00] with your mind. I was, I was meaning to talk to you about this so you could put it on your marketing thing, but, uh, I, I, I thought that nobody’s been to understand the implication of this should, let’s, let’s talk about AI so it can, can draw the attention it needs.
[00:11:15] Uh, but I’ll explain myself right away. So, um, I have been studying semiotics and, um, you know, how do we mark things and what is design and what is communication in [00:11:30] this situation? So this is, this is a beautiful picture. Um, all nature, like, and there is, there’s nothing designed in this, in this premise. The only thing that is, has been somewhat designed is the frame that the photographer designed, uh, designed it to, um, to, to make their, uh, to make the frame about.
[00:11:54] But in itself, there is, there is chaos in here, and this chaos [00:12:00] is reminding us that there is, there has been no inter intervention here. So this is something different, like the, this is, um, equally a picture of nature, but now there has been some artificially placed order, which is, you know, in this situation it’s a mirror thing.
[00:12:20] It’s, it’s very, it’s very immediately obvious, but everything in this picture is generally natural, is putting things in order. Is, [00:12:30] it’s, it’s arranging things in a different unexpected non-car and non-natural way, which is the, the point of design in this situation. So designing is about making these choices.
[00:12:45] Like it was really, really important for me to, um, art give you this context in order to, to assess what is going on with, with artificial intelligence and how it helps us in design. So in order to [00:13:00] make, in order to make pictures in this situation, we, we use certain tools. This is an Unreal Engine five.
[00:13:08] This is a video game development tool. Um, the, the people use it to make worlds and video games and put trees and make all these mans and water do their things and. What is very familiar to, to branding designers and everybody else is that we use this sort of [00:13:30] interface of panels on the right, on toolbars, on the top.
[00:13:33] Uh, the menus like this is, this is the language that we have been using so far to communicate with the computer and tell it, okay, I need this, this rock in on the edge. I need the, the water to, to have this, this kind of reflections, et cetera. So, um, this has been the interface, uh, we have been using up to, to Delhi.
[00:13:57] And the big thing about [00:14:00] artificial intelligence in this situation is just it can enable us to create a different, a different kind of interface. So let’s imagine that we we’re, we’re in control of the matrix, right? We’re in control of the, um, a very complex system that can generate. Anything we want in this situation.
[00:14:24] Like for the designer, the Matrix is a very, very powerful design software. So [00:14:30] we start from the beginning. We, we, we have, we have a blank slate, which is terrifying for most designers. And now, now we have to, um, now we have to make things happen. So, uh, the most intuitive way in order to, to make, to make something in this situation and, and matrix-like situation is like asking it make me a house.
[00:14:56] So if you ask it to make your house, just gives you a house, [00:15:00] the most boring, the most undifferentiated kind of house, because this is the only thing you’ve asked for. So the point of the designer is to further add specificity and more specificity and more specificity to what they’re asking for. So, You can ask for a red roof, you can ask the house to be, um, to be two stories sold.
[00:15:27] You can ask it to be in the desert. You can ask it to have [00:15:30] a, a tree. Uh, you can ask it to, to have red trees, uh, to be in summer, winter, like making the choices is what gives us the, the design that, that we need. So in this situation, what AI changes for us a lot is that it takes us from a situation where our tools are panels and icons and [00:16:00] tick boxes and menu balls, and it enables us to use natural language in order to make pictures.
[00:16:08] So these four pictures have been made with artificial intelligence and. Basically, the more specific we are with these, the more, um, like the desired effect, there are going to be other ones. They’re, they, they aim to be as generic as possible. So there are, there are, there are two things [00:16:30] that have been happening in current technology.
[00:16:32] The one is language generation. The, the other is image generation. Both, both are really, really important for designers. So I will talk about the, the most important tools and get them out of the way and then show practically how, how we use them. Language generation. The most important tool right now is chart G P T four.
[00:16:53] Um, it generates language. It can, uh, you can ask it to write an email. You can ask it to, uh, to give [00:17:00] you recipes. You can ask it to do a lot of things. It is a language generational tool. It’s, it’s not a search engine yet because it’s prone to hallucination. Like if you, if you ask, you can ask it questions and it will try, its best to answer.
[00:17:16] But sometimes just because it is about creating the language, if you ask it something, um, it, it might give you a, a, a bs kind of answer, which is, it will make up stuff that [00:17:30] sound plausible. Like you can ask it about, you know, penguins in Australia and it’ll, you know, give you something that sounds really, really plausible.
[00:17:39] So for people using this focus on creativity, not productivity, like, don’t depend on this, uh, so much. And the, the fourth, the fourth generation of this language model understand context up to 32,000 words, which is like 80 pages of a four paper. [00:18:00] So let’s get this out the way, go to whoever’s generation.
[00:18:03] The, the tool I use and the, the tool I, I think it’s strongest right now is mid journey. It creates, Images from noise. I will talk to you about this. The, the main point is doesn’t understand context at all. Like if you, if we, if we go back in this situation and it generates this image with the house, with the red roof, you can’t ask it to do, just do this roof gre [00:18:30] like, it doesn’t understand the context.
[00:18:32] Every, every image generation prompt is, um, is uh, is new. It’s happens again. So the the times are rapid right now. It used, it didn’t use to be the case. I remember when I was, when I was using stable diffusion on Google collab, uh, notebooks, um, I would be a, i I would be having something like [00:19:00] an image a day or something like that.
[00:19:02] And it would take, it would take, uh, about. Two hours to generate, and then I wouldn’t be able to use it again for, for several hours. Uh, but, um, now the times are rapid, JE could generate four images in, in one minute, and it’s excellent for placeholders. So I’m, I’m just giving you information. I will come straight back to it.
[00:19:28] I just wanted people [00:19:30] to understand what tools I’m going to be, uh, demonstrating in this situation. So,
[00:19:35] diane: so let me ask you this doc has this question, if don’t answer it, if you’re gonna be covering it, but it says, um, doc says, my understanding was that you could reede that image and ask it to make it a green roof.
[00:19:46] Is that incorrect?
[00:19:49] Dimi Lazarou: Uh, it, it’s not incorrect, but for, for most intents and purposes, for people who are going to be using it for the first time, if [00:20:00] they have experience with chat, g p T, that. You know, directly, you just, you know, you type back your, your feedback about what you’re getting, um, it’s not gonna work.
[00:20:11] So I didn’t want people to feel confused. There is there, there are advanced things that I’m not going to capture today because, um, it does, it does warranty a few hours of your study people and your research, uh, you know, over several weekends, I guess. Um, [00:20:30] but, uh, I, I just wanted to, to show you option things.
[00:20:33] So the big, the big situation is how does this work? Like, how does the machine understand what is going on? The thing is, it doesn’t, uh, so first of all, like, um, uh, it, it is like, uh, neural digital neural networks. They work as emergent intelligence compared to centralized intelligence like humans have.
[00:20:57] They are more like ants. So [00:21:00] ants have this way of working together without each ant having language and coordinating with another, or there’s a master, uh, ant person working on things. So a ants work with pheromones. They understand what is happening in their immediate surroundings, and based on that, they, they use certain paths again and again, and again and again.
[00:21:23] And repetition and multiplication of these repetitions is what generates [00:21:30] this emergent intelligence. So this is, this is, this is the way neural. Networks and the computers do this as well. They don’t really understand yet what, uh, what we are talking about, what they’re giving us. They just behave as a, as a whole and colony in, in certain respects.
[00:21:51] Uh, but none, uh, nobody, you, you can ask no and on on what’s going on. So the way, the [00:22:00] way they work is, uh, especially for image generation, essentially, there’s a, a bunch of switches which are randomly turned on and also, and on and off. And, uh, the, the, the, the computer tries to understand. Um, it, okay, understanding is the wrong word.
[00:22:21] Uh, it tries to use statistics to guess. If the result we want, uh, um, can be, can be [00:22:30] acquired with slipping these switches or flipping these switches. Um, I can give you a sample of how it works. And especially in the early days in the, in the early days of stable diffusion, this was actually one of the first steps, like this was actually something we would get people using me journey.
[00:22:49] Now don’t really observe the, the vagueness of things. So if I, if I click on, on this to play and we can, we can pause it several times. You can understand. So first the [00:23:00] computer flips some switches at random generating noise. So these switches in visual, uh, in, in these visual elements, like they have 256 uh, points.
[00:23:12] So each of the corners is based on something like that. So, So based on that, they generate this noise. And when I click on it, you will see what, what you’re going to be seeing is the opposite of, of, um, well, the equivalent of using, [00:23:30] sharpening your image in Photoshop. So it, the computer first tries to find edges, and then from the edges it tries to make this into a coherent image.
[00:23:40] So you can see how this is turning into this ballerina from, uh, you know, inspired by
[00:23:49] diane: Star Night. Yeah, yeah. Star,
[00:23:51] Dimi Lazarou: yeah. So you can, you can see this again. I’ll just play it one one more time so you can understand. This is just finding the edges and sharpening [00:24:00] them, finding edges and sharpening and just changing, like the ants in here are trying to slowly manipulate the, the, the, their very neighborly pixels in order to, to generate this image.
[00:24:14] So this is, Vaguely how it works. I don’t want talk
[00:24:18] diane: too much to, and you, we put in text, right? Yes. You’re putting in text for this. You’re, um, yes. And when you were first doing this, I, I would get these images from Demi, which I thought were [00:24:30] awesome. Um, and he was like, in the style of blank and blank and do this and this.
[00:24:36] And so, um, there was some interesting, uh, it really is a lot about language and for, for me, this is a harder, um, you know, I, I can see it, you know, and a lot of our clients will say, oh, I know it when I see it. And you’re like, that’s so frustrating. That doesn’t help me at all to get to what I need. So, um, So [00:25:00] how, uh, IANA says, how does it know how Van Gogh looks like?
[00:25:04] I guess you can
[00:25:06] Dimi Lazarou: tell him. Yes. So essentially, essentially what is happening is, so you, you ask the question and you say, please make me ball, and the style of van, oh, and what is essentially tries to do is guess whether what it gives you is manhole or not. So that’s the, that’s the big challenge. So, um, [00:25:30] this has been developing over the years and the big, big, big problem with it was we didn’t have enough people telling, telling it, this is not what I want, but applied to millions of s in this situation, uh, with, with many, many times, and giving it the reinforcement we, we want it can start to, to generate these things.
[00:25:53] So, um, so it has been. It has been specific images that we [00:26:00] have been answering to computers for 10 or 15 years. Like, do you remember? Most people will, um, will remember that, you know, back in the day there was this, um, this challenge by Google, uh, fire, find the fire, uh, point me to the buses in this image.
[00:26:19] Like how many, well, how many wheels are existing? This thing. So, uh, this is part of the human reinforcement training in this situation. So that’s, [00:26:30] that’s the, the way it comes to. So, Um, last year. Last year, uh, so when I started, this was the peak of, of technology back in the day, like the, the most I could do with this kind of, this kind of tour.
[00:26:47] Um, this was one of the first images I, I generated. Like, this is a Greek temple and how long
[00:26:53] diane: would it take to generate something like this? It took three hours.
[00:26:56] Dimi Lazarou: It would take three hours. Wow. About about three hours. [00:27:00] And now things are very different. Like all of these have been made with mid journey fairly recently.
[00:27:07] Uh, it’s not even the, the latest version and the, these things generate in under, under a minute by themselves. It takes me a little while to, to customize my prompting, but this is, this is mostly how it works. So I will, uh, we were supposed to be talking about copyright and, [00:27:30] and, um, you know, how, you know, how the, the training itself in this situation fits, um, if copyright is, is a challenge, it’s not.
[00:27:40] I’ll just give you a series, series of pictures and then we can, uh, go back and discuss them. So this is, uh, this is, uh, Uh, I call it, um, you know, the thing is just ready made. Uh, I have call it, this is how it’s called. We have [00:28:00] remixes of things. We have depiction of existing things. We have inspired by things like this picture, for example.
[00:28:12] And, uh, I’ll talk to you about this old gentleman and we have to understand the, the limits of copyright. So in order to ready made images are the most rhythm made. Art [00:28:30] is the most challenging in terms of the. In terms of the copyright, like this urinal was made by someone else and it has been appropriated as, um, already made art.
[00:28:46] Data is object and art has been infused into it in this situation. So I think this is the most, the most extreme of situations in this, in this spectrum. So remix is [00:29:00] slightly better. You still use images somebody else made, but you’re changing them in a very sufficient way. And this is also data’s art. Um, nobody had any challenges calling this art or, uh, calling any copyright, uh, claims in this situation.
[00:29:18] And then you have depiction, which is different, like you create unique. Things, not by using them, by using the object itself, but just [00:29:30] depicting the object itself. And then you have something like, um, inspiration, which is, uh, you know, um, this is, uh, Danita, uh, by, uh, Picasso. And Picasso was, was not present in the bombing golf guga.
[00:29:50] He had to, he had to find out by news reports and photographs like this, which is one of the main inspirations for this piece. This is also whi [00:30:00] the pieces in black and white. So the situation is like, is there, is there a valid reason for, um, you know, uh, for the, the producer of this photograph to be, uh, Upset in the situation when, uh, when Picasso is using his work as inspiration.
[00:30:22] So from what I’ve understood, we are not even there with, uh, [00:30:30] we’re somewhere along this, this range. When we talk about artificially, uh, generated art or pictures, uh, I don’t wanna call it art, if it is an artistic endeavor is something different. But in terms of in, in terms of productions of the pictures, essentially the computer doesn’t, doesn’t combine images from somebody else.
[00:30:54] It’s not a remix like this. It’s not a depiction like this. It is inspiration. [00:31:00] So, um, it is for people who have been using it for quite a while, they will understand why. Um, that the pictures, the pictures that are inspired by that the computer does not take them and merge them or remix them in certain situation because we, we have observed this phenomenon of the many fingered men.
[00:31:25] So the, the situation is none. Uh, so it is [00:31:30] very common that in pictures generated by mid journey or other artificial intelligent, uh, engines, the number of fingers in a person’s hands are wrong. So the thing is, nobody paints people with the wrong number of fingers, or nobody photographs people with the wrong number of fingers.
[00:31:52] The thing is, The, the computer is inspired. It tries to understand when you flag something [00:32:00] as so this, this could be, you know, flagged as dramatic. Uh, you know, the inspirations might be flagged as dramatic. So when it generates it, it tries to make it like the pictures people flag as dramatic or black and white or, uh, painting or, um, the old man.
[00:32:19] But it’s all about the flas. It doesn’t, it doesn’t take any images and blends them around in a mixer and, and make this is, this is how, this is [00:32:30] how the, the software is generating these things. I guess we have questions. I see the number
[00:32:35] diane: of, uh, it says that, um, uh, uh, doc said it, not sure. I would say the computer was inspired.
[00:32:42] Um, and then Iana says, what about the part where people say it has been trained on 5 billion images,
[00:32:49] Dimi Lazarou: but this is, this is true, but the thing is, the, the training is done on marking these images as a certain kind, and it just tries to [00:33:00] produce it. So in, in a sense, it’s not, uh, it, it is about taking these pictures and humans marking them, and then it tries to compare what it made to the markers that other people made.
[00:33:15] So, um, this is, this is the, this is the bit which is, which is. It, it, it, it is like a, uh, a student trying to impress their professor, trying to [00:33:30] do what the professor wants. And based on how the professor mark the other students, the, the, the kid is doing his part to, to write an essay that the, the professor will, will, like.
[00:33:45] None of the words specifically have been taken by the other students. Trying to understand what your assessor wants is, is what is training the system in the situation. I hope I’m explaining you [00:34:00] this right? Yes. So I have this image of many cranes. So there, there is the other thing, the other conversation is, is, um, you know, our computer’s going to, to take our, our livelihoods and if this is something we want, so, So this is for me, one of the, one of the, um, staples of nineties [00:34:30] ni growing up during the nineties, like is the many colored crayons, which is a very new phenomenon in artistic endeavor.
[00:34:39] Like we didn’t, we didn’t have that in the past. People used to mix their by and the existence of, you know, uh, uh, a set of many corners, uh, ready to be picked without needing to mix, without needing to know that you need to, uh, need, you need blue and yellow to [00:35:00] make green, or you need this, um, you know, this rock to, to merge with egg.
[00:35:05] Um, it, uh, having crayons of multiple colors have hasn’t taken from our creativity in this situation has just added, uh, simplicity to the use of many more people. More kids get to play with cranes than, you know, in the, in, in the Renaissance where, you know, you had to find cobalt and blend it and [00:35:30] just put it in.
[00:35:30] Oil painting is just so different. So it’s again, all the same with, uh, you know, with photography. Like, um, um, I, I don’t think, you know, having pictures taken at the party takes from the very old school way of having to find an expert who knows, uh, who knows everything about metal plates and photographic film and, um, and composition and everything like it.
[00:35:58] Uh, I, I don’t think [00:36:00] that having the acc giving access to many people to do the thing I, I don’t think it has ever taken from our creativity in this situation. So the, these two, these two examines of my statement were on, on whether, uh, were going to be, you know, completely suppressed by, uh, um, MUEs or not in this situation.
[00:36:21] So I have. I have a few ways to, um, I have a few things to say about how we can, we can use [00:36:30] image generation, uh, as designers, but are there any questions yet? Nope. Keep
[00:36:35] diane: going. Okay.
[00:36:36] Dimi Lazarou: So, um, so one major thing, uh, that we can immediately use, uh, mid journey for is generating compositions and understanding. So this is a grid, uh, it’s based on chat moder.
[00:36:51] Uh, again, uh, it was, um, really important in developing grids. I, I, I guess, uh, so we can, [00:37:00] we can use, uh, we, we can use images we generated in order to guide us in more, um, unconventional ways, which is, is, uh, something that I do very often now, now that when I have to design a poster, when I have to design something, I can.
[00:37:19] Uh, I can be inspired by different unconventional grids in this situation. So I love this, uh, this kind of thing. So I would recommend it to everyone. [00:37:30] Um, there are also more, um, deliberate things about, you know, you want to, you, you, you want to come up with a layout about something you would just want, uh, uh, some, some way of setting up the pages.
[00:37:43] Uh, you can, you can be inspired by different, uh, by different outputs. So these are, uh, financial magazines. Uh, the, the problem was financial magazines, um, websites, I think. I, I think [00:38:00] this is, this is where it demonstrates, you know, there is a certain, uh, homogeneity, like what you would expect to find in, in, uh, um, a dribble or different, uh, cycle like that.
[00:38:13] But, um, generating layouts, generating colors, generating ideas on how you could approach a project itself visually, um, is very important. So I, I do think. We’re, uh, we’re going to be seeing this. So I recommend all these things [00:38:30] just generate layouts, generate, uh, the grids or, you know, even more obstruction compositions.
[00:38:36] This is, uh, this is something that, uh, works. So, um, but I can also show you how we’ve used, uh, how we’ve used the language generation models in, in our business and, and how, uh, how it helped us develop branding for, um, a self-initiated project. So this is, this is the idea of what would happen [00:39:00] if we could grow KFI and MAs.
[00:39:02] Uh, and so we, uh, my business partner and I, we, we have this thing where we, we want to, we want to make a coffee packaging, uh, project and we didn’t have a real live client in order to do this. So we. We decided to, uh, to make, uh, judge g p t our client and start working on branding based on that. Uh, so we could run all the workshops, we could run all the questions, [00:39:30] understand, you know, make, uh, a business out of nowhere, out of where it didn’t exist, and just testing scenarios and asking at different things.
[00:39:40] Uh, so this is very, very, very useful. So I, I will,
[00:39:44] diane: I’ll, so in this, um, cuz this is what really clued me in on the possibilities. So when you were telling it, Hey, I want you to be our client, right? How, uh, one thing I love about Dimi is when he’s speaking [00:40:00] to the machine, is what I’m gonna call it for right now.
[00:40:04] He is very polite. Hello, how are you today, please? Thank you. Um, and he has that conversation and I was like, why? Because he was just doing it on the screen for me. And, and tell ’em what you said.
[00:40:19] Dimi Lazarou: So, so the thing is for, for text, the way it works, it works like predictive text in your phone. So essentially when you’re typing up a message [00:40:30] and your iPhone or Android phone just, uh, prompts you to use this word or that word, this multiplied the few thousand times is what Cha Chatt is about.
[00:40:40] So it’s predictive text. So when, when you use in your prompt, uh, words says, please, thank you. Good morning, goodbye. It understands this, um, this, uh, willingness to, to to be, to be part of a businesslike context in this situation. So you are [00:41:00] being professional with it. So it gets the prompt that the language you, you wanted to generate is gonna be professional.
[00:41:07] Um, so that’s the, the reason, that’s the reason if you’re being impolite, it will lead you to conversations where. Your, your counterpart in the conversation is also impolite. So using please for
[00:41:21] diane: abrupt, right? Abrupt, yes, exactly. Maybe too direct if you
[00:41:26] Dimi Lazarou: unhelpful in this situation. There are, there are biases in [00:41:30] this, uh, developed by the, the developers of, of ity.
[00:41:33] But, um, it, uh, generally it would, even in an unbiased model, it would pay to be, uh, to be, uh, timed,
[00:41:43] diane: I guess. So in, in the setup, how long was your paragraph or paragraphs of text that you set them up as the coffee? Uh, or is that it? So we are building a case study. Is that all you gave them?
[00:41:59] Dimi Lazarou: So, no. [00:42:00] Uh, so all of this is just, um, so we have a small screenshot here, but you, you will see how it works.
[00:42:06] Like we, we developed a lot of, a lot of different things. So basically we told it, uh, we’re building a case study. We’re, can you help us? And you know, it, it went with it and then we tried asking things about the company and how it works. Uh, so, um, we tried to understand the business model. Uh, so all the images generated from now on [00:42:30] are also mid journey.
[00:42:31] Uh, I will tell you the point will stop, uh, but, uh, business model canvas is essentially, Uh, a way that we already use with our clients to understand their business. So we used it with, and it gives us, it gave us much more than what I’m giving to you now, but essentially, you know, who are the key partners for your business?
[00:42:52] So it can be nasa, can be sustainability partners, or who is your client? Our clients are high coffee, high end [00:43:00] coffee enthusiasts, environmentalists, uh, whoever. So we, we basically run this conversation the same way we would run with, with our clients, but it didn’t have performance anxiety. Uh, it was, uh, to the point really quickly.
[00:43:16] Uh, we have been. Uh, using it, uh, all, all everything you see now, we have been using it alongside our, our real life clients as well in order to supplement some, some of the time [00:43:30] that we don’t get to have it.
[00:43:32] diane: How long did this take from start to
[00:43:34] Dimi Lazarou: finish? Uh, it took about two hours, uh, from start to finish, uh, just because we were.
[00:43:41] Trying every, every workshop and every, every, uh, exercise in the
[00:43:46] diane: book. I, this is the stuff that I love though. This is the stuff my friend Faye and I, we’ve, she’s like, had real clients and she’s put them in and, but I love, so I love these images and I didn’t know that these were AI because of the [00:44:00] hands. I knew the thing with the hands is weird and AI doesn’t do hands really well.
[00:44:04] And I was like, well, where’d you find these images? And he’s like, I made them on mid journey. And as you see them all together, this was one of the images that I used for the promo. It, it has the same sort of feel and this background and I, I definitely want Demi to explain how he got some of these, cuz he in the image before, so we’ll have to come back to that one if, if we don’t address it in a minute.
[00:44:27] But I love that he’s able to [00:44:30] create these profiles, these customer profiles with artificial intelligence. Because usually what we’re doing is just going through stock photography or Unsplash or something and we’re just trying to find something that is sort of what we think. But I love that you can create this feel, you can create the kind of cup that they would be serving.
[00:44:51] You can, you know, have the professional person and you can really. Get what you’re wanting more. [00:45:00] Anyway. I love that the images were connected to chat, G p t telling, um, Dimmi about the company that they were making this brand for. Okay. Keep going.
[00:45:09] Dimi Lazarou: Yeah, so for, for us, and it was really important to understand who Sarah was.
[00:45:13] Again, this is not. Every, this is just, uh, 10% of what we, we manage to develop as a, as a profile. And developing audience profiles with our clients has been the most, uh, taxing, uh, in terms of time with them. So it [00:45:30] usually takes 50% of our time with them and Wow. So many more questions because they’re slow to respond and don’t really know what we’re looking for.
[00:45:39] Uh, in this situation. They sometimes, our, our clients don’t have the imagination to just develop a, a person from scratch and, you know, find out what brands they like or what are they like or whatever. So we developed this. And with the, with the help of a real client, we can give them these profiles almost [00:46:00] in real time.
[00:46:01] And we can modify these based on their feedback rather than, um, rather than ask them to create this from scratch. And this can be game changer. Uh, so we have two profiles. Uh, uh, one is US based, the other is, um, UK based. Like we we’re trying to, to understand what kind of people would go for this brand.
[00:46:25] And then we use the exercises that we have to create delight. [00:46:30] So how do we create talent plans to, to reach our target markets? And this is very important. So, um, the, the, the creativity of the language model and give you ideas on how do. I as a brand, make something that Sarah will love. So this is, uh, a list of suggestions that our clients now can, well, in this situation, they’re, the clients are fictional, but our clients can take and immediately, you know, [00:47:00] start, uh, assess employee.
[00:47:02] And again, the computer itself is hallucinating, doesn’t know what it, what it, what it says. It’s not, it’s not accurate, it’s not proven. But these situations, they’re. There are some, some things that are real clients can assess whether, you know, having a loyalty problem, uh, program is, is gonna help or creating a community platform or whatever.
[00:47:24] So wait,
[00:47:24] diane: wait, wait. We have some questions, so Yes. Before you keep going. So what, um, Tate wants to [00:47:30] know, like, we’ll get back to that. I guess when we get to the images, and I’m gonna write it down Tate. Um, but Alexius says, could AI help fill the gap with a client if a client used it in a similar way? Like, if, if so, a client’s having trouble coming up with the persona, their typical client, typical customer, could they use AI or would you.
[00:47:54] I Is that something that you would, we could a, have our clients take? Well,
[00:47:58] Dimi Lazarou: the, the, the [00:48:00] thing, the thing is, uh, it’s all about asking the right questions. I think the, the expertise of the brand identity design is just to make sure they have the, they have the good questions in order to do that. But the, the client can absolutely, um, you know, if, if as a client you’ve seen this many times, or especially if you have a branding team or things like that, you can absolutely ask the right questions and just go along with it.
[00:48:27] Uh, because, um, yeah, it [00:48:30] can, it can give very, very valuable, uh, insights to the client. So, yes, please. Our clients shouldn’t, uh, should ask it, uh, should ask it, uh, right away, so, yeah. Um, so. Then we started developing during our brand attributes, um, uh, exercises with them. So we have this thing where, uh, it’s based on, on, uh, core, which is an existing framework.
[00:48:59] We have, [00:49:00] um, we have a conversation about words that describe our audience, our voice, our impact, our uh, our culture, our emotions, and we, we use this in order to. Uh, and, and we, we use the machine in order to generate them. And also, if you are using similar exercises, now we have been tempted and we want to test with our appliance whether instead of asking them to come up with these as we go, because that’s the, the form of the [00:49:30] exercise, we can give them, uh, a list of, you know, 200 and ask them to pick, uh, what they, what they feel is more relevant.
[00:49:38] So it just, it, it eliminates performance anxiety in terms of the clients, uh, because they, they’re not sure we have a, a timer of two minutes, like what is going on, uh, just giving, um, and the sometimes the second guest themselves, we can eliminate all that, uh, just by, um, you know, doing some, some, uh, [00:50:00] some work before we meet with them.
[00:50:02] So we’re prepared in, in our conversation about this. So, so based on these. We take them and we put them together in a brand statement. So this brand statement, we did it by hand. By using the, the, the words that have been already artificially generated, uh, before. So this we did by hand because, um, the format of the brand statement is [00:50:30] still kind of, um, uh, kind of limited in this situation.
[00:50:33] So it doesn’t, you don’t really need the machine, you know, in, in order to make it. Um, so this, uh, this is the logo that we, we developed for this business. This is, uh, the planet Mass, and it’s two, uh, moons, Steven and Foggo. Uh, so, uh, and the type first is sinking sand. So all of this has been generated by us.
[00:50:58] It wasn’t made by. [00:51:00] Um, mid journey or an artificial intelligence, like we came up with a symbolism and the way it should work and, and how, um, how it gets put all together. But the thing is now we have this, uh, you know, in order to make the brief, it took, uh, seconds instead of, uh, instead of ours in, in this, uh, situation.
[00:51:22] So again, the color system, uh, developed by us, uh, it wasn’t, it was made by ai. [00:51:30] Um, the, you know, uh, some information, but the look construction is general things. And again, the packaging we came up, yeah. Yeah, so this is, uh, so basically is what we wanted to, to showcase. So in all of the packaging we have metal, uh, elements with a, a alongside a splash of Martian dust.
[00:51:52] Yeah, I love that. And, uh, which is, uh, both inspired by, Itself and [00:52:00] by the ground of Mons in this situation, um, the, the machine helped us, uh, come up with, uh, some BS about one, you know, which station facility and which source and which farm in Mars. All these are when, when it’s harvested in winter 2037 and things like that.
[00:52:19] diane: So did you, did you generate this image in ai? No, we didn’t
[00:52:22] Dimi Lazarou: generate, no. This image, this image is a mockup designed by us. Okay. Uh, well, the, the, the only thing that [00:52:30] is generated here is the, essentially what is loam ipsum, but in a very specific way is just coming up, uh, coming up with real sounding, uh, real looking text in this situation.
[00:52:42] So all of this was designed by us. Uh, same for, um, uh, same for here. Mm-hmm. Like mm-hmm. These are the ground, the, the ground coffee, uh, or in beatings packaging. So again, cold brew. Um, all the, the taste profiles and all this have been [00:53:00] generated by the ai, the barcode is, is working, actually takes, uh, it takes you to our website right now.
[00:53:08] Nice. And, uh, so again, different, uh, different diff different, uh, pieces of packaging that we, we made again, with a, with a, um, ground, uh, of Mars. Uh, we had some ideas for cold brew. All of these are designed by us. It’s just, uh, we, um, it made our lives so much easier, uh, [00:53:30] to do this in, in this way. So were,
[00:53:32] diane: were the mockups already existing or did you make Yes,
[00:53:35] Dimi Lazarou: the mockups were already existing.
[00:53:36] They’re from Envato. We, we bought them like, uh, right, like we, we used to do, but essentially like in terms, in terms of, of using loam, ipsum like this, uh, this is beyond, you know, grapes and it helps. So these are, uh, pan, uh, um, for cold brew coffee that can be, you know, uh, certain in, [00:54:00] in a really nice way. So, um, container, uh, for
[00:54:05] diane: this, okay, so can you go back to the pencil?
[00:54:08] Yes. Um, Sarah and the guy, I have my text, my chats over his name, so I don’t know, but Tate asks what kind of prompts or, or go back to the one where it’s like the. It’s like the GoPro. Can you do that one first? Yeah. Because that was one I, I asked the almost exact same question, Tate, when we were talking before.
[00:54:27] Yes. So Tate Ash, what kind of prompts are you [00:54:30] getting? Are you using to get the photos so real? So tell ’em kind of what you were getting here using to get these. Okay. So,
[00:54:37] Dimi Lazarou: so in order to make re realistic images, you have to understand, again, this is, this is all about, this is all about the tagging and not the images themselves.
[00:54:47] So many pictures of the internet are clear, but they’re not tagged as clear. So most of the tags used by people to mark these images is, um, are things about, uh, about [00:55:00] am lens camera purchase. Uh, so I can, I can put the, the whole thing in the. Uh, in the chat, uh, in just a second so I can, you can find it, but it’s mostly keyword words that have to do with, you know, GoPro and F1 stop and, and all that.
[00:55:20] So I can just give you the whole prompt here. Can read it to you as well. Uh,
[00:55:26] diane: that would be great because I think this is, this is where [00:55:30] Dimi using this for a year and a half or using an image generator for a year and a half, he has, this is what he’s talking about us practicing. It’s not like we, our first day on mid journey are gonna probably be able to create something like this.
[00:55:46] But one of the things I asked about the other image, like with Sarah, I said, oh, did you find her? And he said, no, these are all, um, generated, which I also, [00:56:00] uh, I thought that was. I thought that would be really nice for us to be able to create something that wasn’t someone else’s photograph. And that’s what I asked.
[00:56:09] Uh, and when Demi and I were talking about copyright was, so we have, I, I said, is the computer just taking somebody else’s photo? And they’re doing so can you explain that a little bit? Yep. Because it is completely new, not. There’s probably not a person that looks like Sarah.
[00:56:29] Dimi Lazarou: No, Sarah [00:56:30] Sa Sarah doesn’t exist.
[00:56:31] Uh, so that’s, that’s a hundred percent unsure right now. Uh, what is, what is happening is, uh, basically we we’re using, we’re using, we use the same things, uh, that, uh, Dan will, will post in a, in a second. So, um, basically, uh, the, the machine is trying, is trying to guess if this is what I want and if I’m very specific about what I want, it’ll try, its best to [00:57:00] do that first and then, and get, then give me the answer.
[00:57:03] So, um, uh, I can, I can read it to you as, uh, um, epic, beautiful scene, cinematic post reduction, depth of field CMA photography, 55 millimeter lens. F uh, eight ISO 100, shutter speed lighting, accent lighting. Like there’s probably a hundred, a hundred keywords that we put in this in order for this to [00:57:30] be so, um, you know, so realistic in, in the situation.
[00:57:34] And, um, that’s basically the basically, uh, you know, the, the, the way, the way we, we handle this, this conversation. So understanding what makes a good picture. If you have been using Mid Journey, you can start seeing what other people are making with mid journey. Start to understand what makes it prompt work.
[00:57:55] The first 100 are going to be bad, the same as everything that [00:58:00] we are doing in creative life. Uh, yeah. But that’s the, that’s the, that’s the, the way we handle things. So we barely have time. I’m
[00:58:09] diane: trying to get it to do it, but I will put it underneath. Um, when you are in. Um, if you’re watching this on YouTube, it’ll be in the show notes that’ll be on the page.
[00:58:23] I am also having trouble, uh, Pacing it in. So I will make sure that it’s there. And if [00:58:30] anybody wants to text me right now, if you’re just really
[00:58:32] Dimi Lazarou: tate, just I think it’s a spam, it’s a spam safety measure from, from the Zoom, I think. Mm-hmm. I think so. Do we, do we have any questions?
[00:58:41] diane: So, um, side question. Tate said, any resources you would suggest to help learn prompts for Mid Journey?
[00:58:47] So, I think that’s a great question. Uh,
[00:58:50] Dimi Lazarou: well, mid Journey itself, so you ha you will be using Mid Journey and the application called Discord, but make a point of going to the [00:59:00] actual website of Mid Journey logging in, into your account. And when you log in into your account, you will see the feed of the most beautiful pictures generated right now in Mid Journey.
[00:59:12] And you will see what other people are making with it and try to be social on the Discord service about this. Like people talk about, about from generation. But, uh, the main, the main thing is, um, the, the main thing is instead of trying different [00:59:30] keywords, try adding more and more and more and more.
[00:59:32] There’s no shame in using 100 keywords. Uh, it, it only makes the thing more specific. So if, if you have, uh, and, and, and also just be, uh, you know, uh, knowing a lot of art history, knowing a lot of, of design history obviously helps because, again, some things are tagged so you can, you can get a sense of, of how these are, are, uh, used.[01:00:00]
[01:00:00] diane: So I just, I was able to add it as an image, as a, um, text document. Um, that was the only way. So hopefully, um, you guys can get, um, and I, you know, I noticed that there’s like some things that are not spelled. Correctly, who cares? It doesn’t, it didn’t, didn’t stop him. So we don’t have have to be so specific and so exact in our, um, just, um, in as we’re trying to create [01:00:30] when you were creating the picture of Sarah or the other guy who, I still can’t remember his name, Blake.
[01:00:35] Um, um, oh, so Josh, uh, Gooch said he has a LinkedIn contact that’s publishing some eBooks on this topic. So, um, maybe we can can go there and check that out. I super appreciate you sharing that. Uh, Josh, um, Michael, so to get that same sort of warm tone, it’s just, you’re just writing sort of what you [01:01:00] are wanting.
[01:01:00] Dimi Lazarou: So I’m, I’m, I’m asking for a graphic designer. From London. I, I give the same, the same information about the, uh, the camera. Like the camera will make it look very realistic. Uh, we are very lucky because coffee, um, um, when people are holding cups, the engine understands better how many fingers a person could have.
[01:01:25] This is why, this is why we saved it, uh, thankfully. [01:01:30] Uh, so, um, and then from from there on it, it was just a matter of us understanding, uh, understanding the person more and making sure that we had the right information. So it’s graphic designer, uh, lives in shortage in London. We gave them these words because, you know, there’s a very specific culture that that happens there.
[01:01:52] Uh, we hand, we hand put a few brands of specific coffee shops in this situation. So [01:02:00] that helped. Uh, it was just, uh, um, yeah, I, I, I know I covered a lot of things, uh, but I really hope it wasn’t too superficial,
[01:02:09] diane: Paul. No, no, I think this was great and I hope you guys, um, had fun watching this. I, I really liked that he was able to create the, uh, the look and feel without having some of the things that.
[01:02:26] That we would normally need. Um, I love [01:02:30] that. This is great for a fake project. Um, Tate says, this was awesome. Just signed up for Mid Journey. That’s awesome. I did have two, uh, links that are the chat, G p t and um, mid Journey. And then if you wanted to watch, uh, Demi’s last episode, he’s been on three times now.
[01:02:49] Um, I think just three, but they’re, those are all in the chat and they will be down below if you’re watching on YouTube. And I wanted you guys to know how you can get in touch with Dimi. So [01:03:00] Syon branding.uk.no.co uk. Co
[01:03:04] Dimi Lazarou: uk or at syon dot branding in Instagram, Twitter, uh, 1:00 AM So that’s the, that’s the way to reach me.
[01:03:13] I, uh, my, my final word of advice is, especially Mid Journey, is don’t be lazy. Don’t use it to replace yourself. Use this as a tool. That, you know, use it to do what’s Illustrator or Photoshop or [01:03:30] InDesign panel yet do for you? Don’t use it as a replacement. It will be gener generic. It’ll be boring. Use it to understand your clients better and just make sure that you give them uh, something excellent.
[01:03:43] Um, Uh, that’s my final way of
[01:03:46] diane: advice on this. Well, I also think it doesn’t replace you necessarily using an illustrator to create the, oh, no. An illustration. It, it is not gonna be as good as a person. But for something [01:04:00] like this, I think this is a great way to use it where you’re not kicking anybody out of a job, it’s just that you are enhancing what you are able to do.
[01:04:12] Or, um, helping a client not be so limited because they just didn’t have the right thoughts or they were having, um, some anxiety during the workshop.
[01:04:23] Dimi Lazarou: But, can I, can I interrupt? Please. It’s very important. I, I don’t want to miss this. So this is, [01:04:30] um, for, for everybody who is actually a designer and they have tried working on a laptop, they will understand, they will understand this.
[01:04:38] This is going to be a game changer for people with limited mobility, uh, because, um, manual skills are still very, very relevant. Whoever has tried replacing their usual setup with, uh, a laptop’s tracked, will understand it. Understand it. This is just a new interface that we’re going to be using and people who [01:05:00] can only use language or can only type, uh, or can only, you know, can’t move their hands in the way we do.
[01:05:07] This is super, super, super helpful for them. It’s just a change of interface. It’s really, really, uh,
[01:05:14] diane: I wanna ask you this, um, before we go, what is the biggest hurdle you faced as you were learning? Um, is, was it chat g p t harder or was mid journey?
[01:05:26] Dimi Lazarou: Um, mid journey, uh, image generation in the [01:05:30] beginning was really slow.
[01:05:31] Um, and it needed, uh, a lot of, a lot of effort to make pictures that just weren’t cutting the mustard in, in any case, like, oh, I’ll just put it back again. Uh, Chad Gvt is different. Like, look at this, thing’s just terrible.
[01:05:49] diane: Yes,
[01:05:51] Dimi Lazarou: uh, uh, t was, uh, game changer in the beginning. I think the, uh, what is problematic right now is the [01:06:00] scale where it grows.
[01:06:01] Like when we have adoption spikes, when more more people get to, uh, get to, um, to sign up, um, in, in big spikes, it gets dumber in some way. Just doesn’t have that, the amount of memory it, it, uh, normally has. Uh, so it really doesn’t understand what you’re talking about in certain situations. Uh, but no, the, they have been, um, the, the hurdles are more [01:06:30] about, um, time now than anything else.
[01:06:33] Like, they’re really, really easy to use.
[01:06:36] diane: Okay. So it, they’re really easy to use if you’re good. Explainer, if you can explain. Well, and I think Tate said this and I saw Josh has a question. I’m gonna make sure I get that. Um, Tate says, it’s amazing as a designer that doesn’t draw well, but can explain what I need.
[01:06:52] And this is, I think one of the things that is really, um, that’s really great. Um, Josh asks, [01:07:00] is this case study the one, the red planet available anywhere to review more in depth? And I’m, is it on your website
[01:07:07] Dimi Lazarou: yet? It’s, it’s not yet. So you are the first people who’ve seen this. We’re going to be releasing this, uh, from Monday.
[01:07:15] Uh, it’s going to live in our website. Uh, eventually it’s going to have its own website, uh, so that people can, uh, join. Because for us, this is, uh, this is something that is going to evolve. So what we’re thinking right [01:07:30] now is, uh, Every service that we are going to include, like, uh, naming a brand or, uh, you know, we can use Red Planet as a client.
[01:07:39] And every idea we have about packaging ev everything that we want to demonstrate, uh, it can live within the same ecosystem in, in the situation. Uh, so yeah,
[01:07:50] diane: so this is a great, um, thing. So, um, I have a alumni who had started illustrating, I don’t know, back in [01:08:00] 2014, I think, and he illustrated this robot and it was pretty rough.
[01:08:03] And then every year on that same day that he posted this robot, he illustrates a new robot. And he’s a much better illustrator now. Yeah. But it would be interesting if you took that, the, the first image that you did, the Greek simple, and you use, now you use your experience based because you’ve practiced.
[01:08:21] You now know how to tell it what you want. Um, Matt says, it would be neat to see how this image would evolve based on Demi’s [01:08:30] experience with the prompts. And I do. So you can,
[01:08:32] Dimi Lazarou: you can already see this on my Instagram account, my personal Instagram. I put it, I put it on the, on the, on the chat. Yeah. So you, if you go.
[01:08:42] If you like the, this, I’m making triplets on Instagram and you can see how the, my style has evolved as time passes instead of this, uh, so people can, uh, can have a look like this is, uh, you know, you can scroll in as just, you [01:09:00] know, very, very different as time goes by. And the, the things that are, you already see here, they’re four months old.
[01:09:07] Like, um, uh, it’s, it’s getting even better and the technology’s getting better and I’m getting better in, in asking, uh, the relevant, uh, prompts. So
[01:09:17] diane: yeah, I love that. I hope that this was interesting for y’all. This was interesting for me for sure. Um, Alexia says, personally I like the openness of mid Journey site.
[01:09:28] Everyone’s tries are there to [01:09:30] see. And I also think that that’s interesting as well because we do, um, it’s better when we don’t think that everything comes out perfect. And when we see other people struggling as well, it makes us not feel so alone or like, we’re never gonna get it. Ha. Right. So just using simple for maybe you’re writing an email and you need a, you want a better, um, uh, subject line.
[01:09:58] Dimi Lazarou: When I, when I was in [01:10:00] design school, uh, um, one of the professors said, like, the point is to have a stack of papers in your portfolio, which is as, as tall as you like. If you don’t develop that, you’re, you’re not
[01:10:13] diane: there. Yeah. Well, um, next week I am excited to be back and I’ll be in this same room here in Georgia, so, Um, I am just blanking on who is next week, but Oh, no, I know who it is.
[01:10:29] [01:10:30] It’s, uh, Nathan Yoder, who I think I had him on in 2014, maybe even 2013. So very, very beginning. He, um, had been on a few times, so I’m excited to have him back on. Um, and just thank you guys for tuning in and definitely thanks for staying up late, Demi. I appreciate it greatly. And Of course, of course. Uh, and so does Tate.
[01:10:59] Dimi Lazarou: [01:11:00] Thank you everyone for your attention and your time. Uh, it’s always a pleasure being here, Diane. I’m very, very happy to be here again and yeah, do, do, have you had people come, come over more than three times?
[01:11:14] diane: Yes. Well, Mari’s been on a whole bunch.
[01:11:16] Dimi Lazarou: Um, uh, I need, I need to break this record of nothing and
[01:11:19] diane: see.
[01:11:20] Okay. Well we will, we’ll have to work on that. You just keep coming up with great presentations and I am, I’m just, [01:11:30] I’m glad to see people, Laura, haven’t seen you in a while, so it’s really nice. Um, and says you’re in Chicago. So I was right at my guessing. Um, I, I’m, it’s just cool to see this. It’s cool to see.
[01:11:43] And for me, I started with chat, G P T I, our painter at, uh, where I teach at the University of South Alabama. He uses Mid Journey and he’s a traditional oil painter. So if you’re thinking, oh, you know, like this is not, he, he’s embracing it and now he can have [01:12:00] things, um, visual things to draw from that are not copyrighted.
[01:12:04] So that’s also. For me as an illustrator, if I’m trying to figure something out or a robot with a really cool, um, maybe panel of buttons and levers that I can actually get mid journey to do it for me and I’m not taking somebody else’s image. And to me that’s a, I’m still gonna draw it myself, but I want to draw, I wanna see [01:12:30] what it would look like as realistic as I can.
[01:12:33] And I think that’s another, a good way for us to use it as a tool. It’s a tool to me. It’s a tool like, A, uh, like my power drill or my power screwdriver. I do not wanna build a house with a nail and a hammer. So that’s, that’s what I’m thinking. And I like that Dimmi is very clear. Uh, he’s not always so jumping on every bandwagon, but he jumped on this [01:13:00] and I saw him really grow, and then I saw how I could use it, and I thought, Hey, I bet other people might be apprehensive.
[01:13:08] Maybe this would be a good thing to, because sometimes, you know, we’re, we’re working alone. We don’t always get to talk about this stuff. We see it happening, but we don’t really have time to play. So which things are the best, um, um, uses of our time that could actually make a big bang. So anyway, I hope, I think you guys really [01:13:30] liked it, so I’m really glad that, um, Dimi could come and share, and I think the Red planet stuff is just really cool.
[01:13:37] And I will see you guys next week with Nathan Yoder. You wanna say any last bit, Demi?
[01:13:43] Dimi Lazarou: No, thank you everyone. Um, uh, I’m, I’m always happy
[01:13:46] diane: to see you. It’s always good to see you, and all the links will be down below if you’re watching on YouTube or you’re listening on SoundCloud there right below.
[01:13:53] Obviously, if you’re here live, you know that they’re in the chat. You can always come live. It’s free. You just have to give your email. [01:14:00] Bye y’all.
[01:14:01] Dimi Lazarou: Bye-bye.