Ep 12: How to push personal creativity with AI

Episode Categories:

May 10, 2023


Jordan [00:00:03]:

Will human art exist in the future without AI? That's just one of the things that we're going to talk about today on Everyday AI. My name is Jordan Wilson and I am the host of Everyday AI Your Daily Livestream podcast and newsletter that goes over the latest in AI news, trends and everything else so you can understand what's going on in the world of AI and actually use it. Very excited today to have Rachel Moore with us, a contract research contract researcher at Brand of a leader. Rachel, thank you for joining us.

Rachel [00:00:38]:

Hi, thank you so much for having me. Looking forward to having a little chat today.

OpenAI implements copyright system with new GPT model

Jordan [00:00:43]:

All right, so, yeah, we are definitely going to be taking a deep dive into using AI for personal creativity. But before we do that, want to talk about what's going on in the world of AI news, so we're going to get to it. So, kind of the first big development of the day is the OpenAI CEO. So OpenAI is the company that obviously owns Chat GPT. So Sam Altman, their CEO, is going on a world tour of the future of artificial intelligence. So he said some pretty interesting things on the first leg of this tour recently. So saying that OpenAI is working on a copyright system that ensures creators are getting paid and also even working on a new GPT model that respects copyright, which should be interesting. Rachel, what's your thought on this new kind of lend toward copyright with these models?

Rachel [00:01:34]:

I think it's really important, honestly, a huge criticism, I think within the AI community are people who look at AI say like, well, it's disrespecting people who have already created content and all those things. So I think it is an important step forward for sort of AI leaders to be taking these kinds of things into consideration if they want them to continue and be successful and all that.

Google's Integration: What's Next for Search?

Jordan [00:01:56]:

Sure, yeah, extremely murky. If you're an AI lawyer, it's going to be a nice couple of years for you. Second big event, and this one is very timely. So in a couple of hours, Google is going to have their I O event, their yearly announcement where they talk about all things hardware and software. So Google's been obviously one of the leaders in the AI development space, kind of hand in hand, I think, with Microsoft. So they're going to be talking about the very highly technical advancements to their large language models and everything that makes their AI go. But two things I'm really looking at is what improvements they're going to be making to Bard, which is their equivalent to Chat GPT that didn't exactly get off to a hot start. And then also how they're going to be integrating AI into their normal Google search. Rachel, what are you going to be looking at today from Google?

Rachel [00:02:49]:

Yeah, I'm really interested in seeing sort of how Google is going to integrate as you said into search. But then also I'd love to hear if they're going to be doing stuff with the whole Google suite. Like, are we going to have integration with Docs, are we going to have integration with presentations, things like that.

OpenAI's New Shape Rendering Model

Jordan [00:03:05]:

I'm ready for it. Personally. It's something I saw Microsoft doing. I'm like. Should I be considering Microsoft Speaking? Going back to OpenAI. So they just over over the weekend announced a new model called Shape E. Essentially OpenAI. They have their text generation, they have their image generation in Dal E, and now they have Shape E essentially taking text prompts or even uploading 2D images and getting 3D renders. Wow. The possibilities are endless. Rachel, as a creative, what are your thoughts on that?

Rachel [00:03:41]:

I think it's very exciting, especially in the world of design and crafting. I could definitely see huge amounts of applications for this. It's almost endless in my mind. So super exciting. Interested to see how well it does, if it's going to be something that kind of exceeds expectations or does not. So that's going to be interesting, definitely, yeah.

AI Image Generation Sparks Creativity in Writing

Jordan [00:04:05]:

Rachel, I think that's a perfect segue into talking a little bit about your journey. So there are these new AI platforms and just capabilities that are allowing people, whether it's creators, to be even more creative or people who generally aren't very creative at all, to really dip their toe in the space. So I'd love to hear, just with your personal experience, how have you used these new technologies to really push your creativity?

Rachel [00:04:35]:

Yeah, of course, from the beginning, when image generation AI was sort of becoming popular, I wanted to use it as a tool immediately. I'm somebody who has always been a little bit more than a little bit, I think, creative. I like to draw, I love music and I love writing. I really thought it would be an interesting sort of thing just to play with. And as I was playing with it more and more, I realized that it was really great for taking something that was in my imagination that I was not able to put out onto a page or draw myself and sort of be able to create that at a very high quality and give myself inspiration from my original ideas.

So the use case that I use it most predominantly in is my own creative work, which currently I'm writing a fantasy novel and I've been using it to generate images of my characters and my settings and some of the creatures that inhabit the world that I struggle with physically drawing. And instead I'm able to make these very photorealistic versions of people that don't exist, but exist in my head. And I'm able to sort of use those images as reference for when I'm writing. And that's been really exciting for me.

Transforming Sketches with AI

Jordan [00:06:02]:

Yeah, let's actually get right into that because I think it's a super creative process that you took here. So if you're joining us live on the show. This should be really easy for you to follow along and kind of as Rachel talks us through her process. But if you're listening on the podcast so what I'm going to do right now is Rachel's kind of shared a little bit of her artwork that's now on the screen and just talking about her process. So I'll first paint a big picture, but then, Rachel, I'll let you actually explain it.

So what we have here is we have a series of drawings where for her novel, rachel starts with a sketch, and then she's taking it into kind of a sketching and digital phase, and then taking it into photoshop. And ultimately taking this into the stable diffusion technology, which is kind of a text to image, but also allows you to take your images up a notch. So we're going to just quickly go through your process, Rachel, before we even start going through this kind of step by step, what did this process look like before this technology? Did this novel not really exist? Or was it just an idea that maybe never would have come to fruition if not for this technology?

Rachel [00:07:24]:

Yeah, so I definitely was sort of working on the novel for a long while. I had done little sketches of the characters, but nothing was really solidifying for me because as a lot of artists might experience, I have what is called same face syndrome when you're drawing. So all of my drawings of people often look really similar because their facial structure is kind of something that I don't deviate a lot in my own drawing. So I would draw these sketches of characters, they would all look the same, and it was kind of boring. And that was just where my artistic ability was at.

However, I was able to use Stable Diffusion here and take some of my sketches and turn them into photorealistic versions of people. And I was able to use prompts to sort of tweak where proportions were, use a lot of different even. Sometimes I would use a little bit of celebrity name tagging to tweak different structures in faces. And from there on, it actually made my characters really unique and interesting physically. And I was actually able to realize some of my characters, even though I just thought it was my drawing looked the same, they actually looked the same in their description. So it allowed me to go back and modify them, make them more unique, sort of push boundaries on physical appearance, definitely. So, yeah, that's sort of where we are now.

Creating Photorealistic Character Art: The Process Unveiled

Jordan [00:08:53]:

Awesome. I'd love for you to just quickly walk us through this process. So if you're listening on the podcast, we're going to go from the original sketch hand drawn, and we're going to go through taking it into Photoshop and then all the way to the end. The end result is a highly detailed, vibrant colors. It looks like something that you'd see in an actual movie. Almost like something that's been gone over by a team with CGI and all these other things. But Rachel quickly take us from the process, going from sketch to this finished product.

Rachel [00:09:31]:

Yeah, definitely. So, as I said, I've sketched out my characters quite often. This is a character called Artura who I've been working on for a very long time. So I start with an original pencil sketch, and then I go into digital painting. I just use Critta, which is an online painting tool, and I paint up a couple of renditions. I then took those images and used Stable Diffusions image to image sorry, Stable Diffusions image to image. And I asked sort of make this into a photorealistic version of this person. And it was a lot of iterations. It took a very long time to sort of get one that I actually liked the look of. Stable Diffusion doesn't really like it when you're trying to give it specifics. Like, this is the hair color and this is the eye color, and this is for this character in particular. She's slightly heavier set. Stable Diffusion doesn't like to do that. So it took me a lot of work to get it to sort of give me exactly what I wanted to.

Once I got something close, I took it into Photoshop myself and I adjusted things. So, again, Stable Diffusion sometimes does weird things with pupils or it'll add strange accessories or strange colors and things like that. So I just moved it into Photoshop, fixed up some stuff, and I had the sort of headshot, which was really exciting to me. I just kind of kept that as reference. And then I said, I want to do a pose with this character. So I took that, then again into my digital drawing and drew a really sketchy picture of a body, like a torso for this person. It is not. It was about 10 seconds of me drawing this, and I had a pose in mind. I wanted her hand up, and I wanted to be holding this orb. And I put all this glowy yellow around to try and get a magical looking thing. And then I again took that back into image to image and had it sort of generate all over again. And this time it was changing the face and it was changing the hair. And that was okay for me because I knew I could always go back into Photoshop and fix those things up.

But, yeah, I had a bunch of iterations. Again, took a long while to sort of get the magical orb looking right, to get the body proportions looking right. But eventually, once I had something I was happy enough with, I took it into Photoshop. I fixed up the hands, because you always have to fix up the hands and see what you can. And it fixed up some of the sort of proportions and colored the orb and things like that. It did a little bit of tweaking and from there I went from just a pencil sketch that I'd had for probably a year that wasn't really going anywhere to, I think, a dynamic image that has something interesting and magical about it.

Digital Art: A Stepping Stone to Elevate Creativity

Jordan [00:12:24]:

Yeah, absolutely. So thanks for walking us through that. I'd love to ask, kind of in the same way that this new technology really opened up a whole new world of creativity for you, what do you see? Kind of maybe if we use indie creator or individual creators, how do you see this space changing in the coming months and years in terms of what people are able to produce?

Rachel [00:12:52]:

Yeah, that's a really great question. I think, as I said, I've sort of always viewed it as a tool. A lot of people are resistant to new technology when they come out. I think when photography came out, people would have criticized it, saying, well, that's not real art. When digital painting came out, people were saying, that's not real art. I think what we're going to see over the next few months and years is artists beginning to actually embrace the technology. So ones who are extremely talented now hopefully will embrace the technology and be able to use it for inspiration, for foundations. I see it as a way to sort of get a stepping stone into something that is much more elevated. So where I had a simple kind of sketch and digital painting, I was able to then take that into something that was photorealistic and I could see a lot of people be able to use that to sort of push themselves in different creative directions. I could see them using it as references for different art styles. Yeah, I don't know. I think it will help elevate people who are a little bit stuck in their creative process is how I see it most useful?

Embracing AI in Art with an Open Mind

Jordan [00:14:06]:

Sure. Given your experience through this. So, again, a sketch in a project that you had been working on for a long time, maybe not needed, but you really used and leveraged this technology to get it to a finished state. So what is your advice to other people that maybe need a creative boost and haven't really used AI much? Or maybe they haven't even maybe they're not people who sketch or people who spend time in Photoshop. What's your advice to people to actually go in there and use some of these new tools and techniques to push their creativity?

Rachel [00:14:47]:

Yeah, I think the heart of it is that the technology is super new and I think it's something that can be really helpful to you to get in on early. Something that I've said to some of my friends who are a little bit resistant to using this technology or say like, oh, I'm not really good at computers. I don't really know how to do that is I think the world of art is changing and I think most industries are going to be impacted by AI and not a lot of people in the world have been successful by being resistant to change, unless maybe you count, like, politicians. But I think it's important for people to sort of dive into how the world is changing and how creative processes are changing and being open to that.

So my advice is sort of explore it in a way that makes you happy. If you're enjoying exploring AI through image generation, just mess around with it. Maybe you want to make something like a funny little image for your friends and you don't know how to use Photoshop, so how about you just go in and start there? Maybe you want to make really elaborate pieces of art and you want to build up your skills that way. Then, of course, start there. I think it's just starting where you're going to have fun with the technology first and then moving forward without any higher expectations of how it's going to turn out.


Jordan [00:16:24]:

Yeah, that's great advice because, yeah, you never know how it's going to turn out. So, Rachel, thank you again. We're coming up to the end of the show, so we will have a lot more information that Rachel talked about in our daily newsletter, as well as some practical tips and examples of some of the tools and techniques that she was talking about. So make sure to go to your everya.com sign up for our newsletter. And also we kind of usually mention Chat GPT on the show, so we're actually giving away two premium year long memberships to Chat GPT, so sign up for the newsletter, we'll send you that and we'll send you a lot more information and kind of next steps playing off of Rachel's Conversation Nation. Rachel, thank you again for joining us.

Rachel [00:17:07]:

Thank you for having me. It was great.

Jordan [00:17:09]:

All right, so thank you, everyone. We hope that you learned something and you can actually take it and leverage it. Thank you for joining us and we hope to see you tomorrow and every day on everyday AI. Thanks.

Gain Extra Insights With Our Newsletter

Sign up for our newsletter to get more in-depth content on AI