Lightly edited from the recording for readability. Wording is the speakers’ own.
Hello, everybody. Welcome to the very first AIGA New York and FLORA Lab with Catherine Chung. I'm Olivia, leading education here at FLORA. We are so excited to have everybody here tonight from all over the world to get started in the FLORA Canvas. And before we dive in, I wanted to say a huge thank you to AIGA New York, especially to Stacy and Justin over at the New York Chapter, who have really believed in this partnership with FLORA from our very first conversation. AIGA New York has really been home for design in New York City for decades. And we really do share the same goal to ensure that designers stay at the center of how creative work gets made as tools change.
And all of that, we believe, starts with access and education. So that is why we are doing these next six labs over the next six weeks. Every Tuesday evening from 6 to 8 p.m. Eastern, we will get hands on with a different part of the creative process led by Catherine and Sasha here at FLORA. All right. With that, over to you, Catherine. Hi, Olivia.
Hi again, guys. Good to see everyone. I'm Catherine. I'm a forward deployed creative here at FLORA. And I come from about three and a half years of experience over at Pentagram as a brand designer.
But today, first of all, thank you for joining us. I'm going to be walking you guys through Canvas workflows for logo design and expressive typography. So without further ado. I'm going to talk a little bit about the partnership with AIGA New York. As Olivia said, we're super excited about this partnership. And we're super thankful and just kind of really happy to kind of have everyone in this room together.
But a little bit about our partnership, FLORA is proud to partner with AIGA New York and provide designers across the country with free resources to feel ready for what's next in their field. So small agenda for today. First, going to start off with a little bit of introduction, a little bit of about us. And just in case you haven't like this is your first time interacting with FLORA. Or AI as a creative technology. Second is why nodes and preparing to use them. Third is going to be your first generation and then building a real workflow on the canvas.
So a little bit about us. FLORA is a node based creative tool that brings the top creative AI models across text, image, video, audio and 3D into one intuitive interface. To accelerate creative production without compromising on craft. And FLORA has a few entry points. We have canvas, which is a node based infinite style canvas that allows you to visually program repeatable workflows, freely control how media interact or generate and allows for a more hands on experience. MCP uses a third party LLM application to use FLORA to execute workflows and generate images with existing context already built out. Create slash generate is a chat interface that allows you to generate images using any model that we have on the platform.
And then we have a few more that's coming down the pipeline. So stick around for that. Today, I'm mainly going to be focusing on canvas. I will kind of like speak to the MCP and eight more agentic kind of workflows kind of like naturally but. Yeah, understanding the basic function of nodes on our canvas helps you get the most out of the tool when you explore agentic workflows later on.
So. A few things to understand about AI and typography because I want to make sure that I'm setting some expectations here. I'm not trying to convince you guys you can like typeset beautiful paragraphs with like perfect court kerning. Within FLORA today's definitely more on the expressive side.
So just like some housekeeping things to kind of keep in mind. Most AI image models are pixel based. So if you generate a graphic or a letter, it won't be pixel perfect. And that's okay. The methods that I showed today are most valuable to help visually explore different ideas faster for typographic or logo directions.
So a really quick example here on the bottom. Is a kind of typographic inspiration image I grabbed from a vintage Brazilian poster. I really liked it. I kind of used this example in a graphic design webinar I did a while ago. And what I did here was through a workflow, I created a word mark for the words Cerny based off of those letter forms. However, if you look at this, it looks okay, but if you start to kind of dissect it a little bit, you start to notice the nuances that are obviously not kind of giveaways to tell me that this was a little bit more of like an AI generated image, or at least something that wasn't as refined or kind of pixel perfect when it was being made.
So for example, the diagonals on the ends kind of don't match up with a diagonal in the R. The width of the E is different in both of the E's in the word mark. So small little details like that kind of make a difference between like something that's more AI generated or something that's a little bit more pixel perfect. However, I was able to kind of get a quick result and see whether this typographic exploration was worth kind of going down. And to me, it was with a few tweaks, I was able to kind of perfect it from there, which brings me to my next point. If you like a visual exploration and you want to manually edit it, you can regenerate as an SVG. When AI first came out or first came into the picture, SVG models were not as popular or powerful.
However, you can generate any graphic image as a near perfect SVG file ready to take into a vector based program. So a good example of that is this. So this was AI generated with GPT Image. It's just the word daily. I think I generated this to kind of like mimic a specific like digital clock type of type treatment.
But I was kind of ready to kind of go in there and get a little mess and like start to mess around with some anger points. So I went ahead and downloaded it as an SVG, which is something that's possible within FLORA. When you kind of generate an image, when you go to the top toolbar, you can click download and you can select your file format. Of course, like don't download a photograph as an SVG. It probably won't give you the best results, but in this case, since it generated me like a 2D kind of graphic black on white, I was able to just download it as an SVG. However, as you can see, those anchor points are like all over the place. There are more anchor points that need than need to be there.
If I had done this by hand, they'd be way more cleaned up. They would have like rounded corners, like things would be adjustable, all that sort of stuff. So going this direction is probably not the best kind of route, but here's an example where I downloaded the same graphic, but I ran it through an SVG AI model and then downloaded it as an SVG. The particular model that I used here, I believe it's a model by Arrow AI, and I'll be kind of going over that later on as well.
But what I can do here is kind of like open this up in Illustrator and start to finagle with the things like rounded corners, links and widths of specific parts of the letter, all that sort of stuff. So if I compare it like the anchor points between these two, it's like day and night. So another thing, AI can be very expressive, which is great for expressive typography design. Today's class really tries to take advantage of AI's ability to think outside of the box. While pixel perfect explorations are better done by hand, more expressive explorations can be expedited with AI and perfected later.
So here's an example where I took a, the image on the left started off as a sketch on paper, took a photo of it, brought it into FLORA and just rendered it out as a graphic. And then I just kind of put it into an image node and asked it for a variation sheet to kind of explore things like geometry, symmetry, perspective, all that sort of stuff. And what, like not going to say here and say like every single one of them are bangers, but just to be able to kind of visualize what it might look like if, what if those bottom feet were full circles instead of like half circles? And what, I never even thought of like maybe having a circle at the top and like kind of having that relationship of like half and like full circles, but like maybe the whole thing is made out of circles, like all that sort of stuff.
So it's really nice to just kind of have that opportunity to riff off of just quick visual explorations because you never know when a mistake might be a new direction. And then last, if you want pixel perfect graphic explorations, then agentic workflows may be the option for you. So you can use an agent, such as Claude or ChachiPT, to execute a series of explorations using a vector-based program. And I'll kind of touch on this later in class and show you an example of this.
But yeah, today mostly we're going to kind of be in the canvas generating some stuff. And this is kind of an example of that. So the board on the right was created in Illustrator by Claude. So I went into Claude and I said, here is a serif A, and here is a sans serif A. I want you to explore different ways that these geometries can interact to create a new letter form. And I gave it some pointers. I said, I want explorations of serifs adoption, symmetry, thick and thin contrast like studies, all that sort of stuff.
I want 50 examples. And so I was able to kind of ask Claude to do that, execute it as an SVG in Illustrator. And then I was able to visualize this board on the right. So from here, I can kind of decide. I think that I do actually like the idea of like just the serif letter form without the feet.
But I definitely know I don't like when the apex vertex kind of like butts up to the top of the A like that. It just makes it look a little bit too like. I just can't find the word for it just doesn't look right to me. So it's nice to be able to just kind of quickly visualize those things and know that if I wanted to kind of tweak any of that, it was already built in a vector based program that I can start to edit in.
So this is an example of something I made agentically, where all the outputs are already pixel created and ready to kind of be manually edited. I really like that one. Actually, just seeing that now it's kind of nice. But most importantly, I really want to make this point. During today's class, I do. I am providing you guys with a few things on the board that you can start to kind of experiment with. However, if you have your own images, if you have your own inspiration images, if you have your own like projects or typographic studies that you kind of want to mess around with, you're more than welcome to do that.
So I would like to kind of bring that into a board and start, kind of practicing the things that we're showing in regards to that. However, I have to say this and I wanted to make sure that I put this in. If using a font from a type foundry, always check their licensing policy on editing letter forms. Some licenses cover manipulation or customization of letter forms for logo usage. If you have letter forms from an existing font with the intention of making a new font, you may want to check with the font provider. If you can try starting from a design that you made by hand or starting with an open source font, you'd be surprised of how far it can take you.
So really want to make sure that I communicate that to you guys. Please support your type foundries, purchase fonts whenever you can. They're truly a gift. It's an art form that is so like when you're obsessed with typography, it's such a beautiful thing.
So if you can definitely kind of, purchase a weight of a font that you really love, kind of use it as inspiration. But please, read those licenses. If anything is unclear to you about a license, reach out to the type foundry and ask. That's the biggest piece of advice I can give during this workshop.
But you'll see, I gave you guys more than enough to kind of mess around with. But you'd be surprised, download a Google font or download like a, Trojan horse or something like that on your desktop and like start to add syrups to it. You never know. So, OK, great.
Now I got that out of the way. Why nodes? So if you've kind of like been here a little while, you may kind of like understand where I'm getting at. But just bear with me here. I do have a little bit of an exercise at the end of this that I think will be kind of fun.
But so why nodes? It's a natural progression of creative process and design thinking. So we're we've all kind of set the following line at some point. Sorry, my work file is so messy. And why is that? It's because when we are in a creative flow state, we're spitting ideas out left and right and coming up with new visuals faster than our hands and tools can create. And seeing our process from start to finish is how we prove our ideas.
To the people that matter. So we're all used to kind of just like putting things on a wall and seeing what sticks. That's all kind of, it just kind of naturally translates to a canvas where you can start to connect those things together and generate new things out of them. They allow for repeatable workflows and visual systems as well.
So making great visuals is one thing, but to systemize that approach and make it easy to repeat or pivot for a project or a campaign is priceless. And another thing is that chat interfaces can only get you so far. Some of our first interactions with AI have been through a chat interface, but us as creatives have quickly outgrown that box and craved a more freeform way of making beautiful work without having to scroll up endlessly through iterations and finding that right version while our director stands behind you or something. Okay. I'm going to pause here. There are two versions of this because I do this exercise all the time during my webinars and my immersion programs, but I'm going to start off here and ask you guys what a blue square and a green circle equals. And I'm going to be looking at the chat to see what people's answers are.
And I'm going to be reading a few out loud. Okay. Thanks, Olivia. First one is yellow triangle.
Next one is an oval, a scorcel. I've heard that one before a teal school bowl, geo Earth green circle. Yeah. Green square.
Okay. I love how I always love kind of asking this question because everyone always has like some really unique kind of answer to this. An island in the sea. Oh my God, you're such a designer.
Okay. All those answers are right. You're kind of determining or kind of like trying to understand how two things kind of interact. Now I have a different version of this question. If you've seen me ask this question before, I want you to think about this also. What does a serif a and a sans serif a equal, since this whole thing is inter, okay. If this whole thing is about typography, I want to put this in context to typography, a slab serif.
Very good. Okay. We're getting names of serif of fonts now, modern serif. Okay. Yeah.
This one's a little trickier. Like you got to kind of think about it semi serif. Yeah. Exactly.
So we're AI serif. We were all kind of like thinking about how these two things can interact. So whether it's a circle in a square or a serif a and a sans serif a, we can, the world is your oyster in this sense. But if I had to kind of put it in context to how nodes work, it looks like this.
So when you have two things, and I want to preface this, like not all workflows start off by combining two things. It could be by combining five things. It could be by, taking one thing and turning it into five other things. But this I have found usually kind of translates the overall concept of nodes pretty well.
So this is like structurally how this type of connection would look like when you get into a node based program, which we will be doing later on. But just bear with me while I kind of what your appetites here. So on the top, we have the serif a and on the bottom we have a sans serif a and what these nodes look, what these nodes look like in the canvas is you are able to kind of extend out a line out of each of the nodes and connect them into one other node as seen here, or maybe three other nodes or maybe four other nodes, et cetera. So the point is here is that those lines are the key. When you have an idea and you have an interaction that you want to happen between two things or however many things, that is kind of visually what that idea and what that process looks like on the canvas, you're connecting two things to make something new.
So maybe a serif a plus a, sans serif a you, the way that you want them to interact, maybe to apply the serifs of the serif a to the sans serif a, or maybe you want to remove the serifs from the serif a and keep it stick and thin strokes, or maybe you want a sans a leg on the left and a serif a leg on the right. But this is all to say that you dictate the interaction between whatever you put on the canvas. So AI will give you a bunch of different kind of Depending on what you prompt it, depending on kind of, how you prompt it, what the project is, all that sort of stuff. But what will really get you what you want is you determining what that interaction is.
So let's say that, we combine the serif A and the sans serif A to get a serif or like a brand new kind of like serif A dependent on the sans serif one. So what you can do from there is once you have one output, you're able to then kind of build off of that. So maybe if you like that one letter form, you're now kind of experiencing what if I gave it a pointed apex on serif A? What if I kind of applied the thick and thin to the letter form? What if I put slabs on it instead?
So what's nice here is that where you started off, we're combining two images, you can then start to kind of branch off into different explorations. And then if you look back at it, you can kind of see almost like a track record, kind of where that idea originally stemmed from. So then from there, you can kind of continue building out. So for example, here, you can kind of extend that A into another node and just ask for a sharper apex again.
But let's say you want to start to experiment with it, but you're kind of just I don't quite know how to put it into words. I don't know how to prompt how to make a funky A out of this reference image. So this is kind of where you can start to use text nodes to help you write prompts. So what's going on here is that I have this A plugged into a text node. Metaphorically, I'll show you what this actually looks like in the canvas in a bit.
But in this text note, you can ask a, something like cloud or chat hey, write me a prompt that would add a little funk to this A. And it'll write you a prompt. And what you can then do is connect that text node into a new image node, plus your reference and get that same A with a little bit of a funky kind of like leg to it or something like that. So another thing I'm going to kind of be going over once we get into the canvas is using text nodes to help you write these prompts.
So let me show you what this looks like in the canvas first. And then after a little bit after I showed you kind of like where things are, how things work and all that stuff, we're going to send out a link to have people start to join the workspace and start kind of creating things for yourselves. But for now, let me just kind of show you around because since there are some people that have not used FLORA before. So just want to make sure everyone kind of knows where everything is.
So I go over here. After we send out the invitation link later, this is kind of what you should be greeted with an empty kind of project file or empty kind of project folder. And once this is open, this is FLORA. So this is an infinite style canvas. I'm sure you guys are very familiar with that sort of concept.
But as I zoom in here, you can see a bunch of different workflows. I kind of sectioned this out. So I'm going to give you a quick tour of what we're going to be going over and what I'm going to be showing you. And then I'll actually show you like where the nodes are, how they connect, all that sort of stuff.
So. We're going to be going over going from sketch to graphic. So turning a sketch into an actual graphic that you can kind of download as an SVG. We'll be going over combining forms to create something new.
So what that means is kind of the demonstration that I had shown you in the presentation. But what that actually looks like on the canvas. We'll be going over making variation sheets. So taking one letter form and kind of creating a sheet out of that to explore different variations of it. From reference to specimen.
So let's say you have a letter form. This is a better example, probably. Let's say you have a letter form here. You kind of want to isolate the letter forms and then start to create an alphabet or a specimen like this. Another example here is creating abstract letter forms based off of reference images.
So let's say you have an image like this and you want to start making letters out of that particular image or material. Combining with batch to kind of, explore things at scale. So turning my god eight turning 16 explorations into 64 explorations. And then others.
This is kind of a. How to kind of build your own kind of creative tools, custom creative tools in FLORA. And I have a typographic example of that here. So as I mentioned, this is an infinite style canvas. I'm not sure if you guys saw my colleague Johnny here, but here he is. It is collaborative.
So everyone can kind of be in this canvas at the same time. And if I kind of go up at the top right here, there is a project chat. So if you're kind of working with a team and you're all on the same canvas and you need to be Hey, Johnny, can you look at this? This typeface or something like that, I'll be sure, it's great. And he'll be able to see that in the canvas chat as well. Another thing here that I'll be kind of starting off with is on the left, you'll have this little toolbar.
So if I actually hover over at the top left corner, this is where your nodes are. So if I click that, you're greeted with this menu. So you have a text node, image node, video, audio 3D. I'm going to be mainly focusing on text, image, and I'm going to be touching on 3D today as well.
So let's say you kind of pop up an image node. If I click that, then this is where kind of that image node pops up on the canvas and a few things here. Let me zoom in here. On the left side is telling you what is allowed to be inputted into this node.
So since this node is kind of set to a particular AI model, which is defined up here. You can see here that this model is selected nano banana pro. I'm able to kind of type, I'm able to plug in a text node, which is indicated here through prompt or type in a prompt here. And you're also able to kind of plug in up to 14 images here. However, if I set this to something like Korea, then you'll see that kind of changes. I'm only able to plug in a text node with a prompt.
So if you notice that you're nothing, plug anything into your image mode, into your image node, just maybe check the model. And I'll be kind of going over models as well, just so kind of everyone gets on the same page. But my biggest piece of advice today, stick to GPT Image 2.5 and ideogram. Those are the two really good AI models that do really well with more graphic based generations. You'll see me kind of mess around a little bit with some other ones and I'll kind of, I'll flag that to you guys as well.
But yeah. I'm just going to go ahead and set this to GPT. So again, left side is going to be telling you what can be inputted into that node. And up at the top, you can select your model. Again, just for the sake of demonstration today, stick to GPT and ideogram. On the right there, you can also select the dimension or the aspect ratio of that node, which can be set here. You have a few tool things here.
I would ignore that for now and for today. And then you have a few other things, but you can also download your image node as well. So everything that I just showed you for the image node is also applicable to the text node. So if I select the text node, the inputs are much more robust on a text node. You can plug in another text node. You can plug in up to 20 images, up to 20 videos, up to 20 PDFs. And yes, we do accept PDFs in FLORA as well.
But again, at the top here, that's where you can select your models. If you're using the text node, Cloud Opus 5.5 is great. If you're more of a GPT person, feel free. I won't get too nitty gritty about which text models work better, where I just lean towards Cloud.
But this is where you can start to type in, hey, write me a prompt that'll do this, and that. So these are two really big nodes that I'm going to be using over the workflows that I show you guys today. But what this might look like is up here at the canvas, I have a few resources. I think earlier I saw a question about whether FLORA can understand typographic terminology. It's less of a question of whether FLORA can and more of a question of your text node.
So the answer is yes. If you're using a text node set to Gemini or Cloud or GPT, if you type in, you are an expert typographer, it'll know what a leg is, it'll know what an arm is in context to typographic anatomy. I included this graphic here in the canvas. Shout out, it was created by Orana Velardi. I linked to the blog post that I grabbed it from, just so if anyone's kind of curious to check that out.
But I just have this here so that we have a reference sheet. Sometimes I forget what the tail of a letter is. So if you ever need, hey, I don't know how to prompt that one part of the letter, reference I put it into Canvas for all your conveniences. I also put in some open source kind of cool typography pieces that I found online. These are all kind of just I think I got them off of like a public domain website. They're just kind of like there to kind of spark some inspiration. If you do want to kind of like go into Pinterest or go into type in use or something like that, like feel free.
You kind of make sure I was kind of like pulling from, some safe resources just to kind of like give you guys some inspiration. But I will be showing you like if you have a typographic inspiration image, how to kind of feel confident about using that reference image to create something new. But I have some resources on the canvas here. Bear with me. I whipped out the good old notepad and pencil and just started sketching stuff to provide you guys good stuff to just kind of start messing around with.
But I'm sure everyone, if you're here, you're probably, you probably have your own typographic studies to kind of bring in. So please, by all means, like bring them into the canvas, mess around with them and all that sort of stuff. But if you just need something quick to kind of work with, I provided a few kind of like sketch things here. I, God bless me, I provided some like little cute like geometric animal logo sketch concepts. I thought these are cute. I was kind of clever with that. And I also provided some already like graphic, letter forms or logo kind of like forms that you guys can kind of mess around with.
If needed, these were made by me. So feel free, you can just tear them apart if you need to. Interestingly enough, also I brought in some abstract imagery to kind of get the creative juices flowing. Like what if you made a letter made out of butterfly wings or, took a piece of cyanotype seaweed and took a letter out of it, you never know where inspiration strikes you.
So that's kind of fun to kind of experiment with. I also turned those images into depth maps. So to kind of take the color away and so you can focus more on the form factor of what's going on in the images. You can use these as well.
But I think in some of the demonstrations, I kind of just stuck with the more natural kind of material ones. Let's say I actually want to kind of take one of these letter forms. So I'm going to pick I don't even know what I was trying to do, like a two story. I'm going to pick the K because I'm actually kind of proud of that one. Actually, I have that here, but I'm going to do it again here.
So let's say you have a sketch here. When you hover over to the right side of any kind of image node, whether it's imported or generated, you'll see a little plus sign. When you click and drag that out, this is the this is where you see those lines that you kind of saw me do or saw me kind of like allude to in the earlier presentation. So if I click and drag that out and let it go, you're hit with this like little node selection menu.
So I can kind of go. I if I do this and my intention is to render this sketch as a graphic, I'm going to go ahead and select an image node because I want an image out of this generation. So I'm going to go ahead and click that. And so now my sketch is automatically connected into this node. You can see here at the bottom that that is indeed connected. It's kind of showing up there. I'm going to go ahead and set this aspect ratio to auto and what that does is just adopts whatever aspect ratio of the input connected and I'm just going to type in here something like.
Create a graphic. Create a lean black graphic on a white background. Based off the rough sketch attached. Go to my menu selection model selection menu up here. I'm going to type in GPT. Not G.O.T. GPT and I'm just going to select GPT Image 5 2.5 flare.
And then I'm just going to hit run. So what I can do up down here is I can either run. Or I can also hit a few variations here. So if I want to see maybe once before variations of this, I can click the plus sign here and then just hit generate. And what will happen is those four will kind of pop up on the canvas. When you do this and if all four don't show up on the canvas. Alternatively, it might look like this.
They'll start to pop up on the bottom of node instead. There is a preference thing here. I won't get too like nitty gritty about it, but like if that's if you don't see four nodes or however many nodes kind of pop up on the canvas and you see this instead, it's still the same thing. It's just kind of compacted into that.
So that's kind of one way that I would go about it. And so you see, these are a few of the results. I like how it interpreted the serif straight off of that rough sketch. It's kind of cute, I guess.
But a different version of this can be seen here. So let's say I need a text node to kind of help me write a prompt. I'm gonna start over here. Let's say I have my sketch. I can kind of click out, extend, and attach this into a text node. And I'm gonna set this to Cloud Opus 5.5. And I'm gonna write something I'm gonna actually copy the prompt here.
Write me a prompt that would create a clean, uniform, and refined graphic based on the rough sketch attached, black on white. So this is me asking Claude, write me a prompt to do this thing. So if I actually press this, I'm just gonna go up here like a cooking show, as if I already kind of made this in the oven. This went ahead and wrote out a much more robust prompt for me.
So a clean, refined vector, a logo type of a single stylized capital letter K, solid black, blah, blah, blah. Much more robust and much more kind of better understood by another AI model than if I were to write it myself. So I would start off with the sketch, attach into the text node. Now, very important thing, and I always emphasize this whenever I show people this. When you extend this text node into another image node, I'm actually gonna do this the wrong way first to kind of show you guys. And I'm gonna tell you why it doesn't work.
But if I connect this text node into another image node, without my reference sketch, it's gonna create something similar, but not based off of that sketch, because this is not connected there. So if I have this reference sketch plugged into this text node, and I'm asking this text node, write me a prompt based off of the image attached. Don't attach that image when I'm actually using the prompt down the line. It's gonna reinterpret what that K looks it's gonna reinterpret what that curve looks all that sort of stuff.
So what you do want, see what intarnation is that? I mean, what, it's cute. It's giving curly, it's fine. But what you want actually is actually, I'm gonna do this down here. If I drag this down here, I wanna make sure my reference is plugged in here. There we go, because again, you wanna make those connections. Whatever you want influencing your generation, you need to make sure it's connected, and you need to make sure you're intentionally bringing up how that reference is gonna be used.
So now if I generate this, I'll get something much more kind of close to my sketch. An example of that can be found here. So just a kind of reminder, when you do find yourself using this sort of thing, make sure things are kind of connected. So for AI models for this particular type of stuff, definitely go for editing. You can search up things like ideogram precise edit 4.5, GPT Image 2.5 sunburst, Flux 3 image, or even nano banana pro 4K for photographic edits.
And then for creating brand new things, ideogram 4.5, GPT, image 2.5 sunburst, those are really great for creating as well. So if you see here, I have a series of all those kind of sketches plugged into this blue box here. This is called our batch node. So a really cool function of this node is that it's almost like you're hurting. You're almost hurting like a group of images into a room so that you can walk into that room and give them a particular action that they can all kind of apply to themselves at the same time.
This second row is demonstrating the exercise that I presented earlier. So for example, these two images I have, one that's a little bit more expressive and one that's a little bit more serifed. But so creating a variation sheet, again, it's kind of as simple as using a prompt. So if I have a letter form, I can extend this out into an image node and the prompt is really the thing that will kind of do the heavy lifting for you here.
So actually, I'm going to re prompt this from here. So as if I'm starting from scratch on the left, I have a letter form. And on the right, I have an empty image node. I'm going to go in here and say something like create a, I'm going to pick a 12, create a 12 variation sheet that explores different typographic explorations or different typographic studies based on the letter form attached. I would like to see variations in line thickness, geometry, symmetry, I don't know, let's see. Color, no, stuff like that. I don't know.
And if I go look at this. Oh, inverted stress. That's kind of cool. Damn, I'm kind of OK. I'm digging that one. I might actually use that later for a little experiment.
OK. Yeah, that's really cool. Here's an example where I actually took one of the letter forms that was a letter A and I actually use a text note to do to create a variation sheet. But not for the letter A. I was asking it, in the eyes of a highly trained typography designer, write me a prompt that would give me a variation sheet with what the letter B would look like based on the letter form attached.
So now I'm not only just kind of riffing off of the letter form from the reference image. I'm starting to ask it, hey, look at the letter form. What would the letter B look like based off of what you see here? Or what would the letter C look like? Or what would the letter-- whatever, or what would the word "Catherine" look like based off of letter forms attached? I'm going to copy and paste this prompt.
But I'm going to say, in the eyes of a highly trained typography designer, write me a prompt that would give me a variation sheet. If it's a word, I'm going to say with what the word "cat" would look like based on the letter form attached. So I'm going to do that, let that write up. In the meantime, I'm just going to set this up to an image node so that once that's populated, I can just hit Generate.
So if we take a look at this-- oh, this is kind of cute. I think I like the bottom right one more. I do wish that the C kind of finished out at the bottom, though. So I might actually-- I think from here, let's say if I like one, I'm just going to go ahead and crop. Oh, sorry. One more time. If you have a variation sheet and you just like one of them besides all the other ones that were generated, you can go to the Tools at the top toolbar.
And there's an option to Crop. So if you select that, you can go ahead and just kind of focus in on the one you like and go from there. This is just a cropped view of that particular part. If you don't want to lose the rest of it's OK. It's still there. If you go back to the Tools and go to Crop again, you'll see that the rest of it is still kind of there.
So don't worry in that regard from here. But I'm going to try and say-- I like the design of the expressive type here. But I feel like the bottom part of the C terminates too soon. I want the C to look-- a little more complete. Write me a prompt that will do this.
So I'm going to do that. I'm going to make sure-- OK, I think that's all good. So while that text note is prompting, I'm going to attach this into another image node and then set that up. So hopefully, this will kind of fix that. If not, then I might try again with the text node and try to kind of re-prompt that.
But I think this should fix it. I'm going to try a few ones. I'm going to click this maybe three times. Not bad.
OK, that looks a little bit more complete to me if I uncrop this. I took this reference from that little pile of references I had at the top of the canvas. And I kind of honed in on that typography example up here. And what I did was just plug that into another image node. And I said something using the same handwriting from the reference attached, write the following in black on white. Graphic design is my passion.
So I kind of went ahead and got that. Pretty happy with it. If I had to get really, really nitpicky about it, the lowercase s is not cursive in the reference. And the a's-- the a's are kind of close. And you'll see here, if I wanted to start to experiment with what this would look like as inflated balloons, that's really easy to kind of prompt.
So attach that into another image node and just prompt something inflate the letters in this composition as white balloons against the black background and you get something like this. So what's nice here is that if on the left is where you have your layer selection. So if I go ahead and click layer one, which is the image that I just plugged into this, there's an option to split this image into layers. So if I select this, I'm going to go ahead and select C dream five pro layer eyes. Quick split is okay.
But if, rule of thumb, the C dream layer eyes, is a lot cleaner, you'll get much cleaner edges. So I'm going to go ahead and select that. And once this is going to do is turn this into a layered image. This is obviously like a kind of like a, there's a bad example because it's such a clean image already.
So what I'm actually going to get is just the black type on its own layer with the white background on its own there. You'll see up here, I have my bottom text on its own layer. And up here, I have my top text in its own layer. And so you'll see that these two images have different materials.
So if I kind of reset these back to their black type, Oh, they're kind of resized. My bad. Let me go back here. If I look at this and I'm Hmm, I don't want that type at the top to be chromatic anymore. You can select a layer in layer editor and just regenerate that particular, whatever is on that particular layer.
So if I go ahead and select graphic design text, on the left and I move over to the right, there's a section here called edit layer. There's a little option here called generate. And if I select this, I can start to specifically prompt something that will affect everything that's on that particular layer. So since I'm on the graphic design layer, I'm going to go ahead and regenerate this as give a water droplet effect to the text. And so now you'll see that kind of shining and that will kind of be regenerated while the rest of the composition stays the same. You'll see here, there are a few things that you can kind of move around. I have like a shape layer here.
That's a purple square. I have a green circle. I have a yellow square over here. I can move things around. I can kind of resize them as needed. That's kind of showing up there because this was generated on top of a background color.
So kind of just want to reposition up there. So that layer editor or image editor kind of helps you get much more nitty gritty about editing image compositions and stuff. So I'm going to pause there and see if there are any questions in regards to this. So let's see.
So one can't manually edit a generated graphic like a vector inside FLORA. We'll have to first download it as an SVG and then edit it. So technically you can input an SVG into FLORA and then edit the anchor points in layer editor. So actually if I select the circle, I double click the circle. I am able to kind of finagle the anchor points of that circle. However, I will say the capabilities of editing anchor points in FLORA are not as robust as something like a layer editor in kind of like Illustrator or Figma.
So just keep that in mind. Like we are a little bit limited in regards to like how manipulative you can get with like some of the vector points. But for sure, you can start to kind of if there were small tweaks, you definitely could kind of do that in image editor as well. I thought this cyanotype image was really cool and already it kind of looked like an A to me.
So what I did was I kind of just like extended that reference image into another image node. And I said something like make this look like the letter A. And I was pretty happy with that. Kind of just attach that into another image node. And I wanted to try and get it to a black and white kind of version of it. And I thought it looked okay, but I actually enjoyed the cyanotype effect.
And then I did the same thing with the letter B. What if I just wrote a word out of it? So I kind of just attached that image into another image node. And I said, make this look like the word ocean. The letters should be connected. And I got something like this. And I thought that's really pretty.
So I wanted to see what it would look like as a basic black, logo. So really nice. I've been seeing a lot of re-brands lately of like way more organic kind of typography for kind of like word marks. I kind of pre-generated this, but I have a 3D model of that cyanotype letter form here.
So I can kind of like start to spin this around. I can like look at it from the side. It's all dimensional. I love this little motive here. It's kind of hard to see because it's black, but this little like endpoint I could totally see isolating that and making a little graphic element out of that.
Let's pretend I'm gonna delete this. Let's pretend I have two groups of graphics. I have one group where the A's look like this, and I have another group where the A's look like this. A realistic opposition to this is if you have maybe serif A's.
Let's pretend that this is eight different serif A's, and this is eight different sans serif A's. What you can do is put each of those groups into their own batch nodes, and then plug both of those batch nodes into an image node. So actually, I'm gonna redo this. So if I have two batch nodes here, I can select both into an image node. And what you see on the bottom is that you see both of these batch nodes attached, and then you see an option where you can pair or you can cross them. Pair means that let's see here. Image one of batch one will interact with image one of batch two.
Image two of batch one will interact with image two of batch two, etc. So that's kind of what pair does, and that's why you only get eight variations of that or eight generations out of that. If you do cross, that means image one of batch one will interact with all eight images in batch two. So what does eight times eight equal? It gives you 64 generations.
So what the hell do I get 64 versions out of, right? So what I've done before while doing this is combine the geometric qualities of image one with the qualities of image two and do something like that. I can hit generate from here. So I'm gonna go ahead and hit generate because I work here. And this is just for the sake of demonstration. I don't recommend that you do this without a lot of intention because that's kind of how you go through your usage. And you can see here, I've reached my usage limit.
So be cognizant of that. I wanna show you this as a demonstration, but there are ways to kind of use this to your strength instead of just like kind of willy nilly. So what I end up getting is 64 variations or 64 versions of this generation. But if I look at these altogether now, I get this whole spreadsheet. Yeah, these didn't go through because of usage, so there's that.
But if they all ran, it would kind of look like this. And I have this matrix view where you can see where each one kind of lines up with each other. But yeah, the last thing I'll show you guys is a custom action I built where you can input a layer editor with text in it. So the input can be something like this. And I attach this into an action node. That's creating a collage of the letters in that editor.
So I kind of use that custom action node. And I asked it create me a letter collage maker that can input a image of a word. And isolate each of the letters and make them adjustable. And I need an option to kind of composite the seed, adjust the canvas width, change the background colors, etc.
