311 | How to generate professional graphic designs and videos for any need in seconds instead of hours or days (web design, interior design, marketing assets, brochures, architecture, etc.)

0:00Hello and welcome to the Leveraging AI podcast, a podcast that shares practical ethical ways to leverage AI, to improve efficiency, grow your business, grow your business, and advance your career. This is Isar Maitis, your host. And today I wanted to share with you something that I did just recently. I've done a workshop for a group of interior designers, architects, landscaping designers, and so on. And it was very well received and the feedback was fantastic. And I thought of sharing with you at least the highlights of that workshop, which was a few hours long and included to learn how to do each and every one of those things in more details.

0:34And we did a whole session about other stuff. but the guts of the design side of it is extremely powerful and it can be applied for more or less any kind of design work you do and i know what you're thinking i'm not a design person for this doesn't apply to me well if you need to create any kind of graphics whether it's for powerpoint for social media for brochures for pdfs for your website for whatever reason you need to create any kind of graphical design this is going to be incredibly valuable to you now i know what you're thinking this is a podcast, I'm currently driving my car and I can't look at the screen.

1:11So what are you going to talk about design? Well, I promise you that you're going to learn a lot. Even if you're not watching the screen, I will walk you exactly what's on the screen and why is it so powerful. And if you want to watch this, well, this is available, first of all, on Spotify as a video as well. But also you can check out our YouTube channel, which is available in the link in the show notes. So let's get this started. As I mentioned, the presentation was done to interior designers and architects and so on. That's why the focus is that. But I promise you that everything I'm showing, I will share how it can be used for other aspects as well.

1:50So the first example I want to share with you is the ability to take any 2D object and plant it in 3D environments. So there are multiple 3D environment design tools out there today. But in many cases, you want to take a real object, something in this particular case that your client actually likes, that he or she found online, and you want to plant it into your 3D environment that you can then screenshot, get design, render actual 3D outputs of it, and so on. In this particular case, what I've done is I've taken an image of a cool sofa off the internet, and I put it in two different tools. This is why you see two different options here.

2:30one of those tools is a free tool by Microsoft Copilot. So Microsoft Copilot has a 2D to 3D converter and what it allows you to do is it allows you to drop in a 2D object and then it runs through it for a while and then it exports this 3D object. So in this particular case you can see the sofa and you can see that I can move it around and rotate it and zoom in and zoom out and I can export that to in several different formats that then I can import into other 3D tools. Now, what you also can see is that while it's very good at capturing the essence of this, it is not highly accurate as far as the details and the fine look and feel of the actual object.

3:11So if you go back here, you can see the original ones have these little buttons and lines and so on. And the object inside of the Microsoft environment is not perfect. That being said, there are other tools out there today. And the coolest and best one that I found is called MESH-E. and what MESH-E enables you to do is basically the same thing only MESH-E is not a free tool there is a free version it is just limited with the amount of samples that you can create but you'll be able to see now that it actually captures the details in significantly higher level and it allows me to do the same thing again rotate this back and forth now because this tool is built exactly for that meaning it is built to generate highly accurate 3D versions of a 2D image.

3:57It has a lot more functionality and layers and texturing and a lot of other capabilities built into it. And again, you can get it through the paid version and get all the different capabilities. But what it enables you to do is now I can take this 3D tool and I can then go to any 3D tool out there. In this particular case, I've done a similar thing with a chair, like a little sofa thingy. And I can take it, create it in 3D and then drop it into a 3D tool, even free 3D tool like SketchUp. So those of you who haven't done any 3D design ever probably are not aware of this, but SketchUp is a tool that Google has bought and integrated probably over a decade ago, maybe even two.

4:40And this tool is really awesome for 3D design, both internal and external or objects or anything you want. It is really easy to use and many, many people use it today to create 3D designs. So what I've done is I've taken a picture of a chair from the internet, I converted it into 3D with Meshi, and then I'm able to drop it into my SketchUp design. So I'm taking a 2D image of an object I like and putting it into a highly detailed 3D environment. This by itself is really cool, but as you can see, SketchUp doesn't provide high-level, high-quality rendering, but it gives you just a solid idea of how things look like.

5:18However, there are other tools. In this particular case, I'm showing you another really awesome AI tool if you're into design that is called Rerender. And what I did here is I took the image from SketchUp, downloaded it and uploaded it into Rerender. And in Rerender, I can now render a fully realistic view of the same thing. So a process that was not impossible, but required a lot of design skills before goes from I see an image of something I like that I want to integrate into my design on the internet, I turn it into a 3D object, I can drop it into my 3D object environment in SketchUp or in any other tool.

5:57And then if that tool already renders fully high resolution things, I can do that. Or I can take it to a third party tool like re-render and then just drop it in there and it will generate, as you can see, a highly realistic, beautiful photo of the environment I'm trying to create. This can be used for any kind of design, interior design, buildings, objects, products that you want to sell. All these kind of things can be done in this methodology. But then the next really cool process that I want to take you through in a little more detail is the process of going from inspiration for a design to a mood board to a final design.

6:36So those of you who are not familiar with the concept of mood boards, mood boards allow you to capture the essence of a style that you want. This can be used, again, for product development. This can be used for web interface development. This can be used for, obviously, interior design and landscaping design and so on. Now, the process works this way. Either you, if you like a specific style and you're trying to figure out what you want to do, or if you provide this as a service, then your clients will give you multiple images of things they like from the internet. You can do this with different tools by building boards, but you end up with a bunch of images of different things that you like, whether specific design features that you like or specific furniture that you like or entire rooms and styles that you like.

7:22and you now have a bunch of these things. Going from that to a mood board that captures the essence of it, including the textures and the materials and the colors and the layouts is a work that used to take professionals a long time to capture the essence of it. Well, now AI knows how to do this very well. So in this particular example, what I took is I took X number of images off the internet and I dropped them into ChatGPT in this particular case. ChatGPT has a very powerful and very capable image generation, very good at reasoning and understanding what is it that you're trying to do. So all I did is something really simple.

8:00I uploaded all the images in this particular case. There are seven, but I could have uploaded a lot more. And I wrote the following prompt. Create a photograph of an interior design mood board that incorporates elements from all the uploaded images. The mood board should be organized, thought out, and crafted like a professional interior design mood board for design purposes. You may minimally overlap elements when applicable and use design techniques like collaging and transparency. White background, do not write any words yourself. And as you can see, or those of you who can see, it created a pretty cool mood board because it understands what a mood board is.

8:37It has all my inputs and it knows how to pull specific elements from my initial images that it understands what the style that they all combine together looks like and it generates a pretty good mood board. It has a color swatch on the right. It has different single individual components that it pulled out. It has materials and textures and it has multiple examples and they're elegantly blended together with a white background and no text as I requested. Now this is just step one. Step two of this thing is saying, okay, let's say I have a mood board and I want to apply that to different rooms in the house or to different whatever it is that you're designing.

9:18This could be an office, this could be a yard, this could be anything else you want to design. And so I took, in the second step, the image of the mood board and added an image of an empty room, in this particular case, a living room. And the output is an incredible interior design that is fully aligned with the mood board. How did I do that? Well, same thing. In the same Chachapiti chat, I just continued and wrote the following prompt. So now I have the mood board. I added the image of the empty room and I wrote, create an interior photograph of a room that incorporates interior design elements and furniture and decor from all the uploaded images.

9:55The photograph should be hyper-realistic and high quality. The lighting and mood should reflect the lighting and mood from the mood board and not from the current room image because the current room image has a specific style and I want to create a new style. I continued with the prompt, keep the room identical to the image I uploaded other than the lighting, keep the aspect ratio of the current room image. And I got this incredible outcome that looks like a professional designer took the mood board and applied it to the room. Those of you, again, who are not watching, it took elements from the mood board, including the desk, including some decorative components, including the stuff that was on the wall, including the plants.

10:36And then it added the overall mood with earth tones and cushions and a sofa and a throw on and just applied it to the room overall. And it looks absolutely fantastic and professional. So it took elements that it found in the mood board and turned them into the final design of the room. All with Chachi PT, all in a few seconds, all with just a simple prompt that I just gave you. But if you need to do this at scale, this will take you a lot of time, every single time. And if you need to do this on regular basis and not once, then there is a much better way to do that. So let's learn that other way.

11:13That other way is a tool called Figma Weave. It's a platform that used to be called Weavey until Figma bought them and turned them into that. Those of you who are watching the screen, you can scan the QR code to go to that platform. Those of you who are listening and want to see that platform later on, there's going to be a link in the show notes to go to Figma Weave. But what this tool can do is it allows me to build these workflows. In this particular case, the workflow does exactly what we did with ChatUPD, only it is pre-built. What do I mean by pre-built? I can upload a new batch of photos, click go, and it will run the entire process.

11:52In this particular case, I've done two things that are slightly different. I've made the prompts a little more detailed because they need to be more generic because I really don't know which images people are going to upload. So I've have some more detailed prompts. And I also connected more than one image generation tool to create two different variations in this case of the mood board. So how does Figma Weave works? The way it works is you have these boxes that are connected on a canvas that you can drag and drop. So you can see every one of these boxes, I can grab it and move it around. And they're connected with lines that basically shows you how the data is flowing.

12:32So some of those elements have just an output, some have an input and an output, and so on. So how does this thing work? And I'll explain what we see on the screen so those of you who are just listening can understand. On the left, I have the same seven images as before, and they are just input images. And to create these input images, I can just right click and then it opens a menu of all the different things that I can choose and then I can put it in or I can just drag images from my desktop or files and folders straight into here and it uploads them into the canvas. Then what I had to do is connect the lines from those into an image generation tool.

13:12In this particular case, I took two different options. One of them is cDream 4.5 Edit and the other is Gemini 3 Nano Banana Pro. Why did I choose those? Because both these models can get as input multiple images and still work with that. How do I add one of those nodes, one of those boxes? Again, just by right-clicking. And then here you can see that I have multiple different options, including tools. And in the tools, I've got editing tools, I've got text tools, I've got interiors, I've got helpers, I've got image models, I've got video models, I've got 3D models, I've got audio models, I've got custom models.

13:48All of these are here. And so So I can pick and choose here or go on the left side menu. There's a menu that does something very similar. The only difference in the left side menu, if I go to images and I put my mouse over one of the image generation tools that they have, it gives me a little cheat sheet that tells me which model it is and what does it know how to do. So you can see there are more or less every image generation tool in the world is in here. So I don't have to get specific licenses for them. I just stay within the Figma Weave environment and I can get access to all these tools.

14:18And the way I've built this is each and every one of the two models that I connected, again, one of them is Seadream 4.5 and the other is Gemini 3 Nano Banana Pro. Each and every one of them got as inputs the seven images. And again, how did I connect the seven images? It's as simple as dragging a line. You connect the line from one to the input of the other, and now it's connected as a source. And the other thing that image generation models need is a prompt. So I have a separate box that is called a prompt. And in there, I have a, like I said, a much more detailed prompt than before. but that's all I have.

14:49It's a prompt. How did I create the prompt? I've used a third-party tool, probably Claude or ChatGPT to help me write a better, more generic prompt. But you can see that with the same prompt, using two different image generation tool with the same input images, again, just by dragging lines from one to the other, I got two really good, really solid mood boards that are very different in their setup, but very, very similar from a design and style perspective. Both of them have textures. Both of them have elements. Both of them have images. Both of them have materials and specific elements that it pulled out.

15:28I must admit that I like the Gemini one better. I think it looks more professional and more detailed, but both of them are very good. Then what I did is now I went to Nano Banana in this particular case, just to show you that this is possible. And in this particular case, in the next step of Nano Banana, I connected two things. I connected my recently generated mood board that is now an item inside my canvas. So I have a box that is the mood board. I connected the output of that to the input of Nano Banana. And I also connected as a second input, the image of the empty room. So I have two sources connected to Nano Banana, again, together with a prompt.

16:05The prompt is slightly more detailed than we've seen before, but you can see it generated it and it looks absolutely amazing. The cool thing about this versus doing it in ChatGPT is a few things. One, this process is now a machine that does it. If I want to put in new inputs and just run it again, I will get a completely different output. So those of you who are watching, I will show you a second example that is the same exact process. I just uploaded several different images. So the first one was very much earth tone, laid back, modern style. The second one has a very poppy color kind of scheme.

16:42It has bright yellows. It has lime greens. It has orange. It has pink. And so it's much more lively. And you can see that the machine generated completely different mood boards. Again, both of them are really good. Both of them from the same models. I connected the same room as before, and I got a very different image that uses that style that looks really cool, very colorful, very poppy, and uses components from the mood board. Now, the other benefit of doing this not in ChatGPT is when you run the model, you can tell it how many times to run it. So when I click on any of the boxes, if it's an image generation model, video generation model, anything else that I'm doing, on the right side, it opens a menu for me where I can define many different capabilities.

17:27In this particular case, I can define the resolution. So 1K, 2K, or 4K, I can define the aspect ratio or what I want it to be. And I can define how many runs to run it. and it tells me how many credits it is going to consume. So if I want X number of options that will all be based on the same thing, all I have to do is select the number of runs and then hit go. And again, that becomes the default and I can run it again and again as many times as I want. So down here, you can see six different variations of the same room, all with the same color styling. So different colors of the walls, it changes different images in the background, different furniture, but all very obviously following the same exact style.

18:08And so if I want for myself to figure out different options for design of anything, and again, this doesn't have to be interior design. This could be for designing a webpage. I want to see how the webpage will look like in several different formats or several different styles or several different color schemes and so on. I can do this with this tool. The same thing for a brochure I want to design, the same thing for a proposal I want to create and I want to create a cover. Whatever the case may be, you can create it with this flow and every time you put in new inputs, you run the whole thing. You're going to get X number of variations of the output depending on what you define.

18:42This is taking a process that used to be manual and took literally days to seconds and it is really, really affordable too. So you're buying tokens and then you're just consuming your tokens or credits in this particular case to create whatever outputs you want. So a fully automated design process that I can now put different inputs and get different outputs. Now, while this tool is really, really awesome, and like I said, it has integrated all the different image generation and brain generation and prompting and video and audio and all these other things, it can be used for a lot more than just interior design.

19:23So I want to give you six different examples. Most of them are design-related because, again, this piggybacks on a workshop that I did. But I will give you two examples that have nothing to do with interior design and how to do with different types of promoting a product, whether you're selling things online, whether you're selling things in an actual shop and you need to promote them in different ways online and so on. So the first one is what I call sketch to video. And what I mean by that is in this particular case, I took a third-party tool that works in a CAD tool. So what it does is it knows how to take a sketch that was done by hand of a room.

19:59So this is, those of you, again, are not watching, there's a fast running person that is sketching on a paper an interior design with a bunch of rooms and a bunch of furniture but then using ai this hand drawn kind of like back of a napkin interior design turns into a fully 3d rendered object inside of a true design tool and this doesn't actually change the design tool this is a design tool that a lot of people are using this is just an AI plugin that connects on top of it and knows how to take the flat 2D hand-drawn thing and turn it into a 3D rendering that then you can move around because it becomes an actual object inside of this tool.

20:42But what I did then is I took the output of that, so a screenshot of the 3D object, and I created a promotional video for something completely different using that 3D model. So I went from a flat back of a napkin sketch on paper to a fully rendered promotional video in this particular case for high-end speakers. Now I'm going to hit play. You're going to hear the music if you're just listening and I'll walk you through what it's showing. So it's starting with the actual 3D model, but then it goes into the room. Now in real time, furnishes the room. It looks like a professional, beautiful design room and a beautiful young couple walks in the door and they're sitting on the bed.

21:24The camera is rotating around. The speakers show in and they turn into this sign and then that turns into a promotional ending that says who is the vendor of the speakers and so on. And all of that was created inside Figma with, including the music, including the sketches, including everything. So this is one example of what you can do with this tool. Another example is architectural angles. In this particular case, what I did is I took an original image of a room, and then just by giving it prompts, and the way it works is exactly like before. I have an image of a room with all its different components, and then all I did is told it how to change the angle.

22:10So the model, the image generation model in this particular case is cdream version 4 edit and then it gets as inputs two things one input is the actual original image which is a eye level view of the room and i wrote as the prompt which is the second input change the camera angle of this shot to a high angle shot and it gives me the high angle shot of that room and everything else stays exactly the same the lighting, the furniture, the accessories, the cushions on the sofa, the rug, everything stays the same, just the angle changes. And then I've did this multiple other times. A few of them are really, really cool.

22:49In the second one, I asked it to give me a close-up of the kitchen table, and it did. So in the original image, the kitchen table is really far in the back after the living room. And in this image, I've got a close-up image in an angle of the actual kitchen table with the flowers, with the sink, with the bowl of fruit that is on it. Everything is kept the same, just a really good close-up image of that. In the second one, I created like a cropped image. I put a square around a section of the original image. I didn't actually crop it. I just highlighted the section that I want, and the prompt was to give me a frontal shot off the couch, rotate the camera 90 degrees left.

23:29And then I got the outcome that actually looks straight at the sofa and the window behind it with all the different components. And it looks perfectly realistic and so on. And then in the last one, I combined two images that I had. One is the high shot and one is the zoomed image. And I combined it into a video model. In this particular case, Kling 2.1. And I wrote a simple one line prompt that says camera dolly in from the living room to the kitchen. And what it's doing it's showing think about a very slow drone shot that goes from a top view of the room sliding slowly forward into a perfect close-up view of the kitchen table which were the two original images that I gave it as the starting point and the ending point so cling 2.1 is one of the tools that knows how to take a first frame and the last frame as the inputs when it comes to generating videos so it knows how to take both of those and creates a really amazing beautiful shot that you You can show somebody the design of their house or, like I said, anything else you want.

24:31This could be a website. This could be a product. This could be whatever else you want. And you can generate all of these things. But as I mentioned, in addition to all these cool things, if I'll put in, if I'll drop in another image in the source, the very first image that showed the room, as long as the room has a kitchen table, because otherwise one of those prompts won't make sense, I will get all these outcomes with a single click without having to do anything about this ever again. And then obviously I can take and add other components in the room and continue the motion, the other video going into other shots so it can continue the flyby if you want the drone shot inside the house to show more and more components.

25:11Another use case that is really cool is in this particular case, I took a image of a 3D wireframe from the internet of a house design. And I've asked AI to generate multiple outcomes of it, including like a hand created house, like a little carton board like architects do as to show what the project would look like. And then I took a image of an actual house in the world and have it blend together and work together. This particular example is based on a workflow of a guy I follow online. So it's not my workflow. I just played with the details a little bit, but what it enables to do, and that's by the way, the other cool thing about Weavey is that there are a gazillion examples on how to do things online.

25:57And most of these allow you to copy and paste the template. So in this particular case, this is what I did. This is not my workflow. I made minor changes to the workflow, but I was able to take a new image that I like of a different house online and place it in an actual street and get to a professional high resolution image with people walking on the sidewalk and a car and a beautiful evening lighting showing this new house in a location where in reality right now a different house exists. And one of the things that you can see here is that I have two versions of the same image because one of the or several of the models that you have here are upscalers.