How to Create Concept Art with AI
Using Dall-E mini or Craiyon, and other AI image generators, then polishing with Photoshop.
Have you ever been faced with creating some detailed concept art to convey ideas to an audience? Do you want to be able to complete multiple pieces in days instead of working on one piece for a week? Are you lacking the tools, skills, or time it takes to make your art and also have it look accurate and visually appealing? In this article I will teach you how to create concept art with AI, as well as provide detailed examples of my own work going through this process.
When it comes to creating concept art, you often need highly detailed images with quality lighting and backgrounds. For character concepts you can usually get by with a sketch, but most times this just isn’t good enough. Sometimes you wish you had the skills, or the necessary equipment to render on the canvas exactly what you see in your mind. For those that are not good at drawing, or don’t have the time, AI can work wonders in this regard.
Disclaimer:
I do not work for Adobe, Craiyon, Dall-E, OpenAI, or any other image generation site. I am not being paid for any links, and this post is not sponsored by any of the sites used herein. To that end, this is also not a review of the sites I am using. This article is for education purposes only. I am simply informing people how to use these tools like I have to get similar results. I am not affiliated in any way with these web-based or commercial software. However, I include links to some of these websites for two purposes: To help guide my readers to follow the instructions, and to give credit to these companies whos software I have used.
Easily Avoid Plagiarism
Many times, creators fall behind and want to publish lots of work quickly and get it done. When this happens, they often resort to editing pre-existing artwork without consent, or plagiarism. In this article I will teach you how to create concept art with AI that is your work, free of copyright or trademark, and also have it look stellar.
Are you this guy?
This was me back in the day. I had pencil, pen, and paper, and access to Microsoft paint. I was limited by my ability to draw, and I was limited by my software skills. In any case, I wanted to make concept art. And this was the result. This is the main character of the story from way back in the day that I am attempting to fix in my “How to fix terrible writing” series. If you want to do way better than this with minimal effort, then keep reading.
Dall-E 2 on OpenAI created these in less than 30 seconds each. While not perfect, they are remarkably better, done faster, and can be improved with a little extra input.
What you need
To accomplish this, all you need is your PC and some rudimentary image editing software. By the end of the article you will have the skills to create your own unique and detailed concept art for your audience. Firstly, we will be using Craiyon.com, formerly Dall-E mini. This AI image generator will do the bulk of the work for you. All you have to do is have the ideas and put it into the right words for the AI to translate.
Note that while I was completing this, the full Dall-E 2 project on OpenAI was not available to the public yet. I have since tested it out as you see above above. It can work wonders if your art needs to be photorealistic or cartoon, but you can find yourself paying extra $ to generate more ideas. Craiyon is totally free at the moment. One of the drawbacks of Craiyon is that it’s terrible at generating faces. Often the output is just monstrous. With that said we will still need faces.
So, we will use https://generated.photos/ for generating faces for our characters. This is optional. If you don’t need faces, or are using personal photographs you won’t need this. You don’t have to pay or sign up if you just need to download some previously generated faces. However, if you want to create an in depth face, you’ll need to pay or sign up.
Lastly I will be using Adobe photoshop to assemble and polish the images into completed pieces. Photoshop is not required. I use it because I have it. But you’ll see as I’m going through the steps that any image software that can cut and paste accurately can accomplish the same with maybe just a little bit more or less effort.
Before we get started, I need to post a spoiler warning. If any of you are currently reading my Web Series eBook, Star Stream then there are minor spoilers below. For any of you that have not read the series and would wish to check it out. Follow the link below.
Getting Started
To get started, we are going to go get our faces. You don’t have to do this first, but I find that compiling a folder full of faces helps save time. This is because I usually stock up on so many that I can skip this step on later projects and use a face I have already downloaded. In some cases you might have a particular face in mind. It’s perfectly fine to do this step later if you want to match it to the body of your work.
Generate Faces
So lets head to https://generated.photos/ now.
Since we don’t want to pay right now, we will click on “Browse Photos”. Otherwise, you can click “Generate a photo” to sign up and pay to make custom faces. We are going with the free option here.
Once you’re on this page you can select any of the faces you like. If you click on one, it will show you all of the faces similar to the one you clicked. You can fool around with the settings on the left for other options. For this example we will pick a male face and a female face, and then click on Download.
Create a new folder and store your Faces.
As you can see above, I’ve stored my AI generated faces in a folder made for Faces. Moving on you will see why this is the easiest part of “How to create concept art with AI”. Now we’ve gotten it out of the way and can focus on the more complex AI images.
Craiyon
Let’s go to https://craiyon.com/. Here you can type in just about anything and have the image generator go to work. One thing that this generator excels at is not only giving you the concept that you were looking for, but make it look fully rendered and have a good backdrop.
Examples
First we want to tell it what we want. In this case, I use different forms of phrases until craiyon gives me something acceptable. Each phrase will take about 30-45 seconds to generate 9 images. Sometimes you might get exactly what you’re looking for. Other times you need to re-phrase your query. And some times the 9 images are almost exactly what you were looking for. But, you click to draw again on the same phrase over and over until you get the perfect output for your needs. All of the above are illustrated in the examples below.
Once one of the 9 results suits your needs you can click on it. That will expand it to fill the block. Then you can click the screenshot button below to download your result. Lets go through this process a few times. Take note of some of the phrases I use in my screenshot, and how they differ from subsequent queries of the same object. This will help you envision how to re phrase your queries to get desired results.
1. Characters
Male Character
I want it to create a suit of sci fi armor with a cape, and I want it to be all white.
Female Character
Here I want it to create a female that is projected as a blue hologram in a pilot suit. I also want her to have a nurses outfit in one of them.
I also want her to have a virtual chair to sit in.
Combining our Images
Now we take our bodies, our faces, and our props and get them ready in photoshop. The first thing we want to do, whether in photoshop or another image editor, is get rid of the background behind the face. Background needs to be transparent to paste it over the body. In the case of the holographic female, I also want to change the hue to match the blue color. I’ll also do some blurring around the edges of the head where the background was cut out. This will soften those edges to help them blend later.
Next, we want to bring up our body and use that as the main canvas. For the male character, I need to extend the height of the canvas to fit in the head and then paste it. For the female character, I position the face until it fits right where I want it on the head of the body. Then I adjust the transparency of the face until I am happy that it looks “holographic” enough.
Lighting
Now that we have our faces aligned over our bodies, we will want to work with the lighting to further blend the faces with the body and background. I want the woman to look like a lit up hologram, so I focus on rendering a bright central light. Now, instead of the ghosted face pasted on the wire frame, it almost blends perfectly. I do this for all concepts of this particular character. For the males, I simply fine tune the placement of the heads and paint in some minor shoulder alterations if needed to connect. Then I render a soft directional light to help make it all seem like one full piece of art.
2. Props
If you are satisfied with your characters, or if you want to move on to something else, we will begin with “How to Create Concept Art with AI: Props”. Following the same process as before, input phrases and edit them until the output matches what you are mentally imagining. In the case below, I had imagined a cockpit dock for a spherical metal object. After just a few tries, the output matched to about 99% of exactly how I imagined it in my head.
Next I was working on a weapon. But I was being very specific. This one took a large amount of time and tries to get right, but it was worth the effort to get the pieces I needed.
Finally I needed a glowing red rectangular stone. This one Craiyon managed easily.
3. Settings, Locations, Abstracts
This is one of the areas where Craiyon really shines. Not only can it construct the structure, or the building you are imagining, but if you give it the right input, it can make great background settings behind your structures. Craiyon has a distinct way of taking your input and generating an image that closely resembles a dream. I find that with concept art, the abstract nature and the blurred features can oftentimes work wonders. The level of detail that can be achieved in the surrounding areas of your structure is just there, with no additional work or miniature painting of little people or smaller structures. Yes, it is probably a little out of focus, but with a little work you can make it perfect in no time.
Structures
For my first structure I want a metal building in a sci fi surrounding that resembles a nuclear reactor stack.
My next query will be huge domed cities of the Sci Fi variety. It took a good deal of time to come up with results that I was satisfied with. But I wanted multiple different versions of cities, and not just one particular look. These will all be separate places so I need them to be somewhat distinct from each other.
Locations and Abstract Concepts
We are almost at the end. Some people would normally stop collecting generated images at this point and polish what they have. But I think it is worth it to add more to the compilation. Some ideas are very abstract, or just really hard to visualize, and this is another area the Craiyon really shines. Just a few queries and it’s possible to have displayed something from your imagination that would have been impossible to put on paper. I’ll display a few examples below, and then we will move on to the final step of “How to create concept art with AI”.
Finishing Touches
Photoshop in your final touches. Do some touch ups.
At this stage we have gathered a significant amount of images to compile into a concept. We have characters, props, places, and ideas to fill our imagined setting. We’ve come to the final stage. Put it all together to show off your concept to the world. To do that, we first need to put some finishing touches on our artwork. Go back into photoshop and put some stylized logos or text that fits your theme. If you don’t have photoshop, try an alternate image editor that will help you achieve these same goals. With that said, the following are the final pieces and a brief explanation of how I finished them.
Male Characters
Most of what I’ve done here is explained in the character section above. All that was left to do was add my logo and names for the character concepts.
Female Concepts
For the holographic female. I had a lot to conceptualize. For starters, getting her to look like a hologram. That is explained above in the character section. But then I needed to express her versatility. She has other outfits and props. I combined one of her bodies with the chair I had Craiyon produce. For the Nurse, the hat just looked awkward so I got rid of it. I also put her orbital sphere inside her body for normal size and below her for her miniature projection.
Props
The props I ended up making were the ones I felt needed to be conceptualized as they are unique to this story. These aren’t everyday items. They are objects I created for my story. To fully convey the appearance of these items I felt having some concept art was the best route.
For Val’s orbital, I made some minor edits to the lighting, but the main significant change was adding the lettering for VAL on the sphere. This is a crucial identifier in the literature of the story. The floating aspect and the docking were also on my list of things I wanted to display.
The weapon I had envisioned took forever to get the right results. In the end I’m not as happy as I want to be with what I ultimately ended up with. It’s not for lack of trying. This was just one of those items where my mental image was so much harder to put into words that matched with what Craiyon output. Having said that, I still wanted to display the idea somehow. So I put something together that is not 100% but at least satisfactory for the concept I was going for.
Contrary to the results for the Lance Noir, the Artifact came out to be precisely what I intended. Craiyon did a good job with my inputs, and I was able to take that and make a few examples that 100% match with my mental image. For the Artifact, I combined the cubes from the beginning of this post with the glowing red crystals. I’m really impressed with how it came together.
Structures and Locations
First I want to showcase the “stack like” structure. Some of the outputs from Craiyon were quite tall. What I had initially envisioned was about half the height of a typical reactor stack. For the examples on the left, I touched them up with some color changes and reduced the height of the upper stack from the original output. This setting will exist at different times in the narrative, once green and lush and after that spoiler, so I did want to showcase some green in the surrounding area.
Pristine cities, or the domed mega cities, were one of my more lengthy projects. It took me quite some time to get them, often querying the same phrase over and over until I was satisfied. The first 4 I didn’t edit, and just combined. However, for Draconis Quar’d, I added a ship landed with the agent walking away, because that output came out exactly 100% matching my mental image of that scene for the book (not yet released).
Abstracts/Mental images of concepts
Of all the abstract ideas in my narrative, the concepts I wanted to convey early on were stream travel and the star streams themselves. I also needed to display the concept of the multi-faceted shield array, because the complexity of my visualization is hard to convey in words. Since these things are central to the narrative, I wanted a visual representation of what these ideas and actions looked like in my head. There will be more but I felt these three instances were very crucial to put a visual to at this point in the narrative. None of the Craiyon outputs for the following were edited. These are all compilations straight from the results of my queries.
Conclusion
If you’ve made it this far then I hope you have enjoyed the artwork, along with the accompanying instruction. While I know that this isn’t the only method of “How to Create Concept Art with AI,” it is the method I used. Because I was so happy with the results I felt that I just had to share this process with others. You will be more than capable of coming up with concepts of this quality or better following these steps of instruction. You can also do the same with varying success using other methods, or skipping/adding to this one.
In the end, it doesn’t matter how you achieve your desired result, as long as you are satisfied with it. To that end, I hope this guide helps you and many others produce artwork that they never thought they were capable of. That is my ultimate goal, teaching someone that with today’s technology, with computers and AI, the impossible just got a little more possible.
Farewell
For those of you that enjoyed this content, please subscribe to my newsletter and visit my various social media pages listed on the homepage.
Everyone Please check out Star Stream if you haven’t already.
Here are some useful links relevant to this article.
Thank you Everyone and have a wonderful day.
~Steven
Comments are closed