I’m a visual thinker, big time. If you’re explaining something to me, I’m probably picturing what you’re talking about so I can follow what you’re saying and make something useful of it. If the picture starts to fade on me, then you may as well be speaking jibberish. That’s why rapid prototyping concept art is such a gamechanger for me, at least, in storytelling and game design. And it can do much more than prototype, as long as you’re not afraid to spit and polish.
In the last few months, AI art generators have taken off like a rocket and are rapidly improving in functionality, customization, and capabilities. Stable Diffusion, Codeway’s Wonder, and Artbreeder are my favorites right now, depending on the functionality I’m looking for. Midjourney and Dall-E have stolen all the oxygen out of the room as far as the media running with this narrative, but for fantasy / speculative fiction concept art they don’t offer the styles and datasets I need.
At Grailrunner, we’ve recently incorporated AI-aided art into our workflow for marketing images, for the website graphics, and to some extent in our products. I’m sorry if you’re an artist who feels threatened by this marketplace shift, but it really is a technology that is unlikely to go away or accept a lot of regulation. At this point, with millions of images generated per day across multiple apps, it feels more like an unstoppable tsunami you should probably figure out how to surf.
We just added another Lore Card to the Story Arcade here on the site and thought it would be fun to show a behind-the-scenes on the work, mainly to show how we’re using this fantastic new tool in what we do here.
Photobashing is a technique where artists merge & blend photographs or 3D assets together while painting and compositing them into one finished piece. This is used by concept artists to speed up their workflow and achieve a realistic style. –Concept Art Empire
Stephen Gibson, Art Director and designer of Grimslingers makes an interesting comment about this: “My current style for Grimslingers is photo/3D bashing. I collect images to splice together and keep painting over it, splicing in new images to fill out the character until I can’t stand to look at it anymore.” -Interview, Nov 2022 ImagineFX
Funny, huh? Anyway, photobashing is a big part of my concept art journey. I started off trying to paint everything myself and realized that’s not where my talent lies. Things accelerate and honestly look a lot better if I pull together stock images or 3d assets from places like Turbosquid, Shutterstock, Archive3d.net, Free3d.com, Nasa (nasa3d.arc.nasa.gov/models), Sketchfab and Daz Studio into Blender and work out compositions there. A lot of museums are starting to upload their collections as 3d models as well, typically for free!
Pulling assets together
Blender is still my go-to tool for compositing 3d assets into something useful because you can manage the placement and lighting and mess around with textures in Substance Painter. Incredible flexibility, but it’s time consuming. It’s also a place to build up assets that are unique to a world I’m creating (and so won’t be available anywhere. Then Photoshop. Always Photoshop. Nothing is done till it’s been through Photoshop. It’s a thing.
This is an example of what I’m talking about. The ball lightning carbine is a distinctive weapon, strapped to the arm of practically any adventurer in Grailrunner’s Salt Mystic universe. I made this thing in Blender with some parts from various models (a motorcycle, crutch, and I forget what all). The frame of it and the leather straps were just cylinders that I squashed and pulled into place, then added textures. Now I’ve got this thing in a hundred styles and orientations.
1. For our new Lore Card, it started with an idea: a dramatic aerial view of a carbine gunslinger on a mountaintop with a wide valley below him. Sometimes the story comes first, but in this case only the mental image. I’d write the story behind that guy after I saw him.
2. With Stable Diffusion, I experimented with a series of prompts suggesting the aerial view, the mountain top and valley, and the “fantasy gunslinger”. It took patience, not going to lie about that. And I cycled through several artists incorporated in the prompts to try different styles as well (you can mix artists too!).
3. Once I got something that looked like it would work for me, it needed some basic touch-up painting and color & tone adjustments in Photoshop. There was also an annoying misshapen character standing there (instead of my gunslinger that I asked for!) that had to be just erased. Photoshop has vastly improved capabilities now for easily removing stuff.
4. I still needed that gunslinger though, and went back into the cycle loop trying various iterations and prompts to get a guy in the right posture and wearing the long, gunslinger coat I was looking for.
This is the character I eventually went with. They can turn out with three arms and nightmarish faces, the fingers often run together and look like tendrils. Seriously, the output isn’t always mind-blowing, but here at least I saw the outline I wanted and general textures.
A bonus was the weird almost rectangular thing he had on his left arm, which if I squinted looked to me like our carbine without any extra editing.
5. I cut him out of his background and placed him on the mountaintop, gave him a little shadow, and darkened him to almost a silhouette. I was close, but it was bugging me that he was up on the high plateau without a clear story of how he got there.
6. I already had an asset for an airship from a previous work that I repurposed here. There’s a fantastic feature in Photoshop for automatically adjusting color grading and tone to match another image, so much of the hard work was done for me with that option. I just needed to trim it up a bit and place it in context.
You can read the accompanying story here on the new Lore Card. Here’s the final piece:
Let me know what you think, and if you have your own art journey going on. I hope you liked the peek behind the scenes.
Till next time,