Andy Walsh shared his thoughts on what it means to be a concept artist in the industry, experience with Blender, and the workflow for the project Owl Sanctuary.
In case you missed it
You might find these articles interesting
My name is Andy Walsh and I studied in my bedroom mostly. While I painted and drew a lot in high school, I stopped altogether at about eighteen years old and pursued more of a graphic design/web design career with some 3D art thrown in for fun. I always loved the games industry but could never figure out what my path would be. After getting quite tired of that career (more than ten years on), I decided to get back into art and see if I could make that work.
Since then I've worked for companies such as Tendril, Territory Studios, Illfonic, Freemantle Media, and with such clients as Sony, Microsoft, Mattel, and various others. Projects include Film, TV (such as American Gods), games, book illustration, advertising. Quite a wide gamut.
Finding a Place in the Industry
I'm not quite sure what I specialize in really. I've never fully figured out if I'm an illustrator or concept artist. There's some grey area where the two overlap and I feel like I'm there. Mostly, clients come to me because they like the particular vibe of my portfolio and want me to work on their project and so over the years I just respond to whatever projects come in, always hoping that one day it'll be a big one. I didn't really 'do' anything to get in the industry as I don't quite know how one gets in. I just get offers from time to time.
What It Takes to Be a Successful Concept Artist
Speaking of the essential skills of a concept artist, I'm not even sure if I'm fully qualified to say something here. I think I know what they are but it is like asking a brown belt what skills it takes to be a master. Having been in the industry long enough and seen enough job adverts (I even advertised jobs myself and played a role in recruiting), I can say that a concept artist pretty much has to be a kind of superhuman freak of nature. What I mean by that is they have to be able to draw/paint/design anything that walks, crawls, locomotes, lives, and everything in between. Ideally, they need to learn it all within a few years so as not to die of old age in the process. Had many humans not achieved this seemingly impossible feat, I'd have deemed it implausible! So if you think you're able to do all that, then you're on the right track for a great career. Good luck!
Love-Hate Relationship with Blender
I shrugged Blender off until Eevee came along in 2.8 and I saw Florent Lebrun do an amazing scene with it. At the time I was trying to get away from 3ds Max as it was just feeling too cumbersome. I was looking for the right software to handle atmosphere and fog without taking an age to render. This is because just about all the scenes I ever work on have atmosphere – it's a cinematographer's main tool after lighting. So I dabbled with it but I kept modeling in 3ds Max where I felt comfortable. Now I'm pretty much full time in Blender.
Don't get me wrong, I have a love-hate relationship with it! Having used something as old and powerful as 3ds Max for so long, you really are dumbfounded by some of the features that Blender doesn't have. Really really basic things like drawing out shapes and having each object with its own wireframe color. Currently in Blender, when you view in wireframe, everything is black, so you can't even tell the difference between objects. I could go on all day. Plus it crashes while rendering a lot. It got to the stage where I got sick of it and decided to go back to C4D (another old fave) but quickly realised that, actually, Blender is still better. It's crazy. I've become dependent on it now and can't go back. It's largely down to the real-time viewport rendering. I can change materials instantly and don't have to tweak, render, wait, tweak, render, wait. And this is absolutely indispensable.
Some of the latest updates are cool but I just wish they'd focus on some of the fundamentals first.
Owl Sanctuary: Project Breakdown
I wanted to make a medieval house and I thought I could get away with building it in Oculus (well, Adobe) Medium, but that program is awful at doing anything that requires precision and right angles, even with the snap tools on. So I went back to old faithful 3D Coat.
Here's a screenshot of the house as I was building it. You can see that I built it beam by beam and then destroyed it.
And below is the finished house in Blender:
I got inspiration from table-top gaming models. They're so intricate and made by hand!
The snow is a combo of shaders that make everything facing upwards one material and everything else – another material. The chunky snow is a free Blender addon called Real Snow. It's a little blobby but can look great in certain circumstances.
Then I just downloaded as many free assets as I could from Sketchfab to populate the scene and went about lighting it and picking good camera angles. The trees were all downloaded and most of them didn't make it into the final shot as they look a bit too fake, except maybe in the final daytime shot.
The stone slabs were first tested in Blender, then I went into 3D Coat, made decent quality ones, and replaced the test slabs one by one. I also tried that in Medium where I exported some of the basic geometry and sculpted around it, using it as a guide. But I was modeling blindly (i.e. not modeling directly to the camera angle, just modeling and hoping for the best), and when I brought the stone structure into Blender, it screwed up the composition. So that's 0 for 2 for Medium in this particular project. But I think some of the structure remained as a guide for the overall shape of the old cobbled-style road. Or is it a bridge? I'm not sure.
I don't want to get into too much detail as I may release a full tutorial on this but let's look at some of the lighting for the two scenes. In the night scene, there's essentially one main area lighting pointing down on the house and a few warm lights representing the fire. The fog comes from a volume shader in the World settings:
Which works better with Eevee actually. For some reason when you use global fog, Eevee won't render the HDRI as the fog goes to infinity, cutting off the environment light. And this is my pro tip. If you want to get the most out of Eevee, use a sun and a daytime set up with cast shadow. If the lighting is diffuse and the fog is thick, Eevee can start to look thin and fake and when you turn on Cycles it's miles better.
The third shot took forever, and yet everything was pretty much done already. I had an entirely different shot in mind but tried many, many camera angles, none of which worked, so I went back to the drawing board and started the lighting from scratch.
The post-production was also tricky as I had to render out each depth-layer separately. Everything that overlapped was rendered separately and comped in so it had its own layer in Photoshop. As such, much of the fog was baked into the render. It was kinda complex. Then I just set about matte-painting it so anything that looked too 3D was massaged into realism.