Using Blender For Crafting 2D Magical Cave Illustration

Introduction

I got into 3D during my time working in publishing, ultimately, I chose to learn Blender so there was more versatility to what I was making and how I was making it. I’ve come at 3D from a 2D background, for me, it's always been about twisting the package to fit my needs, as it wasn't a decision that was affected by the final product, because I could have drawn it. It was more a case of enjoying the workflow more than traditional 2D, and finding out how far I could push Blender towards a more illustrative style.

The Merlin's Cave Project

As this is a client piece, it was more about telling the right story and surrounding it with something awe-filled. I knew I wanted to stretch out the composition, play with the height, as caves are such an enclosed space. The references I used for this project were entirely for the cave itself. I wanted to make sure I nailed that so I could build on top of it and make the scene feel fantastical.

Modeling

I work with a print-first mindset, so the first thing I usually do is render a cube in place with my shader on to set the scale; that way, I know what my minimum detail level is. From there, it's about getting some rough shapes and shadows using cubes. For this piece, the biggest stick in the mud for me was getting the cave to feel right, so much of the early process went into pushing and pulling to get a space that felt suitably enclosed, but still had that height to it. I try to do as little modelling as possible; I'll use assets for anything I would consider "natural", rocks, trees, etc. I use a lot of Quixel Megascans for that, and I've got a sizable hard drive full of anything I need regularly (lots of armour pieces, for example).

For anything I've sculpted, I'll apply a series of modifiers to bring the poly count down and stylise it at the same time, this gives the line modifier something to grab onto in predictable places. As for UV'ing, I avoid that as much as I can. The view transforms in my shader are sufficient enough in most cases where I won't need UVs as a guide, and I can get away with box mapping any textures.

Texturing

The goal from the get-go was to be able to use as many assets as possible, so my shaders were set up to take the textures from a PBR workflow. The shading comes from mixing together 3 separate wave textures that are offset from each other at 30°, which allows me fine-grained control of how much of each level is applied and where (for example, clamping the darkest values into recesses using AO).

The painterly texture that sits underneath the line-art comes from passing the raw light values from the principled BSDF out into the compositor, and finally through a kuwahara filter. I apply that same workflow to several other passes, such as depth, for use in the colouring process in Photoshop.

After blocking out, it's a case of gradual refinement. The biggest benefit to using a 3D program in this way is that, for the most part, the image is editable right up until the end. Increasingly, I find myself using particle systems less and less, opting to place everything individually, as there is really no beating a deliberate hand.

Lighting & Rendering

I try to keep my lighting set-ups as simple as possible, as render times can run long quite quickly. Typically, by the time I'm getting close to moving into Photoshop for colouring, the render times are around 5-10 minutes, which doesn't sound like a lot, but I'm re-rendering constantly as I move things around. I've automated the process within Blender as much as I can so that when a render completes, I get a folder of images with the light, depth, and individual mattes for coloured items. That way, I can link them in Photoshop so that even if I move a plant to the other side of the composition, everything updates without my input.

Conclusion

I worked on it over the course of a week, roughly 20-25 hours. One of the biggest challenges was getting the water to feel right; the difference between the appearance of translucency and a mess of lines was quite small, and it was quite a balancing act to get it right. My main takeaway from this piece was to have a bit more fun with composition. I think I often err on the side of story over artistry.

As for new artists, being deliberate is key in my eyes. The more you ask yourself why you're doing something a certain way, putting a line in a certain place, the quicker you'll learn about what does and doesn't make a satisfying image.

R.A Marsh, Illustrator

Interview conducted by Emma Collins