Hmmm. What should draw for a dark and bloody book?

A personal work of a great book I read, Slewfoot: A Tale of Bewitchery by Brom.

From the opening chapter, you knew this book was going to be a feast of blood and tears. And between all the blood and tears was this scene, when Abitha tries to cheer up Samson with vulgar songs, being the good puritan girl she is. HA! And Samson, being the bad boy devil he is, had no idea what she’s singing about. Love it. 

Research stage

I wanted to illustrate this scene, but where to start? I thought contrast between Abitha and Samson was important. I wanted Abitha laughing so hard, you know that kind of laughter when you see it, makes you smile. And Samson who's brooding is like WTF? but can't help give a twitchy smile...

That's how I imagined it. Now to make the composition, the poses and the lighting to fit all that. No problem.

Here what my PureRef board looked like:

It's not much, but I just needed some reference for clothing, what Samson could look like (yup, there's Brom's painting there), and most importantly Abitha's expression (there it is!) That middle picture of the girl laughing with her hands. I thought that was perfect, it looks like she's trying to explain something (using her hands) but is laughing so hard, I can almost imagine her shoulder shaking.

Comp sketch in Blender

Alright, on to Blender where I created this scene:

I used Human Generator to create and pose the characters. Samson, I had to cut up and kitbash some creature models I had lying around in my hard drive. I didn't even bother to texture the legs properly. Sorry. But the texturing and rendering part will come later.

I know I went through the above part quickly, because I don't have much to say, but more than half my time is spent on the above part. I believe the work I do in Blender is the most important part because that's what will determine what the final image will look like.

So this is what I output from Blender:

It's hard to get very expressive with 3D models. But working in 2D I can push the expression more.

Rendering stage

In this stage, I focus on one part of the image or character at a time. I used Leonardo.ai, using the Leonardo Lightning XL model, image to image with a high int strength (.6-.7ish) and I used a character reference.

In Blender I have the composition down, but what should the character look like? I like to use Leonardo.ai's photo real with Kino XL model. It creates nice realistic portraits. I can generate a dozen and pick the face that feels like the character. So much easier than looking for reference images online.

Meet Abitha, our dear AI puritan influencer:

Here's what I got: