If man were wolf scene 2

This is the second scene for my passion project to be made into a story-book.
I've build this while learning Blender back in 2020, and continued to come back to it over the last 2 years refining bits and pieces.
Even though the other illustrations will probably be worked in a more 2d medium, this is a special example of me working on a 3d still image (not realtime).

I did most of the assets myself, some are taken from quixel, or free download models (like the fox and deer) which i've rigged and posed for my purpose.
First I did the cabin as a standalone asset, modeled in blender and textured with substance painter, then the trees and terrain, props and characters, including the barely visible ladybugs, which also have been great fun to explore.
It was really important to get every detail perfect, since it was going to be a very precisely suggestive image. The tracks needed to be just right, morphing from a wolf paw shape to a human foot shape, in the right succession and just visible enough.

It's illustrations like these, I think, that show the power of working like this instead of prompting an AI to do it.
Out of curiosity, I tried my hardest to recreate the scene in Stable Diffusion, letting myself be open minded about the result, but nothing good (imo) came back. Not even close. One good thing was doing a high image influence pass (img to img in SD 1.5 with the nvinkpunk style) and getting some interesting painterly artifacts, which i've decided to incorporate into the final image for some extra funk.
Rant continues:
I think this is a good use case for AI image generators to shake things up a bit at the end, to provide a bit of mutation space and maybe add some curated cheeky extra details. Another good use-case, i think, is at the beginning of the process, as will be the case with other images of this series, to provide some inspiration for the mise-en-scene, additional elements you wouldn't think of, etc. While, for best results, I consider that the core of the work should remain obviously grounded in "traditional digital art" (funny as that sounds), I am determined to experiment with AI image generation in the future, but as a side-tool an helper, not relying on it too much.