Mountain - World Creator + ComfyUI

Redefining my DMP pipe with AI:

In this case, from 3D Terrain to AI-Enhanced: My World Creator → ComfyUI Pipeline

After some experimentation, I'm developing a workflow that bridges procedural 3D terrain generation and AI-powered image enhancement.

STEP 1: WORLD CREATOR
I start by building production-ready terrain in World Creator, capturing screenshot-renders from the exact angle I need. The 3D foundation gives me complete control over lighting, elevation, details, textures, and composition.

STEP 2: PHOTOBASHING
A quick Photoshop pass lets me refine specific areas and establish the core visual direction before moving into the AI pipeline.

STEP 3: COMFYUI SETUP
This is where it gets interesting. I extract prompts directly from the render itself and generate essential AOVs (Depth maps, Canny edge detection, Normal maps) to guide the AI with precision.

STEP 4: AI INPAINTING
Using different models in ComfyUI, I perform targeted inpainting on specific regions, adding atmospheric elements, vegetation, or other details while maintaining the original composition's integrity.

STEP 5: MULTI-PASS REFINEMENT
• First pass with Qwen to add micro-details
• Second pass with Flux for overall refinement
• Final upscale pass to boost resolution

This workflow is still WIP, but the results so far are promising. The key advantage? Complete control over the base geometry and composition, combined with AI's ability to add photorealistic detail and atmosphere.

The bridge between procedural 3D and generative AI is where I see the most potential right now. Traditional 3D gives you structural control that pure image generation can't match, while AI fills in the details that would take hours to model and texture manually.