A local AI pipeline for
game-ready asset production.

InterForge is a Tauri 2.0 desktop application. Every stage runs on your machine — the AI models, the image generation, the mesh pipeline. Nothing is sent to a server.

01 / PROSPECT Generate

Write a prompt, choose an asset type (character, weapon, prop, creature…) and an art style. ComfyUI runs Juggernaut XL v9 entirely locally and returns your concept image. Lock it to continue.

02 / SMELT Multi-View

Takes the locked prospect image and generates four orthographic views — front, right, left, top. Background removal via rembg. Each view is locked independently before moving to Forge.

03 / FORGE 3D Mesh

The Python engine converts smelted views into a game-ready mesh. Multi-view depth merging, contour tracing, CadQuery lofting, Open3D smoothing, vertex-color GLB export, LOD0–LOD3 generation.

🎞

SPRITE PIPELINE — IN DEVELOPMENT

Frame extraction, angle tiling, background removal, sprite sheet packing, and Godot / Unity atlas export. Built alongside the mesh pipeline.

PLANNED
Tauri 2.0 Desktop shell — Rust + WebView
React + TypeScript Frontend UI
FastAPI Python backend — port 7842
ComfyUI Image generation engine — port 8188
Juggernaut XL v9 SDXL checkpoint — runs locally
rembg Background removal — Smelting stage
CadQuery Contour ring lofting — hard surface route
Open3D Point cloud + Laplacian smoothing
trimesh LOD generation + format export
imagehash + SSIM IdentityLock quality validation
SSE Streaming Real-time pipeline progress to UI
vtracer Raster → SVG tracing
Local First

Every model, every generation, every export runs on your hardware. The backend lives at localhost.

Private

No telemetry. No accounts required. No prompts or images ever leave your machine.

No Blender

The mesh pipeline is entirely headless Python. No external DCC tools required to produce a game-ready mesh.

Open Core

The pipeline is transparent. Every stage is inspectable, replaceable, and free to run.