Desktop App

Comfy Labs

A whole lab of AI image, texture, and 3D workflows in one local desktop app. Pick a workflow, set a few options, and run it on your own GPU — no node graphs, no cloud, no uploads.

12+
WORKFLOWS
NVIDIA
RTX GPU
100%
LOCAL
Comfy Labs — CHIPPWALTERS.com // comfy-labs.workflows
What it does

ComfyUI power, without the graph.

Each workflow is a guided, single-screen tool. Comfy Labs picks the models and wires the pipeline for you, so you describe the result instead of building the machinery.

01

Curated workflows

Concept art, restyle, upscale, background removal, PBR textures, image-to-3D, ortho views, outpaint, inpaint, and video — each one a focused single-screen tool.

02

Runs on your GPU

Every workflow runs locally on your own NVIDIA card. Nothing is uploaded, nothing is metered, and your work never leaves the machine.

03

Built for 3D pipelines

Outputs are made to drop straight into Blender, Unreal, or Unity — transparent PNGs, GLB meshes with UVs, tileable PBR sets, and ortho reference sheets.

// comfy-labs.catalog

A modular catalog, one clean shell.

Comfy Labs is built on a plugin-based workflow architecture, so the library grows without changing the app. Each tool stays focused, and the interface stays the same.

  • Pick a workflow from the sidebar and go
  • New tools arrive as self-contained plugins
  • One application log instead of three subprocesses
// comfy-labs.local

Local by design, private by default.

Inference runs on your own hardware — no accounts, no per-image fees, no data leaving your GPU. Point it at an existing ComfyUI models folder and skip the heavy download.

  • Reuse an existing ComfyUI models folder
  • Outputs auto-clean on a 24-hour window
  • Hosts the local API behind Concept Magic

Run the lab on your own machine.

A curated, local front-end for the ComfyUI pipelines that matter — generation, restyling, upscaling, PBR texturing, and 3D mesh, all on your own GPU.