Local AI Local AI Foundations

Local Image Generation Landscape

Understand models, UIs, and hardware requirements

61 of 66

What is diffusion?

Diffusion models learn to turn random noise into images matching a text description. You provide a prompt, the model denoises step by step, and an image appears.

ToolBest forComplexity
ComfyUIControl, workflows, automationHigh
Automatic1111 / ForgeQuick prototypingMedium
FooocusSimple, good defaultsLow
InvokeAIClean UX, artistsMedium

Model families

  • SD 1.5: old, fast, lots of community models
  • SDXL: better quality, needs more VRAM
  • Flux: current open-source leader, very capable
  • Stable Diffusion 3: strong text rendering

Hardware primer

TaskMinimum VRAMComfortable VRAM
SD 1.54 GB6 GB
SDXL6 GB8 GB
Flux12 GB16+ GB

Pick your first stack

  • Want maximum control? Start with ComfyUI.
  • Want one-click good results? Start with Fooocus.
  • Want to script generation? Use ComfyUI in server mode.

Working out which model to run this on? See The Codex. Packaging it as a reusable skill? See The Armory.