‹ All workflows
WORKFLOW · #core · mid vram

SDXL Text-to-Image

The baseline text→image graph — the clean, correct starting point every other workflow here builds on.

Result

SDXL Text-to-Image

Workflow Preview

⬇ workflow.json⬇ API format

How It Works

A minimal, correct SDXL text-to-image graph: checkpoint → dual CLIP encode → KSampler → VAE decode → save. The reference wiring for prompt, seed, steps and CFG that the rest of the library extends — and the source of the shared portrait used across these examples.