// Tools

Caption Generator

Generate training-ready captions in WD14, natural language, FLUX, or dataset format. Paste your image descriptions and get formatted captions instantly.

Generate one caption at a time

Caption Style

Example Output — WD14 / Danbooru Tags

1girl, blonde hair, blue eyes, outdoor, sunlight, smile, solo, looking at viewer

Quick Templates

Click any template to add it to your input. Build complete dataset captions fast.

Character — Portrait

Character — Full Body

Style — Cinematic

Environment

Tips

→ For LoRA training, use WD14 or Training Dataset format

→ Add your trigger word to every caption so the model learns the association

→ Describe variety: different angles, lighting, and expressions = better LoRA

FLUX Optimized works best for FLUX Dev and Schnell inference

→ Aim for 15–50 images with consistent captions for character LoRAs