Free Stable Diffusion Prompt Generator
Token-weighted SDXL prompts with automatic negative prompts. Compatible with Automatic1111, ComfyUI, InvokeAI, and Forge. Free, no signup.
Generate SDXL Prompt →What makes Stable Diffusion prompts different
Stable Diffusion — especially SDXL — gives you the most technical control of any image model. Prompts use comma-separated tokens with optional weighting, and a separate negative prompt field for what you want excluded.
Positive:
(masterpiece:1.2), (highly detailed:1.1), portrait of a wizard
in a library, dramatic chiaroscuro lighting, oil painting style,
cinematic composition, 8k, sharp focus
Negative:
blurry, low quality, distorted hands, extra fingers, watermark,
text, worst quality, lowresReprompte generates both fields automatically when you select Stable Diffusion.
How to generate an SDXL prompt in 3 steps
1
Describe your idea
Enter a short concept — e.g. "a wizard in a library, dramatic light".
2
Pick SDXL + style
Reprompte outputs a weighted positive prompt and a matched negative prompt.
3
Paste into SD WebUI
Positive prompt in the main field, negative prompt in the negative field. Works with A1111, ComfyUI, InvokeAI.
SDXL settings cheat sheet
| Setting | Recommended | Notes |
|---|---|---|
| Sampler | DPM++ 2M Karras | Best balance of quality and speed for SDXL |
| Steps | 25-35 | Above 40 = diminishing returns |
| CFG Scale | 5-8 | SDXL prefers lower CFG than SD 1.5 |
| Resolution | 1024×1024 (or 1024×1536) | SDXL trained at 1024px base |
| Token weighting | (word:1.2) | Use for emphasis, avoid extreme weights >1.5 |
| LoRA format | <lora:name:weight> | Append at end of positive prompt |
Frequently asked questions
Is this Stable Diffusion prompt generator free?
100% free, no signup, no credit card, unlimited generation.
Does it generate negative prompts automatically?
Yes — each output includes a style-matched negative prompt (blurry, low quality, distorted hands, watermark, etc.).
Does it work with Automatic1111 and ComfyUI?
Yes. Token-weighted syntax (word:1.2) works identically in A1111, ComfyUI, InvokeAI, Forge, and most SD WebUIs.
What is token weighting?
(blue:1.3) boosts blue; (clouds:0.7) weakens clouds. Reprompte applies sensible weights automatically based on the style.
What CFG scale should I use?
SDXL prefers 5-8. SD 1.5 uses 7-12. Lower CFG = more creativity; higher CFG = stricter prompt adherence.
Can I use LoRAs?
Yes. Append <lora:name:weight> to the positive prompt after generation.
Related resources
Generate your first SDXL prompt
Free, no signup, no credit card. Takes under 30 seconds.
Start generating →