How Automatic1111 prompts work
- Positive and negative prompts have their own boxes; both take comma-separated tags.
- Full weighting: (token:1.3) numeric, ((token)) to emphasise (~1.21×), [token] to de-emphasise (~0.91×).
- Load LoRAs inline: <lora:name:0.8>, plus any trigger words the LoRA needs in the prompt.
- BREAK starts a new 75-token chunk to keep groups of concepts from bleeding together.
Example Automatic1111 prompt
masterpiece, best quality, a knight in ornate armor, castle courtyard, (intricate engraving:1.2), <lora:add-detail:0.7>, dramatic lighting
negative: worst quality, low quality, blurry, extra fingers, deformed
Build Automatic1111 prompts faster with Prompt Builder
- Highlight a token and press Ctrl/Cmd + ↑/↓ to wrap it as (token:1.1) — no counting parentheses.
- Save LoRA tags and trigger-word combos as one-click snippets.
- Drag in an A1111 PNG to recover the exact prompt, negative and model.
- Reusable quality and negative blocks you stamp into every build.
Frequently asked questions
How do I weight a prompt in Automatic1111?▾
Use (token:1.3) for an explicit multiplier, or bare ((token)) / [token] where each layer multiplies or divides the weight by roughly 1.1.
How do I add a LoRA in Automatic1111?▾
Inline with <lora:filename:weight>, e.g. <lora:add-detail:0.8>; add any required trigger words to the prompt.
What does BREAK do in Automatic1111?▾
It ends the current 75-token CLIP chunk and starts a new one, isolating groups of concepts so they don’t blend together.
Keep exploring
- Prompt weighting & LoRA syntax guide
- Prompt glossary
- Other models: SDXL · Flux · Midjourney · Pony Diffusion · Illustrious · Stable Diffusion · ComfyUI