Stable Diffusion Prompts, Tools & Guides
Open-source image generator with its own syntax: comma-separated tags, weight syntax like (subject:1.2) to emphasize terms, and a negative prompt field that actively shapes results. Community checkpoints vary widely -- a prompt tuned on one often needs adjusting for another. Can run fully offline on local hardware.
Stable Diffusion Tools
all tools →
Stable Diffusion Prompts
all prompts →
FAQ
What's the difference between the main prompt and the negative prompt?
The main prompt describes what you want; the negative prompt explicitly excludes unwanted elements (like "blurry, low quality, extra fingers") and genuinely shapes the result -- it's not just a minor cleanup step.
What does the weight syntax like (subject:1.2) actually do?
It tells the model to emphasize that specific term more heavily than the rest of the prompt. Overusing it on many terms at once dilutes the effect -- it works best reserved for the two or three terms that most need to dominate.
Why do my results look different after switching to a different checkpoint?
Different community-trained checkpoints have genuinely different strengths and aesthetic tendencies, even given the identical prompt. A prompt tuned on one often needs real adjustment, not just re-running, to get a comparable result on another.
Can Stable Diffusion run without an internet connection?
Yes -- unlike most other platforms here, it can run entirely on local hardware once installed, which is a genuine practical difference if privacy or offline access matters for your use case.