Stable Diffusion prompting without the English vocabulary test

What it does

sd-webui-oldsix-prompt is an interface plugin for Stable Diffusion WebUI, Forge, and ComfyUI that surfaces a categorized dictionary of prompts in Chinese. Users click terms to inject them into the positive or negative prompt boxes, bypassing the need to write or translate English manually. It also supports custom user dictionaries, weight tuning, and a #[option1,option2] syntax for dynamic random selection.

The interesting bit

The plugin can optionally wire into a local llama.cpp backend, using models like Qwen to translate Chinese inputs or generate random inspiration entirely offline. That turns a simple phrasebook into a self-contained prompt assistant that does not phone home.

Key highlights

• Categorized library of thousands of prompts maintained by a community group, with a dedicated yours folder for user customizations that survives updates.
• Left-click sends terms to the positive prompt box; right-click sends them to the negative prompt box.
• Built-in random prompt syntax (#[red,blue,yellow]) for slot-machine-style variation.
• Optional local LLM integration via llama.cpp for translation and inspiration generation.
• Hotkey-driven UI (Alt+Q) that toggles a floating panel without cluttering the main interface.

Caveats

• The core translation mechanism is ambiguous: the README advertises both a static, pre-categorized phrasebook and an optional local LLM for translation, without clarifying which layer handles the default Chinese-to-English mapping.
• The optional LLM path requires manual installation of llama-cpp-python wheels and downloading .gguf models, which is not exactly plug-and-play.
• All documentation and community support are in Chinese, so non-Chinese speakers are on their own.

Verdict

Worth a look if you generate images through WebUI or ComfyUI and prefer thinking in Chinese rather than guessing at English prompt vocabulary. Skip it if you are already fluent in Stable Diffusion English or need a plugin with extensive multilingual documentation.