首頁 › Stable Diffusion
Stable Diffusion Prompts
Stable Diffusion is the most popular open-weight image model family (SDXL, SD 3.5), with a huge ecosystem of LoRA, ControlNet and ComfyUI/WebUI workflows. Every prompt here includes the full English prompt, a matching negative prompt and recommended parameters (steps/cfg/sampler) you can reproduce directly in WebUI or ComfyUI.
Product 奢華香水瓶棚拍
Product 漂浮花瓣面霜罐
Product 球鞋潑濺動態拍攝
Product 奢華腕錶微距棚拍
Product 無線耳機漂浮概念
Product 汽水罐冰塊飛濺
Product 保養精華滴管瓶
Product 鑽戒珠寶微距拍攝
Product 煙霧中漂浮的智慧型手機
Product 高級包裝盒棚拍展示
Product 智慧手錶主打棚拍
Product 智慧手錶生活風格細節微距
Product 球鞋漂浮主打棚拍
Product 高級皮革球鞋棚拍
Product 無線耳機充電盒棚拍
Product 耳罩式耳機漂浮主打
Product 香水瓶水花主打
Product 黑底琥珀香水潑濺
Product 保養精華滴管棚拍
Product 精華瓶漂浮潑濺
Product 設計師太陽眼鏡主打棚拍
Product 反光表面上的太陽眼鏡
Product 無反相機主打棚拍
Product 復古相機漂浮展示
FAQ
Do negative prompts matter for Stable Diffusion?
Yes. SD models benefit greatly from negatives that exclude low quality, deformed hands and watermarks. Every prompt on this site ships with a ready-made negative prompt.
What steps and CFG should I use?
A good starting point: steps 25-35, cfg 5-8, DPM++ 2M Karras. Lower CFG looks more natural for photorealism; higher pushes style for illustration. Recommended values are listed on each prompt page.
Can SDXL and SD 3.5 share the same prompts?
As a starting point, yes. SD 3.5 understands natural language better while SDXL leans on tag-style keywords; run the same description on both and tweak by result.