Black Friday

任意套餐全年立减 50%

立即领取
eroq

Prompting

Negative prompt

A negative prompt lists what the model should avoid — text, watermarks, extra fingers, blur. It steers generation away from those features instead of hoping the main prompt implies their absence.

Negation inside a normal prompt ("no text") often does the opposite, because the model attends to the word "text". A separate negative prompt is weighted against, which is why it belongs in its own field. It is most effective on diffusion image models and on the video engines that expose it.

在 eroq 上

A field in Settings on the engines that honor it (Anima among the image models; Veo 3.1 and Kling 3.0 among the video engines), hidden where an engine ignores it. Typical: "text, watermark, extra fingers, blurry, low quality".

隐私

私密,且不受限制

  • →eroq 自有引擎没有额外的过滤,只有一份成文政策。
  • →不保留任何内容,也不用你的作品训练模型。
  • →隐私模式不留任何文件。

常见问题

What should go in a negative prompt?

Artifacts you never want: text, watermarks, anatomy errors, blur, duplicate subjects.

Do video engines accept a negative prompt?

The Veo 3.1 and Kling 3.0 engines take one natively. The API accepts negative_prompt on every video model but drops it on engines that have no such knob, and the studio hides the field there. Among the image engines, Anima takes one.

Should every render carry a negative prompt?

No. Start empty and add only the artifacts you actually see coming back, because a long blocklist pulls the model away from your subject too. Text, watermarks, extra fingers and blur are the ones worth keeping.