Lesson Content

The negative prompt field.

Leonardo AI provides a dedicated negative prompt field in the generation interface – separate from the main prompt. Negative prompts instruct the model to avoid specific visual elements, styles, or qualities in the generation. These are not embedded in the main prompt with "no" or "without" language – they go in the dedicated field for maximum effectiveness.

Writing effective negative descriptors.

Effective negative prompts are specific: "blurry, out of focus" (not "bad quality"); "watermark, text overlay, signature" (not "no text"); "extra limbs, deformed hands, anatomical errors" (not "bad body"); "overexposed, washed out, blown highlights" (not "bad lighting"). Specificity targets the actual issue rather than providing vague quality guidance.

Common negative prompt vocabulary.

Standard negative prompt elements useful across many generations: "blurry, out of focus, low quality, jpeg artifacts, watermark, text, signature, extra limbs, deformed anatomy, bad proportions, overexposed, flat lighting, amateur photography." Many experienced users maintain a standard negative prompt block that addresses common AI generation artifacts, with additions specific to the current generation goal.

When negative prompting helps most.

Negative prompting is most valuable when: the model consistently adds an unwanted element despite the positive prompt not requesting it; specific generation artifacts (anatomy issues, watermarks, blur) keep appearing; or a specific style needs to be explicitly excluded. It is less useful for fundamental composition or subject issues that require positive prompt changes.