98 words to cut from your prompts, and what to write instead
98 terms across 20 failure groups. If your model has a negative field, these belong in it. If it does not — and most media models do not — cut them and state the opposite instead: 'tack sharp' rather than 'not blurry'. Free PDF at the bottom, no sign-up.
The problem
Most generation models have no negative prompt field. Type 'no blurry faces' and the model reads 'blurry faces' — you have just asked for the thing you were trying to avoid.
What this gives you
The full list of terms worth removing, grouped by the failure they were meant to prevent, with the positive phrasing that actually prevents it.
Video — Text and branding
- text
- logos
- subtitles
- watermark
- watermarks
- captions
- on-screen text
- text on screen
Video — Anatomy and faces
- extra limbs
- distorted faces
- extra characters
- hands
- extra fingers
- extra animals
Video — Content and safety
- violence
- scary imagery
- scary elements
- weapons
Video — Audio
- music
- narration
Video — Other
- dialogue
- copyrighted characters
- talking
- cuts
- horror
- humans
- clothes
- flickering
- frightening imagery
- morphing
- camera shake
- zoom
Image — Text and branding
- logo
- labels
Image — Anatomy and faces
- arms
- extra people
Image — Frame artefacts
- frame
- border
- mockup
Image — Content and safety
- weapons prominently displayed
Image — Style and clutter
- people
- props
Image — Other
- overlays
- slogan
- letters
- sunglasses
- dramatic pose
- helmet
- police department names
- cinematic color grading
- character redesign
- new clothing
- new pose
- perspective
Audio — Text and branding
- spoken words
- words
- other words
- harsh textures
- ad-libs that obscure the words
Audio — Content and safety
- scary sounds
Audio — Audio
- vocals
- lyrics
- sound effects
- speech
- spoken narration
- voices
- melody
- singing
- ad-libs
- background music
- vocal-like sounds
- prominent lead melody
Audio — Other
- medical sounds
- abrupt peaks
- dark
- dramatic tension
- heavy drums
- long instrumental passages
- sung parts
- chanting
- tense passages
- harsh timbres
- abrupt stops
Identity and faces
- identity drift
- face morphing
- inconsistent clothing
- wardrobe changes
Anatomy
- duplicate limbs
- warped anatomy
- blurry faces
Surfaces and texture
- plastic skin
- over-sharpening
- AI look
- cartoon style
Physics
- floating objects
- incorrect shadows
- unrealistic expressions
Motion over time
- random camera jumps
All 98, as a PDF
Every entry on this page, in the same order, over 1 printable page. Nothing is trimmed to make it fit — the count in the title is the count in the file.
Common questions
Do negative prompts work in AI video generators?
Only where the tool exposes a separate negative field. Where it does not, the words stay in the prompt and are read as part of the request. Check the model's own documentation before relying on one.
What should I write instead of a negative prompt?
The positive form of the same thing. Instead of 'no blur', 'tack sharp'. Instead of 'not cartoonish', name the medium you do want: '35mm photograph'.
Where this comes from, and what to read next
This page is a list. It does not explain the material — Prompt craft does, group by group, with what each decision controls and why. Research on how people actually prompt is collected under how people really use AI.