Universal Negative Prompt

write a universal negative prompt for image generation

Here’s a **universal negative prompt** you can reuse across most image models (Stable Diffusion, SDXL, Midjourney-style systems, Flux, etc.). It’s designed to suppress common failure modes without being overly restrictive. --- ### 🔧 Universal Negative Prompt (Balanced) ``` low quality, worst quality, bad quality, low resolution, blurry, out of focus, pixelated, jpeg artifacts, compression artifacts, oversharpened, overexposed, underexposed, bad lighting, flat lighting, washed out colors, oversaturated, undersaturated, noise, grainy, muddy colors, bad anatomy, distorted anatomy, malformed limbs, extra limbs, missing limbs, extra fingers, fused fingers, too many fingers, long neck, twisted torso, deformed hands, poorly drawn hands, poorly drawn face, asymmetrical eyes, cross-eyed, bad proportions, unnatural pose, mutated, disfigured, cloned face, duplicate, multiple heads, extra arms, extra legs, poorly rendered hair, poorly rendered eyes, poorly rendered teeth, watermark, signature, logo, text, caption, username, artist name, border, frame, cropped, cut off, tiling, repeating pattern, 3d render, cgi, plastic skin, doll-like, waxy skin, uncanny valley, unrealistic, cartoonish (if aiming for realism) ``` --- ## 🧠 If You Want a Stronger “High-Quality Only” Version ``` low quality, worst quality, lowres, blurry, pixelated, jpeg artifacts, compression artifacts, noise, grain, bad lighting, flat lighting, oversharpened, overexposed, underexposed, washed out colors, bad anatomy, bad proportions, distorted anatomy, deformed, malformed limbs, extra limbs, missing limbs, extra fingers, fused fingers, too many fingers, long neck, twisted body, broken hands, poorly drawn hands, poorly drawn face, asymmetrical eyes, duplicate, cloned face, extra head, extra arms, extra legs, watermark, signature, logo, text, border, frame, cropped, out of frame, cgi, 3d render, plastic skin, doll, waxy, uncanny valley ``` --- ## ⚙️ Minimal Clean Version (for modern models like SDXL/Flux) If you're using newer models that don’t need huge negative prompts: ``` low quality, blurry, bad anatomy, distorted hands, extra fingers, watermark, text, logo, cropped, out of frame ``` --- If you tell me: - what model you're using (SDXL, Flux, Midjourney, etc.) - what kind of images (photorealistic portraits, anime, product shots, landscapes) I can optimize a version specifically tuned for your workflow.