پرامپت عکس پسرونه توی رستوران
۰ ۰ تصویر
پیشنمایش پرامپت
A copy-paste kit of master negative prompts and quality-boosting tag blocks for Stable Diffusion, Flux, and prompt-based image tools - engineered to remove common artefacts without strangling style.MASTER NEGATIVE (long, high-detail images):"lowres, worst quality, low quality, jpeg artifacts, watermark, signature, text, blurry, out of focus, deformed, disfigured, mutated, extra fingers, missing fingers, extra limbs, bad anatomy, bad proportions, duplicate, cloned face, bad eyes, bad mouth, bad hands, fused fingers, merged limbs, warped, glitch, noise, oversaturated, undersaturated, flat lighting, harsh shadows"MASTER NEGATIVE (clean/short, for style variety):"lowres, watermark, text, bad anatomy, extra fingers, blurry, deformed, oversaturated, flat"QUALITY PREFIX (add BEFORE style words):"masterpiece, best quality, extremely detailed, sharp focus, 8k, intricate details, professional lighting"STYLE STABILIZER (append to keep style consistent across a series):", consistent style, same lighting, same palette, focused composition"Per-model tweaks:- FLUX: drop "masterpiece, best quality" (they read as noise); keep only the negative artefact list, short.- SDXL: use the short negative + quality prefix with LoRA name at the end.- Midjourney: not applicable as negative prompt - instead use --no text, watermark, blurDeliverable for each model: recommended prefix/suffix/negative, a 3-line rationale, and 2 test prompts showing before/after expected effect.
