Stable Diffusion Prompt Generator — Build Expert Image Prompts
PromptBro builds structured Stable Diffusion prompts — covering both positive and negative prompts, token weights, quality boosters, and model-specific syntax that actually improves your outputs.
What Makes a Great Stable Diffusion Prompt?
Stable Diffusion is unique among image generators because both your positive prompt and your negative prompt matter equally. A mediocre positive prompt with a well-crafted negative prompt will produce cleaner results than a brilliant positive prompt with no negatives. Token weights using the syntax (keyword:1.5) let you boost specific elements — values above 1.0 increase emphasis, below 1.0 reduce it. Stack weights carefully; over-weighting multiple elements causes competing artifacts. Quality boosters like "masterpiece, best quality, 8k" at the start of your positive prompt work consistently on anime and illustrative models (like Anything V5 or Counterfeit), though they matter less on photorealistic SDXL models.
SDXL handles longer, more natural language descriptions significantly better than SD 1.5 — you can write more like a sentence and less like a tag list. SD 1.5 benefits from pure comma-separated keyword prompts in the CLIP vocabulary. Negative prompts commonly include: blurry, low quality, deformed hands, extra fingers, watermark, signature, text, jpeg artifacts, overexposed. For fine-tuned models, LoRA trigger words must appear exactly as documented — check your LoRA's activation word and include it explicitly. CLIP skip settings (typically 1 for photorealistic, 2 for anime models) affect how the model interprets your prompt at the clip level.
Example Stable Diffusion Prompts
Portrait — photorealistic SDXL
Positive: RAW photo, (photorealistic:1.4), 8k uhd, portrait of a 30-year-old Japanese woman, sharp intelligent eyes, subtle smile, wearing a structured navy blazer, (bokeh background:1.2), soft window light from the left, skin texture detail, professional headshot, shot on Sony A7R IV, (85mm f/1.4 lens:1.1), natural makeup Negative: (deformed iris:1.4), (deformed pupils:1.4), semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime, text, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, bad anatomy, bad proportions, watermark, signature
Landscape — cinematic
Positive: (masterpiece:1.2), (best quality:1.2), epic cinematic landscape, ancient Scottish highland glen at dawn, mist rolling through heather-covered valley, dramatic storm clouds breaking, single shaft of golden light illuminating a ruined stone castle, (atmospheric perspective:1.3), moody color grading, Ansel Adams photography style, (fog:1.1), photorealistic, 8k Negative: blurry, low quality, overexposed, flat lighting, watermark, text, signature, people, cars, modern structures, deformed, ugly, low contrast, washed out colors, jpeg artifacts
Anime — illustrative model (Anything V5)
Positive: masterpiece, best quality, ultra-detailed, 1girl, solo, (silver hair:1.2), long flowing hair, violet eyes, wearing a traditional white miko shrine maiden outfit with red hakama, (cherry blossom petals falling:1.3), standing in a sunlit Shinto shrine courtyard, soft dappled light, (detailed fabric:1.1), anime style, clean linework, vibrant colors, Studio Ghibli aesthetic Negative: (worst quality:1.4), (low quality:1.4), (normal quality:1.4), lowres, bad anatomy, bad hands, error, missing fingers, extra digit, fewer digits, cropped, blurry, (bad proportions:1.2), watermark, text, signature, username, artist name, nsfw
Product photography — studio
Positive: (product photography:1.3), commercial studio shot, luxury skincare serum bottle, dark amber glass with gold dropper cap, (perfect specular highlights:1.2), placed on a wet black marble surface, water droplets on marble, black seamless background, (three-point studio lighting:1.1), soft box key light, rim light on bottle, macro detail on label texture, 8k, photorealistic, high-end beauty editorial Negative: blurry background on product, deformed bottle shape, bad lighting, flat, dull, low quality, reflections from camera crew, text overlay, watermark, extra objects, cluttered background, overexposed highlights, blown-out glass
Build your own Stable Diffusion prompt with PromptBro
PromptBro's guided flow helps you build both positive and negative prompts with the right token weights, quality boosters, and style tags for your specific Stable Diffusion model — whether you're using SDXL, SD 1.5, or a fine-tuned checkpoint.
Try PromptBro free — build your first prompt in 60 seconds →