{
  "type": "tutorial",
  "pillar": "usando_errado",
  "title": "Por que seu prompt de imagem falha (e como consertar)",
  "image_urls": [
    "https://cdn.borav.com.br/tut-20260703-1200/cards/01-cover.png",
    "https://cdn.borav.com.br/tut-20260703-1200/cards/02-step.png",
    "https://cdn.borav.com.br/tut-20260703-1200/cards/03-step.png",
    "https://cdn.borav.com.br/tut-20260703-1200/cards/04-step.png",
    "https://cdn.borav.com.br/tut-20260703-1200/cards/05-step.png",
    "https://cdn.borav.com.br/tut-20260703-1200/cards/06-cta.png"
  ],
  "caption_ptbr": "O problema quase nunca é o AI. É o prompt. Três erros matam o resultado antes de gerar qualquer imagem: prompt vago, estilo sem descrição e instruções que se contradizem. A boa notícia: a correção é simples e você aplica agora. Qual desses erros você já cometeu? Comenta o número: 1, 2 ou 3.",
  "en_full": "The problem is almost never the AI. It's the prompt.\n\nMost people get disappointing results from image generators like Midjourney, DALL-E, or ChatGPT because of three fixable mistakes.\n\nMistake 1: Vague subject. 'A photo of a person at the office' gives the AI nothing to work with. Try instead: 'Woman in her 30s, white shirt, laptop open, wooden desk, natural light from the window.' Specifics produce results.\n\nMistake 2: Style label with no description. 'Cyberpunk style' means almost anything. The AI guesses. Describe what you actually see in your head: 'Purple and blue neon, rain on the street, Japanese signs in the background, high contrast, nighttime photography.'\n\nMistake 3: Conflicting instructions. 'Realistic photo in watercolor with flat illustration style' asks the AI to do three incompatible things at once. Pick one visual direction and commit to it.\n\nThe structure that works: subject + what it's doing + setting + lighting + style. Example: 'Orange cat sleeping on a balcony, afternoon light, shot on 35mm film.'\n\nSame AI. Better prompt. Completely different result.\n\nWhich of these three mistakes have you made? Drop the number in the comments: 1, 2, or 3.",
  "hashtags": [
    "#ia",
    "#inteligenciaartificial",
    "#promptdeimagemAI"
  ],
  "image_keyword": "AI image generation prompt",
  "source_url": "https://aiunpacker.com/blog/10-ai-image-generation-mistakes-99-of-people-make-and-how-to-fix-them"
}