PROMPT RADAR / PROMPT产品商业
BREAKING: SpaceXAI just released Grok Imagine Image 2.0, its latest mode
BREAKING: SpaceXAI just released Grok Imagine Image 2.0, its latest model for professional image generation and editing. • Improved prompt accuracy, typography and layouts • Precise edits using Magic Wand and segmentatio
@cb_doge 2.3k 热度 4.1 万 浏览
编辑先看
它为什么值得进入灵感库
先别急着照抄整段 Prompt：真正有用的是主体、空间和镜头之间的约束关系。 产品必须先于氛围被看见；材质反射、空间尺度和品牌留白决定成片是否能用。
01 / 原始方法
完整视觉 Prompt
BREAKING: SpaceXAI just released Grok Imagine Image 2.0, its latest model for professional image generation and editing. • Improved prompt accuracy, typography and layouts • Precise edits using Magic Wand and segmentation • Background removal and smart resizing • Supports up to 5 input images • Better consistency across generations • New templates for product shots, headshots, icons, game assets and more Available now as Quality Mode on Grok for web, iOS and Android.
02 / 复刻路径
不要一次改完所有变量
用 Grok Imagine 复刻时，先固定主体和构图，再逐次替换光线、色调或材质中的一个变量，这样更容易判断哪项描述真正起作用。
- 主体：BREAKING: SpaceXAI just released Grok Imagine Image 2.0, its latest model for professional image generation and editing. • Improved prompt accuracy, typography and layouts • Precise edits using Magic Wand and segmentatio
- 视觉方向：实时 AI 视觉案例；查看原帖媒体与作者说明
- 镜头与构图：查看原帖画面；查看原帖媒体与工作流
- 交付约束：auto，保留原作者方法与归属；复刻时避免水印、低清、畸变、主体漂移和无关文字。
负向约束
保留原作者方法与归属；复刻时避免水印、低清、畸变、主体漂移和无关文字。
模型建议
Grok Imagine / GPT Image 2 / Midjourney / Flux。先用原始 Prompt 建立基准结果，再根据模型对自然语言、镜头运动或文字排版的能力做局部改写。
来源与编辑原则