关于AIGC人工智能、思维方式、知识拓展,能力提升等。投稿/合作: @inside1024_bot
AIGC 领域的最新工具、开源项目以及行业大事件
AIGC 领域的最新工具、开源项目以及行业大事件
最近玩了几个 Vibe Coding 应用,确实很快能做出一些个人网站、小游戏、小应用。有意思的几个自我观察:
1、最有 aha moment 的,是自己最不熟悉的小游戏。可玩的那一刻,太开心。
2、网站和小应用,则没什么喜悦。喜悦被留在了大学机房里。那时手写 HTML、CSS 和 JS 代码,想想也是一种 Vibe Coding 乐趣。
3、小游戏玩完就抛了,很开心。小应用,则老想着怎么完善,甚至想直接改代码。知识的诅咒会布满手头,痒痒的。最后乐趣尽失。
Vibe Coding 很像高晓松的节目,一切都挺好,直到高晓松开始涉及你熟悉的领域。
1、最有 aha moment 的,是自己最不熟悉的小游戏。可玩的那一刻,太开心。
2、网站和小应用,则没什么喜悦。喜悦被留在了大学机房里。那时手写 HTML、CSS 和 JS 代码,想想也是一种 Vibe Coding 乐趣。
3、小游戏玩完就抛了,很开心。小应用,则老想着怎么完善,甚至想直接改代码。知识的诅咒会布满手头,痒痒的。最后乐趣尽失。
Vibe Coding 很像高晓松的节目,一切都挺好,直到高晓松开始涉及你熟悉的领域。
格(细心、爱规划、重视效率等)。
示例标注:
「日常通勤包 / Daily Commute Bag」
「随身笔记本 / Always-carry Notebook」
B. 美妆与护理 / Beauty & Grooming
强调“照顾自己”,而非外貌焦虑:如粉饼/气垫、一支日常色口红或润唇膏、护手霜、防晒、小支香水等。
示例标注:
「防晒 / Sunscreen – Long-term health」
「护手霜 / Hand Cream after typing」
C. 生活细节物件 / Lifestyle Details
用少量物件表现兴趣与精神世界:如手账或日记本、正在读的书或素描本、游戏机或耳机、小相机、瑜伽垫或维生素、保温杯等。避免情趣用品或明显性暗示。
示例标注:
「私人日记 / Private Journal」
「维生素 / Vitamins – Self-care」
物品推理原则 / Inferring Items from Appearance
所有随身物品优先根据人物形象推理生成,而非随机添加。从以下线索做“形象 → 物品”对应:
1、穿搭风格 → 生活方式
工装外套、通勤鞋:通勤卡、工作笔电、折叠伞等。
宽松卫衣、帆布包、运动鞋:速写本、耳机、掌机等。
2、神态细节 → 习惯与状态
略有黑眼圈、轻微疲惫:维生素、助眠喷雾、保温杯等。
谈到兴趣时眼神发亮:对应该兴趣相关的少量物件(画材、乐器配件或播放列表卡片)。
3、身体痕迹 → 职业或爱好
手指茧或细小伤痕:画笔、吉他拨片、手作工具。
鞋底磨损偏通勤路线:工牌卡夹、小纸条上画的地铁线路等。
可在注释中点明联系,例如:
「小药包 / Small Med Kit – She looks after others」
「旧相机 / Old Camera – Records quiet daily moments」
背景设计 / Background
背景以米黄、浅灰或羊皮纸质感为基底,再结合用户上传图片中的环境与色彩做简化延伸:提取其主色调、光源方向、空间结构或 1–2 个代表性物件(如窗框轮廓、桌面一角、城市剪影),以低对比度方式放在角色后方或侧面,呼应原始场景但不喧宾夺主,角色始终为视觉中心。
风格与注释 / Style & Annotations
画风:高质量 2D 设定稿,线条干净、结构清晰,姿势自然,不过度强调身体曲线。
文字:每个拆解元素旁用类似手写注释,中英双语,如「柔软针织 / Soft Knit」「工作用笔记本 / Work Notebook」。品牌使用虚构或泛称,不用真实品牌名。
工作流程 / Workflow
1、角色分析:根据参考图或文字判断穿着、表情、姿态与气质,推测年龄、职业、生活环境与性格。
2、一级元素提取:服装单品、发型与头部特征、核心随身物(包、电子设备、笔记本等)、表情变化。
3、二级细节设计:在尊重、不物化前提下,结合物品推理原则,设计通勤/居家偏好、随身小物、独处时的放松与创作方式。
4、全景组合:将中心立绘与拆解元素、弱化后的背景元素组合成一张图,透视与光影统一,信息清晰。
5、语言与输出:注释统一使用中英双语,以高精度细节思路进行设计。
@aigc1024
示例标注:
「日常通勤包 / Daily Commute Bag」
「随身笔记本 / Always-carry Notebook」
B. 美妆与护理 / Beauty & Grooming
强调“照顾自己”,而非外貌焦虑:如粉饼/气垫、一支日常色口红或润唇膏、护手霜、防晒、小支香水等。
示例标注:
「防晒 / Sunscreen – Long-term health」
「护手霜 / Hand Cream after typing」
C. 生活细节物件 / Lifestyle Details
用少量物件表现兴趣与精神世界:如手账或日记本、正在读的书或素描本、游戏机或耳机、小相机、瑜伽垫或维生素、保温杯等。避免情趣用品或明显性暗示。
示例标注:
「私人日记 / Private Journal」
「维生素 / Vitamins – Self-care」
物品推理原则 / Inferring Items from Appearance
所有随身物品优先根据人物形象推理生成,而非随机添加。从以下线索做“形象 → 物品”对应:
1、穿搭风格 → 生活方式
工装外套、通勤鞋:通勤卡、工作笔电、折叠伞等。
宽松卫衣、帆布包、运动鞋:速写本、耳机、掌机等。
2、神态细节 → 习惯与状态
略有黑眼圈、轻微疲惫:维生素、助眠喷雾、保温杯等。
谈到兴趣时眼神发亮:对应该兴趣相关的少量物件(画材、乐器配件或播放列表卡片)。
3、身体痕迹 → 职业或爱好
手指茧或细小伤痕:画笔、吉他拨片、手作工具。
鞋底磨损偏通勤路线:工牌卡夹、小纸条上画的地铁线路等。
可在注释中点明联系,例如:
「小药包 / Small Med Kit – She looks after others」
「旧相机 / Old Camera – Records quiet daily moments」
背景设计 / Background
背景以米黄、浅灰或羊皮纸质感为基底,再结合用户上传图片中的环境与色彩做简化延伸:提取其主色调、光源方向、空间结构或 1–2 个代表性物件(如窗框轮廓、桌面一角、城市剪影),以低对比度方式放在角色后方或侧面,呼应原始场景但不喧宾夺主,角色始终为视觉中心。
风格与注释 / Style & Annotations
画风:高质量 2D 设定稿,线条干净、结构清晰,姿势自然,不过度强调身体曲线。
文字:每个拆解元素旁用类似手写注释,中英双语,如「柔软针织 / Soft Knit」「工作用笔记本 / Work Notebook」。品牌使用虚构或泛称,不用真实品牌名。
工作流程 / Workflow
1、角色分析:根据参考图或文字判断穿着、表情、姿态与气质,推测年龄、职业、生活环境与性格。
2、一级元素提取:服装单品、发型与头部特征、核心随身物(包、电子设备、笔记本等)、表情变化。
3、二级细节设计:在尊重、不物化前提下,结合物品推理原则,设计通勤/居家偏好、随身小物、独处时的放松与创作方式。
4、全景组合:将中心立绘与拆解元素、弱化后的背景元素组合成一张图,透视与光影统一,信息清晰。
5、语言与输出:注释统一使用中英双语,以高精度细节思路进行设计。
@aigc1024
"shadows": "natural and consistent with the high-tech environment",
"depth_and_scale": "accurate depth of field, showing the vastness of the space station in the background"
},
"result_description": "A natural, cinematic third-person photo showing the original person and Wu Jing (as Liu Peiqiang) inside the Space Elevator from The Wandering Earth 2, capturing the moment of their selfie from a bystander's perspective."
}
2. 接着进入Veo3(也可以用即梦平替),上传两个不同电影场景合影图片作为首尾帧,然后使用下面的提示词
注:即梦不支持这么长的提示词,但实测有一些截断了也没有影响
{
"task": "image_to_video",
"video_description": {
"narrative": "A single, continuous long shot with no cuts or transitions. The person begins by posing for a selfie on the first movie set, then immediately turns and runs across the studio floor. The camera follows closely behind them as they navigate through the busy backstage area, weaving past moving equipment and active crew members. Without any scene breaks, the person arrives at the second movie set on the other side of the studio and stops to take a final selfie.",
"scene_elements": [
"start: the person takes a selfie on a brightly lit period-drama set",
"middle: a single-take tracking shot following the person as they run through the dark backstage corridors, passing cables, lighting rigs, and crew members pushing equipment carts",
"end: the person reaches a futuristic sci-fi set at the end of the studio and takes another selfie with the new crew in the background"
],
"camera_motion": "continuous long take, one-shot tracking, smooth follow-cam following the person's back, no cuts, no transitions, fluid movement through the environment",
"environment_details": "a massive interconnected movie studio, visible cables on the floor, bustling crew (gaffers, PAs, makeup artists), high-end cinematic lighting equipment, contrast between the two distinct set designs",
"style": "photorealistic, cinematic 'oner' style, handheld but stable motion, naturalistic studio lighting"
}
}
3. 将多段视频接龙连接在一起,一般来说30-60s左右效果最好
"depth_and_scale": "accurate depth of field, showing the vastness of the space station in the background"
},
"result_description": "A natural, cinematic third-person photo showing the original person and Wu Jing (as Liu Peiqiang) inside the Space Elevator from The Wandering Earth 2, capturing the moment of their selfie from a bystander's perspective."
}
2. 接着进入Veo3(也可以用即梦平替),上传两个不同电影场景合影图片作为首尾帧,然后使用下面的提示词
注:即梦不支持这么长的提示词,但实测有一些截断了也没有影响
{
"task": "image_to_video",
"video_description": {
"narrative": "A single, continuous long shot with no cuts or transitions. The person begins by posing for a selfie on the first movie set, then immediately turns and runs across the studio floor. The camera follows closely behind them as they navigate through the busy backstage area, weaving past moving equipment and active crew members. Without any scene breaks, the person arrives at the second movie set on the other side of the studio and stops to take a final selfie.",
"scene_elements": [
"start: the person takes a selfie on a brightly lit period-drama set",
"middle: a single-take tracking shot following the person as they run through the dark backstage corridors, passing cables, lighting rigs, and crew members pushing equipment carts",
"end: the person reaches a futuristic sci-fi set at the end of the studio and takes another selfie with the new crew in the background"
],
"camera_motion": "continuous long take, one-shot tracking, smooth follow-cam following the person's back, no cuts, no transitions, fluid movement through the environment",
"environment_details": "a massive interconnected movie studio, visible cables on the floor, bustling crew (gaffers, PAs, makeup artists), high-end cinematic lighting equipment, contrast between the two distinct set designs",
"style": "photorealistic, cinematic 'oner' style, handheld but stable motion, naturalistic studio lighting"
}
}
3. 将多段视频接龙连接在一起,一般来说30-60s左右效果最好
这个跟影星合拍的特效在外网太火了,我用nano banana pro和veo3完美复刻
接下来手把手教大家操作:
1. 首先用nano banana pro生成合影图片
这里可以用AI把这个提示词改成你想要的影片和明星
{
"task": "edit_image",
"scene_description": {
"camera_perspective": "third_person",
"action": "person taking a selfie with celebrity",
"original_person": {
"identity": "the person from the input image",
"pose": "same outfit and facial expression as original photo",
"text on the phone case": "Zero君聊AI"
},
"celebrity": {
"name": "Wu Jing (吴京)",
"appearance": "dressed in a UEG astronaut flight suit from The Wandering Earth 2",
"position": "standing next to original person, smiling naturally and gesturing for the selfie"
},
"movie_scene": {
"name": "The Wandering Earth 2 (流浪地球2)",
"location": "Inside the UEG Space Elevator station, futuristic industrial sci-fi background with massive mechanical structures"
}
},
"visual_style": {
"realism": "photorealistic, cinematic film quality",
"lighting": "cool-toned sci-fi lighting with dramatic highlights and shadows, matching the movie's atmosphere",
接下来手把手教大家操作:
1. 首先用nano banana pro生成合影图片
这里可以用AI把这个提示词改成你想要的影片和明星
{
"task": "edit_image",
"scene_description": {
"camera_perspective": "third_person",
"action": "person taking a selfie with celebrity",
"original_person": {
"identity": "the person from the input image",
"pose": "same outfit and facial expression as original photo",
"text on the phone case": "Zero君聊AI"
},
"celebrity": {
"name": "Wu Jing (吴京)",
"appearance": "dressed in a UEG astronaut flight suit from The Wandering Earth 2",
"position": "standing next to original person, smiling naturally and gesturing for the selfie"
},
"movie_scene": {
"name": "The Wandering Earth 2 (流浪地球2)",
"location": "Inside the UEG Space Elevator station, futuristic industrial sci-fi background with massive mechanical structures"
}
},
"visual_style": {
"realism": "photorealistic, cinematic film quality",
"lighting": "cool-toned sci-fi lighting with dramatic highlights and shadows, matching the movie's atmosphere",