Use @image1 as the exact visual reference for Gugugaga and @image2 as the exact visual reference for Doro. Preserve both characters’ faces, colors, outfits, body proportions, and plush-toy textures. Gugugaga only says variations of “gugu,” “gaga,” and “gugugaga.” Doro never speaks, stays on four legs, and behaves completely like a puppy. First-person real-phone POV, slight natural handheld shake, autofocus, realistic everyday lighting, no subtitles, no watermark, no logos, no character swapping, no humanoid Doro, no extra limbs, and no character redesign. Create a 15-second outdoor life-comedy scene in a breezy neighborhood park. Gugugaga walks along the path holding a red balloon while Doro trots beside her. A gust makes the string slip free and Gugugaga cries {guga?!}. The balloon stays just above head height as its long string trails along the path. Doro sprints ahead on all four legs, makes one small safe jump, gently catches the string in its mouth, and trots back. Gugugaga takes the string, hugs it against the white belly, rubs Doro’s head, and says {gugugaga!}. End with both continuing down the path beneath the balloon. <light breeze> <distant birds>, no music, no balloon popping, no dangerous road, and no flying character.
可直接复制的咕咕嘎嘎提示词模板
从模型更容易理解的单场景提示词开始,再替换地点、动作或结尾。制作连续角色时,请保留锁定角色外观的描述。
早餐桌上的小反应
适合测试角色一致性、细微动作、简单运镜和循环结尾。
推荐设置: Seedance 2 Fast · 9:16 · 5–10 秒 · 可选音频
Use the uploaded image as the exact character reference. Preserve the penguin silhouette, face, coat colors, beak, feet, and proportions in every frame. Place the small chibi penguin on a breakfast table beside an oversized cup. It waddles toward the cup, notices its reflection, tilts its head twice, then waves both flippers with a delighted baby-like reaction. Use one continuous medium shot with a gentle camera push-in. Keep the background simple, the action readable, and the character centered for a vertical short. End with the penguin returning to its opening pose for a clean loop. Add soft room ambience and light playful vocal babble only when the selected model supports generated audio. No subtitles, logos, extra limbs, or character redesign.
办公室从平静到混乱
用清晰的铺垫和反应结构,让企鹅保持居中,同时让背景逐渐变得混乱。
推荐设置: Seedance 2 Fast · 9:16 或 1:1 · 5–10 秒
Use the uploaded penguin reference as the exact identity guide. Preserve its face, outfit, colors, short body, and graphic chibi proportions. Begin with the penguin standing calmly at a tiny office desk. A single blank paper lands beside it, then a playful storm of blank papers fills the room. The penguin stays completely serious, slowly straightens its collar, and looks into the camera. Use a stable medium shot and one slow push-in. Keep all papers free of readable text. End on a still, loop-friendly reaction pose. No violence, no brand marks, no new main characters, no dialogue unless audio generation is enabled.
便利店循环笑点
一个无害的肢体喜剧场景,只使用一次连续运镜,并以可循环姿势结束。
推荐设置: Seedance 2 Fast · 9:16 · 5–10 秒 · 可选音频
Animate the uploaded penguin reference in a bright convenience-store aisle. Keep the same face, coat, colors, beak, feet, and proportions throughout. The penguin waddles into frame, slips harmlessly on a loose paper bag, rolls once like a soft ball, then pops upright and poses proudly. Lightweight snack bags flutter in the air without anything breaking. Track backward smoothly at a low angle with no cuts. Make the action clear in the first second and return close to the opening composition for a loop. No injuries, no readable labels, no logos, no extra limbs, and no character redesign.
咕咕嘎嘎视频推荐模型与设置
先用速度快、成本较低的设置测试动作和角色,再在效果稳定后提高质量。实际可选参数以当前模型为准。
先用 Seedance 2 Fast 测试
先确认动作、构图、参考图和声音是否正确,再投入更多积分制作最终版本。
短视频优先使用 9:16
TikTok、Reels 和 Shorts 使用竖屏;需要更多环境或第二个角色时再选 4:3 或 16:9。
每段只安排一个清晰动作
5–15 秒的聚焦场景,通常比包含多个地点、运镜和剧情转折的长提示词更稳定。
把生成音频作为可选项
模型支持音频时,可要求原创婴儿式咿呀声和简单环境音,不依赖特定受保护音源。
三步制作咕咕嘎嘎视频
三步完成一个短场景;长故事可用同一参考图分段生成。
上传首帧或参考图
首帧控制开场构图,参考图更适合保持角色外观。
描述动作并选择输出
填写动作、镜头、声音和结尾,再选择模型、时长、比例与清晰度。
确认积分并生成
查看按钮上的积分预估,完成后到“我的作品”预览和下载。
如何让咕咕嘎嘎在不同片段中保持一致
重复使用同一组参考图,并在每段提示词里写清不能改变的外观特征。每次只生成一个情节点,角色稳定后再组合片段。
重复使用相同参考图
每次都保留最清晰的咕咕嘎嘎图片;只有场景需要两个角色时才添加 Doro。
锁定关键外观特征
重复脸、服装、颜色、喙、脚、身体比例和毛绒质感。
每段只生成一个情节点
把铺垫、动作、反应和结尾拆开,不要让一次生成承担完整故事。
保持镜头要求简单
每段只使用一个景别和一次运镜,让模型把注意力放在角色和主要动作上。
咕咕嘎嘎是什么梗?
咕咕嘎嘎企鹅是一种粉丝创作的 AI 短视频迷因,以圆滚滚的企鹅小角色、日常喜剧和婴儿式“咕咕嘎嘎”叫声为特点,并在 2026 年通过短视频与社交平台传播开来。
粉丝创作的企鹅形象
这个迷因通常被认为是以《明日方舟:终末地》女性管理员为基础的 Q 版企鹅二创,并不是官方独立角色。
日常小故事加婴儿语
一个容易看懂的生活场景,加一次意外或反差,再用“咕咕嘎嘎”表达情绪。
可以不断改编
做饭、摆摊、跳舞、恶作剧和宠物互动都适合这种形式。
注意素材授权
Promptsref 与《明日方舟:终末地》、鹰角网络或 Gryphline 无关联。请使用原创或获准使用的素材。
常见问题
它可以根据图片和文字说明生成咕咕嘎嘎风格的企鹅短视频。
不是必须,但参考图有助于保持同一个角色。
写清一个角色、一个场景、一个动作、一次运镜和一个简短结尾。
支持音频的模型可以生成,但页面不提供特定的受版权保护音源。
取决于所选模型,页面只显示兼容选项。
费用由模型和设置决定,请在生成前查看按钮上的预估。
可以,但你需要确认图片、声音和角色等素材的使用权。
不是。Promptsref 与《明日方舟:终末地》、鹰角网络或 Gryphline 无关联。

