소개
우리집 반려견('소미' 포메라니안)을 AI강아지로 만들어 교육콘텐츠 영상을 브랜딩하고자 하였습니다.
진행 방법
Gemini: 물결님께서 만드신 '브랜드 IP 기획하기' gem을 이용해 프롬프트를 만들었습니다.
Chatgpt: 나에 대해 많이 학습되어 있는 Chatgpt를 통해 'IP 브랜딩 바이블'과 'System Prompt for Reuse'를 무수히 수정하여 Gemini gem을 완성했습니다.
영상의 일관성의 어려움, 대본 작성의 어려움, AI 영상 작업의 기초 실력 때문에 많은 시행착오를 겪었습니다.
대본은 유사 유튜브의 링크를 공유하여 프롬프트를 수정했습니다.
Grok: 다양한 사진 이미지와 영상 이미지를 생성해 보았습니다.
Mixboard: 영상을 편집했습니다.
Veo 3.1: Scene별 프롬프트를 통해 영상 이미지를 생성했습니다.
Capcut: Scene별 영상 이미지 편집을 진행했습니다(간단 편집).
FINAL PROMPT
A high-quality 4K video designed for YouTube Shorts.
The protagonist is Somi, a fluffy white Pomeranian with very soft, full, natural fur.
Somi interacts naturally with her owner, a Korean woman in her 40s referred to as “Mom.”
Mom’s face is never fully visible. She appears only through a side profile, back view, hands, or partial body, ensuring that Somi remains the clear visual focus at all times.
The setting is a realistic everyday life environment, such as a cozy kitchen or a sunlit dining area.
The atmosphere is warm, bright, calm, and cozy, with natural lighting and a lived-in feeling.
Somi appears intelligent and expressive, gently moving her mouth as if speaking English fluently, but always in a natural, dog-like way — not exaggerated, not human-like.
The video teaches five English expressions through natural situations and interaction, not through explanation, instruction, or on-screen text.
🐶 Today’s Topic: Dinner Time
Each scene is approximately 3–4 seconds, flowing naturally as one continuous daily moment.
1️⃣ Scene: Prepare the ingredients
Description:
In a cozy home kitchen, Somi looks up at Mom while sitting near the counter.
Kitchen utensils and a dining table are visible in the background.
Somi’s fur moves naturally with small body shifts.
Camera & Angle:
Medium close-up shot, frontal angle.
Expression & Action:
Calm, attentive expression.
Somi gently moves her mouth as if speaking naturally.
Spoken Audio (voice only, no text):
“Prepare the ingredients.”
Lighting:
Soft, warm indoor lighting, similar to natural daylight.
Video Length:
Approx. 3.5 seconds.
2️⃣ Scene: Set the table
Description:
Somi remains seated as a Western-style dinner table (soup, pasta, steak) is now clearly visible in front of her.
Camera & Angle:
Medium shot showing both Somi and the table.
Expression & Action:
A small nod and confident, calm demeanor.
Spoken Audio (voice only):
“Set the table.”
Lighting:
Bright, warm kitchen lighting.
Video Length:
Approx. 3.5 seconds.
3️⃣ Scene: Clear the table
Description:
Same setting. The camera moves slightly closer to Somi’s face.
Camera & Angle:
Close-up shot focused on Somi’s expression.
Expression & Action:
Clear articulation with focused eyes and a steady posture.
No props in paws.
Spoken Audio (voice only):
“Clear the table.”
Lighting:
Consistent warm lighting matching previous scenes.
Video Length:
Approx. 3.5 seconds.
4️⃣ Scene: Put the food away
Description:
Somi glances briefly toward the remaining food on the table, then back toward Mom.
Camera & Angle:
Medium shot, slightly low angle.
Expression & Action:
Blinking once or twice, appearing thoughtful and smart.
Spoken Audio (voice only):
“Put the food away.”
Lighting:
Clean, appetizing lighting that makes the food look fresh.
Video Length:
Approx. 3.5 seconds.
5️⃣ Scene: Do the dishes
Description:
Somi faces forward again as if concluding the moment.
Camera & Angle:
Medium frontal shot.
Expression & Action:
Bright, gentle expression.
Somi lightly lifts one paw in a natural, friendly gesture.
Spoken Audio (voice only):
“Do the dishes.”
Lighting:
Bright, warm closing light that fills the kitchen.
Video Length:
Approx. 4 seconds.🐶 Today’s Topic: Dinner Time
결과와 배운 점
Gemini: 영상에 대한 프롬프트를 수정해 달라고 하면 나노바나나 이미지를 만들어줘서 어쩔 수 없이 Chatgpt를 병행 사용했습니다.
다른 유튜브 영상의 프롬프트를 가져올 때 내가 만들고자 하는 영상과 조금이라도 불일치되는 곳이 있는 경우 영상과 이미지가 다르게 형셩되었습니다.
Grok: 50대 여성은 할머니를 생성해 주네요. & '얼굴 안 보이게'를 요청했더니 목이 잘렸습니다.
자막이 깨져서 프롬프트를 수정했습니다.
다양한 영상 편집기가 있다는 것을 알게 되었습니다.
교육 콘텐츠를 만들 때 정교한 설계가 필요하다는 것을 알게 되었습니다.
영상이 원한 대로 구체화되지 않아 프롬프트를 수정해야 할 것 같습니다.
도움 받은 글
물결님께서 만드신 '브랜드 IP 기획하기' gem
