Midnight Confession by Fridge Light
Late-night kitchen, the fridge door open, its cold light on her face as she sings her feelings straight to camera. Fridge light: Rembrandt lighting on a dorm-room budget.
PROMPT · Video Prompt
Create with AI竖屏9:16视频：一名女子深夜在昏暗的厨房里打开冰箱，冰箱灯光戏剧性地照亮她的脸，她自信地唱着一首R&B歌曲，嘴型随着俏皮张扬的歌词变化，带着一股态度伸手去拿零食，身穿睡衣，厨房一片漆黑，只有冰箱的光亮，展现出令人会心一笑的深夜嘴馋日常，自嘲式的喜剧氛围
音频类型：自信的R&B/嘻哈风格歌曲（俏皮、自信、洗脑）
添加文字：“凌晨2点的我，正在为一天中第4顿饭找理由 🍕🌙”
直式9:16影片：一名女子深夜在昏暗的廚房裡打開冰箱，冰箱燈光戲劇性地照亮她的臉，她自信地唱著一首R&B歌曲，嘴型隨著俏皮張揚的歌詞變化，帶著一股態度伸手去拿零食，身穿睡衣，廚房一片漆黑，只有冰箱的光亮，展現出令人會心一笑的深夜嘴饞日常，自嘲式的喜劇氛圍
音訊類型：自信的R&B/嘻哈風格歌曲（俏皮、自信、洗腦）
加入文字：凌晨2點的我，正在為一天中第4餐找理由 🍕🌙
Vertical 9:16 video: Woman in dark kitchen opening refrigerator at night, fridge
light illuminating face dramatically, singing confident R&B song, mouth moving to
sassy lyrics, reaching for snacks with attitude, wearing pajamas, kitchen dark
except fridge glow, relatable midnight snacking behavior, self-aware comedy
energy
AUDIO TYPE: Confident R&B/hip-hop song (sassy, self-assured, catchy)
TEXT TO ADD: "Me at 2am justifying my 4th meal of the day 🍕🌙"
縦型9:16動画：暗いキッチンで夜に冷蔵庫を開ける女性、冷蔵庫の明かりが顔を劇的に照らす、自信たっぷりなR&Bソングを歌い、生意気な歌詞に合わせて口を動かす、態度たっぷりにスナックに手を伸ばす、パジャマ姿、冷蔵庫の明かり以外は真っ暗なキッチン、共感を呼ぶ深夜の間食あるある、セルフパロディ的なコメディの空気感
音声タイプ：自信満々なR&B/ヒップホップソング（生意気で、自信たっぷりで、キャッチー）
追加するテキスト：深夜2時、その日4回目の食事を正当化する私 🍕🌙
세로형 9:16 영상: 한 여성이 한밤중 어두운 주방에서 냉장고 문을 여는 장면, 냉장고 불빛이 극적으로 그녀의 얼굴을 비춤, 그녀가 자신감 있게 R&B 노래를 부름, 입모양이 장난스럽고 당당한 가사에 맞춰 움직임, 태도 있게 손을 뻗어 간식을 집음, 잠옷 차림, 주방은 냉장고 불빛만 제외하고 완전히 어두움, 공감 가는 한밤중 야식 충동을 유쾌하게 표현, 셀프 디스 코미디 분위기
오디오 유형: 자신감 넘치는 R&B/힙합 스타일 노래(장난스럽고 당당하며 중독성 있는)
추가 텍스트: “새벽 2시의 나, 오늘 네 번째 끼니의 핑계를 찾는 중 🍕🌙”
✍️ Editor’s Notes
This is a classic riff on the “opening the fridge for a late-night snack” meme format — a single light source builds that late-night, private intimacy, then a caption with its own built-in punchline gets layered on top. That trio of visual scene plus text meme plus music style is already the standard formula for this kind of vertical short-form video. Breaking AUDIO TYPE and TEXT TO ADD out into their own separate fields also makes it easier for the model to handle music generation and caption rendering as two independent tasks, instead of fishing both out of one long paragraph.
Related Works
׋![]()
›