The other day, our CEO asked me: "Hey, could you create a video with Gemini for our news feature?"
I decided to generate an AI video with Gemini to post a New Year greeting message!
Here's the finished video!
https://www.liberogic.jp/topics/20260101-news/
It captures a nice New Year atmosphere, the CEO looks just like himself, and it turned out pretty well, don't you think?
But I have to say, creating just this one video was quite a struggle.
Prompt: I need a New Year greeting video for 2026. The CEO's face (photo attached) should look like this, and please make it a 10-second video with the CEO speaking.
Video transcript (Japanese)
♪ ピアノのBGM
（背景に「ざわ...ざわ...」の文字。真ん中に社長。）
社長: みなさまさん、新年明けましておめでとうございます。
社長: 本年もどうぞよろしくお願い申し上げますね。
Since the request was just to make the background feel like New Year's and didn't specify the exact greeting lines, I realized I needed to write a more detailed prompt.
So I revised and refined the prompt, but it turns out Gemini limits video generation to 3 videos per day... oh well, I'm done for today.
So we decided to have other staff members create their own.
Video transcript (Japanese)
（社長のまわりにたくさんの従業員がいて手に飲み物を持っている）
社長: Happy New Year to everyone at Liberologic Corporation!
（従業員たちの拍手と乾杯）
Apparently the greeting ended up in English.
Unless you explicitly instruct them to "use Japanese," it can unexpectedly cause problems.
Video transcript (Japanese)
（お正月飾りの前に社長）
社長: 新年明けましておめでとうございます。
社長: 今年もよろしくお願いしますの。
Personally, I love this one the most.
"Please be nice to me" (laughs)
Video transcript (Japanese)
（金の屏風の前に並んだ従業員達と中央に社長）
新年明けましておめでとうございます。
社長: 今年もよろしくお願いしますね。
（従業員達の拍手）
This one was created by a senior colleague, but every time we generate it, his forehead gets wider (laughs)
Then the president requested that we include everyone from the team.
Video transcript (Japanese)
（お正月柄と従業員の似顔絵や写真を合わせた背景の中央に社長）
社長: 新年明けましておめでとうございます。
社長: 今年もよろしくお願いしますよー。
We loaded everyone's profile icons and created patterns with what looked like company staff members surrounding them.
In the end, everyone reached the three-generation limit...
The video we ultimately released worked smoothly once we changed the order to generate images first and then convert them to video!
Prompt: I need a main visual image for a 2026 New Year's greeting. The CEO's face (attached photo morimoto_prof.jpg) should look like this, and make the background feel like a traditional Japanese New Year! The other attached illustration (Group1.png) is a caricature of our staff, so I'd like them positioned around the edges to add liveliness.
Prompt: Based on this image, I'd like you to create a New Year's greeting video. Please create a video where the CEO in the center greets us with "Happy New Year. Thank you for your continued support this year."
Video transcript (Japanese)
♪ お正月のイメージのBGM
（背景にお正月をイメージするイラストと社員のイラスト。真ん中に社長。）
社長: 明けましておめでとうございますね。
社長: 本年もどうぞよろしくお願いいたしますね。
Finally got the video looking good!
It's quite challenging to generate exactly what you have in mind when the audio is missing, or the speech is in English, and such.
Creating detailed, specific instructions requires some technique.
Why don't you all try creating AI videos too? It's really fun!
I focus on frontend development with markup, JavaScript, React, and Next.js. I'm always happy when a site I've worked on goes live successfully! My hobbies are playing guitar, and I love cats and roasted sweet potatoes 🐱🍠
Hiraicchi
Frontend Engineer / Joined 2022