While writing a previous blog, I really enjoyed making AI-generated videos, so I decided to write a blog about how to create one.
If you want to read my opinion about AI-generated video, please go to the previous blog.
Here is the link: The AI Creativity Paradox: Why Humans Still Matter
To do this, I am going to use ChatGPT and Gemini (Veo).
First of all, open Gemini (I'm repeating it for those who are reading this blog only. If you already read my previous blog, then you can skip this part! Go to step 2).
Step 1: Create an image first
Describe what you want to create. For example, I asked them a prompt:
Italian Greyhound puppy, one ear has a white pattern, front shot.
Warning! If you don't provide specific guidelines, you can't get what you want.
Step 2: Download your image
Download your image and keep it. I always rename it so I can find it easily.
Step 3: Ask ChatGPT for a prompt
This is a very important part. Now open ChatGPT and ask it to make a prompt. I wrote it this way:
“Hi everyone! My name is Kkangsik, and I really love basketball. The name ‘Kkangsik’ is Korean—I chose it because I wanted it to sound strong.
I turned on the camera today because I have some big news. Did you know that in 2026, Toronto is getting a brand-new WNBA team? The team’s name is Toronto Tempo!
I live in Ottawa, so I’m already getting ready to buy tickets. But here’s something even more exciting—In July 2026, two WNBA games will be held in Montreal!
WNBA games just two hours away? There’s no way I’m missing that. I’m so excited and counting down to July. I’ll be back with more fun news next time!”
We made a good prompt, but I found a problem. They are too long to put in Veo, so I had to shrink them. Veo only gave me an 8-second video. So I had to change the narration shortly. Here is the version 2 prompt that I asked ChatGPT.
ChatGPT prompt version 2
 |
| Veo(flow) page |
Go to Veo(flow) Google AI and click the + button to add a new project.
Step 5: Select Assets mode
Click the down arrow of 'convert text to video' to create a video with assets.
Step 6: Generate Video
Add the image you created in the first step, copy the prompt from ChatGPT here, and then click the right arrow. That's all. Now Google AI makes a video based on your prompt!
After this prompt, I got an awesome video, but I found another problem! Everything looked clear before I saw HUMAN'S HAND!!!
Yep, here is an important point: YOU HAVE TO GIVE A SPECIFIC PROMPT.
Now we learn we need a better prompt. I got an action prompt: "At the end, the dog sits still, tilts its head, and gently wags its tail. No waving. No gestures."
And in the last sentence, I added: "No human hands, no human arms, no human gestures, no human body parts. Dog only."
Step 7: Modify if needed
If you encounter any problems or don't like your video, you need to modify it. Hover over the video, click the correction button at the top, and add the details.
Step 8: Download
If you like it, hover over the video again and download it. That's all! But remember, there are limitations to generating AI videos.
How were the steps? I think you can make a fun video right away! To me, it was a really fun experience to make a video in 10 minutes. I learned that I really need to know how to write a good prompt, because the results will vary widely depending on it.
Thank you for reading! And here is my final result :) Hope you guys like it.
Comments
Post a Comment