Brothers, what logic? Recently such a video has become a big hit, in the major platforms wild traffic, let's take a look at this account, December 24, 2024 only start number.Just sent more than 40 videos, each video 5 seconds, easily attracted 180,000 fans, get 1.2 million likes!
One of the videos has a whopping 100,000 likes! What the hell is this?
What's more interesting is that imitators soon appeared. For example, there was an account that only started posting similar content on New Year's Day, changing the scene to a high speed train with more diverse clothing styles.
There are several distinguishing features of these videos:
- They are all pictures of a young mom with one or two little girls.
- Mother and daughter are seen interacting closely, often comparing ❤ and kissing together!
- Video length is limited to 5 seconds
- The text mainly plays the warmth card, such as "mother and daughter like it", "daughter is a sweet little cotton coat, accompany you to grow up to see the world."
- The background music is selected from Jitterbug's popular songs
What's the underlying traffic code?
As simple as it may seem, these videos hit the user right in the heart - theWarm, healing, intimate.
What about cute kids and beautiful moms? Let's all find a touch of warmth and empathy in our fast-paced lives
You think these videos are just random? Wrong! The latest AI technology is used behind them! Here, I'll reveal the process of making these videos.
In this post, I'll continue to give the full range of methods, tools, and cue words directly from creativity, to static image creation, to making characters move.
And using completely free homegrown software
1. Mapping cues: the "soul" of AI
First, the drawing should have a cue word. We grabbed the image of the benchmarking account and sent it to AI for analysis. After tweaking, the following cue words were obtained:
We tune in and get the cue word:
"Headshot of a young baby mama female and a little girl in an indoor setting. The female is wearing a yellow halter dress, her hair is down and she is smiling. The child is wearing a pink hat and an orange t-shirt and looks very cute. The figure is facing the camera, some children can be seen sitting at a table in the background, and some children's paintings are hanging on the wall, making the setting look like a nursery or children's activity room. The overall atmosphere is cozy and pleasant."
We then used AI to generate 10 similar cue words, making sure that the scene, season, and clothes were different for each video, but that the characters and actions remained consistent.
Let's switch to deepseek, www.deepseek.com
Enter the prompt: "Generate 10 similar prompts, different scenes different seasons different clothes, but consistent and consistent characters and actions: headshot, a group photo of a young baby mama female embracing a little girl in an indoor setting. The woman is wearing a yellow halter dress, her hair is down and she is smiling. The child is wearing a pink hat and an orange t-shirt and looks very cute. The figure is facing the camera, some children can be seen sitting at a table in the background, and some children's paintings are hanging on the wall, making the setting look like a nursery or children's activity room. The overall atmosphere is cozy and pleasant."
deepseek gives the answer perfectly:
2. Plotting and video generation: the "magic" of AI
Open Corinne AI, go to drawing and select the Corinne 1.5 model.
Now a "Picture Quality Enhancement", and then the third step, "Generate Video".
Enter the cue word: mom and daughter look at each other and smile, then mom puts one hand and the girl's one hand together in a love gesture
And just like that, a heartwarming mother-daughter video was born!
OK, we've finished the first layer and made aShort Video
Let's do the second level below, batch generation
Drawing and mapping
With the AI tool, we can generate multiple similar videos at once, making sure that each video has different scenes and details, but the core content stays consistent with mother-daughter parent-child interactions.
Looking at the third layer below, the second account stated earlier in this article is clearly borrowed from the first one
It's not that complicated to change your face.
Open Drawable, use 1.0 as 1.5 does not support matting (uploading reference drawings)
Careful students may ask: Why do the moms and daughters in these videos all look so much alike? Are they carved from the same mold?
Exactly! There are two possibilities here: either you're using SD Unified figures, or you're using Korin's latest"AI Customized Models"
KeLing's newest "AI Customized Model" is a feature only available to Platinum members, and there was an event in December that allowed Gold members to apply for it, and it costs 1,000 points of Spiritual Power to practice one model. After customizing the model, you can use "Wensheng Video" to call the model!
However, Kerin's "AI Customized Model" has a bug! If there are multiple people in a picture, no matter male or female, young or old, their face shapes will all be the same.
But it doesn't detract from the overall effect of the video, rather it makes the parent-child video look more harmonious.
Through these technical means, we can easily produce cozy and healing mother-daughter videos, which accurately hit the pain points of users and quickly attract fans. And behind these videos is the powerful support of AI technology.
It's the male fans that attract a lot of this kind of video. Although it is said that the consumption of male fans is not as good as dogs. But it can't resist its large volume. And a video is done in half a minute.
So, old-timers, if you also want to suck up fast on Jitterbug, try these methods! Remember.Hot VideosThe underlying logic is: simplicity, warmth, empathy, plus a little AI "magic"!
