It's been a few days in theVideo NumberOn swiped a group of accounts that use AI to make ancient poetry videos, the picture below is one of the best data, just 2 months, the whole network fans more than 20W.
One of the videos that popped up had over 100,000+ likes both on the video number and on other platforms. Also opened the window to start carrying goods.
Today, we're going to break down the process of making these types of videos.
I. Explosive analysis
First of all these types of videos must be something that moms and dads will see and bookmark and then let their own kids watch.
At the same time there are a lot of 80s and 90s will also watch it, and even will follow the read through a large number of people. In a nutshell, that is our once lost youth.
Actually, a lot of similar accounts appeared in '23 and '24. From the earliest days of just using AI drawing to make PPT videos to adding AI video to make dynamic videos.
Until now, very high quality cartoons are generated through finer graphics, and character consistency, and yes, that kind ofShort VideoIt even works better than the early cartoons.
The form of content is the same, the competition is a better quality of presentation, and this is exactly what AI is good at, as AI continues to evolve at a rapid pace, the time needed to make such a beautiful video is, on the contrary, getting shorter and shorter.
II. The whole process of production
1. UseDeepSeekDismantling the split screen
We open DeepSeek, the official website card of everyone can also use other access to DeepSeek tools, there are already many.
And recently each big factory self-developed deep thinking model, the effect is also very good, basically free, you can choose to use.
The following cue words can be used:
- Help me break down "Fude Gu Yuan Cao Sending Farewell" into coherent subplots and generate AI cues for all subplot scenes (all cues need to be in Chinese form only). The cue words should be as simple as possible, but need to include the subject, scene, elements and so on. The fixed character in the scene is a little girl wearing an ancient Chinese dress. Include a detailed description of the subject (appearance, clothing, behavior, etc.). At the end of the prompt add a fixed style description: ancient Chinese cartoon, 3D Pixar style.
The above generates screen cue words for different subplots. The cue word here in Yeh requires a fixed character image, and is meant to maintain the consistency of the subject in the frame.
Here Ye fixed the style cue words: ancient Chinese cartoon, 3d Pixar style. In order to keep the consistency of the screen style.
We can use either Korin or Instant Dream for raw charts and videos.
2、Generating diagrams
Yeh here uses Korin as an example:
For the first time, you can generate just the characters individually, and then choose a favorite image of a character to be processed in HD. Then select the mat image below the picture.
After selecting the mat picture, then add in the cue words to the different scenes, actions and elements in the previous AI disassembled subplot to generate different subplot pictures.
3. Raw video
After clicking Generate Video, we will enter the interface of the graph-generated video, and we can directly fill in the prompt words.
4. Repeat the above actions
The next step is to repeat the above actions to create all the split-screen images and convert them to video one by one.
5、Cutting and editing into a movie
Import all the produced video clips one by one into the cut screen, then arrange them according to the order and input the corresponding copy. Add the soundtrack and voiceover.
Here are some singing audio, you can use AI music to make your own, or you can directly use the audio in the video on the network, download it and then in the cut screen audio separation can be.
Final Thoughts
AI production of ancient poetry tracks can be said to be enduring. The only thing that has changed is that the graphic content is more polished and the voice-over soundtrack is better.
It's really a case of whoever puts more effort into producing content is more likely to be liked by users.
Common Encouragement.
