Children's educationShort VideoIt's really on fire! There's a particularly fierce account that makes AI poetry starter videos withDeepSeekDoing video, only two months has increased 225,000 powder. A total of just 47 videos were posted, and the top video had 176,000 likes in just one platform, Jitterbug, and that's not even counting the data from other platforms!
Now parents love to show their children this kind of traditional cultural content, using AI to do both save time and good results. The speed of this account is like a rocket, and those who want to do the direction of children's education can follow me to dismantle this play!
▌Step 1: DeepSeek to make the picks
We use nano AI that accesses the DeepSeek-R1 model to generate a component mirror scene.
Website:https://www.1ai.net/24025.html
Turning on Deepseek's Deep Thinking and Networking Features
Enter the following prompt words directly:
I'm going to make an AI Ancient Poetry video for babies aged 0-6 years old, but I don't know which ones are for 0-6 years old, can you please help me to recommend 10 ancient poems.
Click[Sending]button, and in just over ten seconds DeepSeek gives the results.
▌Step 2: DeepSeek Generates Ancient Poetry Splits
Let's go ahead and enter the cue word:
I want to make an AI video about the ancient poem "On the Pool", help me design the sub-script for each line of the ancient poem, including scenery, dribble, text, screen, timeline, and output it as a table.
DeepSeek directly generated a table of split-screen descriptions. The table includes the timeline, the original poem, the scene, the mode of transportation, and the description of the scene, and the initial framework of the entire video content has been completed.
▌Step 3: DeepSeek Generates Painting Cue Words
Next we'll have DeepSeek generate the subplot as an AI painting cue word for easy access to the AI painting tool to generate the image.
Help me to generate the AI painting prompts for the "Pool on the Pool" ancient poem sub-scene script. Containing the subject, scene, style, quality words, each cue word at the end of the "ancient Chinese cartoon, bright colors, 3d Pixar style", cue words should be simple and easy to read, do not contain complex descriptions, do not have adjectives and lyrical parts, do not appear in the name of the person, each screen needs to appear characters, always a child, wearing a red shirt, please describe the character appearance (fixed), clothing (fixed), behavior. wearing a red shirt, please describe the character's appearance (fixed), clothing (fixed), and behavior. Then according to the descriptor to generate the picture to video cue words, cue words in accordance with the logic of what the subject to do, in the form of a table output.
Here are 4 split-screen painting cues and image-to-video cues given by DeepSeek:
Subtitle 1: A Little Baby Holds a Dinghy
Painting Prompt Words:Red-shirted boy (round face with double buns) holding Penny in both hands, wooden boat with bow cocked, water ripples, lotus leaves in background, ancient Chinese cartoon, bright colors, 3D Pixar style
Image to video cue word:The camera focuses on the action of the child holding up the boat, and the fixed camera pans to the ripples spreading at the stern of the boat.
Subtitle 2: Stealing the White Lotus
Painting Prompt Words:Red shirt child bends down to pick lotus seedpods, boat tilted, three lotus seedpods hidden under skirt, white lotus petals with water droplets, ancient Chinese cartoon, bright colors, 3D Pixar style
Image to video cue word:Panning to follow the action of the child picking the lotus, the boat rocking to the close-up of the lotus seedpods
Subtitle 3: Unexplained Hidden Trails
Painting Prompt Words:Red-shirted boy looking back, hair band fluttering, stern floater cutting a V-shaped trail, water floater densely covered, ancient Chinese cartoon, bright colors, 3D Pixar style
Image to video cue word:The camera pans down to focus on the trajectory of the duckweed, and pans out to capture the child's panicked expression.
Subtitle 4: The Floating Weed Opens Together
Painting Prompt Words:Red-shirted boy standing on the bow of a boat waving, floating green carpet extending into the distance, smoky village in vain, ancient Chinese cartoon, bright colors, 3D Pixar style
Image to video cue word:Aerial photography pulls up to show the panorama, white lines of floating weed tracks and village smoke form a deep movement
▌Step 4: Instant Dream AI generates images
That is, Dream AI daily gift 66-88 points, generate four pictures only 1 point, equivalent to free to use, is not enough to find a cell phone number to log in again.
Open the Instant Dream AI homepage:https://www.1ai.net/10005.html
Use ShakeScan or cell phone verification to log in and select the[Image Generation]
Enter the Ancient Poetry Split 1 prompt words generated in the previous step into the description field.
Raw graph modeling:Image 2.0.Fineness:10.Picture scale:9:16, click[Generate Now]
Wait for the generation to finish and then continue if there are no satisfactory images.
Click below if you are satisfied with any of the pictures[HD]button to generate a high-resolution image.
After the HD image is generated, then click above the image on the[Downloadicon to save the image locally. This image will be used as a reference image for us to generate the split screen later.
When generating the later split image, the operation is the same as generating split image 1, except that after inputting the split descriptor, you click on the bottom of the[Import Reference Chart]
Then select the reference[Characterization] - [Save
Go back to the Raw Map screen and click[Generate Now]
Select the image you are happy with and click below[HD]button to generate a high-resolution image.
This way the AI generated character images remain consistent. Follow the above instructions to generate other split images.
▌Step 5: Instant Dream AI Chart Generation Video
Image Generation Video We still do it in Instant Dream AI.
Click on our generated image and select[Generate Video], the page will jump to the Tusheng video interface.
In the video generation interface, we directly copy the image-to-video cue words generated by DeepSeek into the input box. Model Selection[Video Video S2.0 Pro]Click[Generate Video]
Wait a little while for the video of this split to finish, click on the[Downloadicon to save the video locally.
Use the same procedure to generate other split videos.
▌Step 6: Edit the video
Open Cutscene, import the produced video into Cutscene and add it into the track in order. Add subtitles, voice reading, background music and you're done.
By this point the video is complete, click on the top right corner[Export]Set the export parameters and save!
