Recently, netizens on TikTok have collectively gone crazy—the videos they come across always start with high-altitude aerial shots, and after the camera zooms in from Earth, it's either an orange cat squatting on the White House lawn giving the middle finger, or a Shiba Inu on the summit of Mount Fuji giving the middle finger...
Videos with Hollywood-level camera work like this easily get tens of millions of views, and the comments are all "my mental state is exactly the same as mine," "tutorial please, I want to make my pet turtle rebellious too."
K姐 immediately went to try it, only to find that the original higgsfield.ai only offers one free trial chance, and before you can even reproduce it, you have to pay...
So I tried other methods and found thatJimengcan get it done, and it's completely free~ Look at this result, isn't it pretty good?
It only takes three simple steps:
- Separately create images of outer space, the atmosphere, and a semi-aerial view of the ground + an image of a small animal giving the middle finger
- Use the first and last frames to generate a video
- Use CapCut to connect the videos front to back, and add BGM.
Next, K姐 will teach you step by step how to recreate this famous scene!
01. First and Last Frame Image Creation
First, enterJimeng'simage generation feature。
We first create thefirst frame image—Earth from an outer space perspective. Jimeng will quickly generate four images based on our description for us to choose from.
Prompt: Earth from an outer space perspective against a black background.
After selecting your favorite image and saving it, next open up your imagination to create thelast frame image, and enter the prompt:
Prompt: An anthropomorphic groundhog, international friendly gesture, middle finger raised, fur fluffy, standing on hind legs, wearing a floral short-sleeved shirt and white shorts, wearing slippers, one hand on the hip and the other giving the middle finger to the camera, presenting a top-down three-dimensional effect. Angry expression, gritting teeth. In a dense forest, stunning image, creative, mythological style, expressive. Masterpiece, realistic, 8K ultra HD.
Here K姐 provides a few prompt examples for everyone's reference:
Prompt: An anthropomorphic lion, an anthropomorphic fluffy hand, retaining the lion's nails and fur, middle finger raised,, one hand on the hip and the other giving the middle finger to the camera, presenting a top-down three-dimensional effect. Angry expression, gritting teeth. In a dense forest, stunning image, creative, mythological style, expressive. Masterpiece, realistic, 8K ultra HD.
Prompt: An anthropomorphic panda, an anthropomorphic fluffy hand, retaining the panda's nails and fur, middle finger raised, palm facing inward, overall presenting a top-down three-dimensional effect. Sitting on a large bluestone in a bamboo forest, still holding half a bamboo stalk in the other hand, stunning image, creative, mythological style, expressive. Masterpiece, realistic, 8K ultra HD.
Testing shows that,aside from monkeys and gorillas, other animals may require multiple attempts~
02. Transition Effect Creation
Next, open the video generation feature, and import thefirst and last frames generated in the previous step separately。
Enter the prompt, wait one minute, and you'll get the finished video.
Prompt: Fast travel from an outer space perspective to the Earth's surface, quickly pass through the clouds from a top-down view, then a vertical perspective shot appears above a vibrant forest, then quickly zoom in to a marmot standing beside a forest path covered in fallen leaves, facing the camera, with its little paw raised in a middle finger gesture toward the camera. During the process, keep the marmot's details unchanged and maintain realistic details.
But K-jie found that no matter how the prompt was changed, the middle footage was very hard to control, and there was always a missing shot of a gradual transition zooming in from the clouds to the forest.
Then I triedusing continuous first and last frames to create it, combining multiple segments, and the effect was actually pretty good!
03. Segmented production
We first create the slice from Earth to the continent. Here we need to create the ending frame image of looking down from the atmosphere.
Prompt: Vertical top-down view, with the curvature of the Earth, realistic satellite imagery style, natural colors, rich details. The perspective faces directly toward the ground, looking vertically down at the continent from high in the atmosphere. In the center of the frame, large forests are visible distributed across the surface, the terrain is clear, mountains, rivers, and woodland interweave, sparse clouds float above, no horizon.
Next, we generate the video through the first and last frames to obtain the segment from Earth to the continent.
Prompt: Vertical overhead view, the Earth rotates in space, the camera begins to zoom in extremely fast, passes through the atmosphere, and the continent comes closer and closer into view.
Once the first segment is done, we can start creatingthe slice from the continent to above the forest. As before, first create the ending frame above the forest~
Prompt: Dense forest, realistic feel, photographic feel, overhead perspective.
Next, we input the prompt to create the second slice.
Prompt: The camera begins to zoom in smoothly and extremely fast, passing through layer upon layer of clouds and mist in the atmosphere, the continent comes closer and closer into view, and finally presents a view above the forest.
Now let's create the last partthe transition from above the forest to the animal close-up。
04. Post-production editingPrompt: The camera begins to zoom in smoothly and extremely fast, passing through the gaps of the dense forest, and sees an anthropomorphic marmot standing in the forest. It stands with its hands on its hips looking at a neck brace, the scene pauses for a moment, and the marmot raises its middle finger.
After finishing the three segments, we import the videos into CapCut in order.
Connect the previously generated videos end to end, and you get a complete video.
Here K-jie teaches everyone a little trick:delete some of the frames at the junctions, then add a cross dissolve transition, and it will look more natural~
Finally, after adding background music and sound effects, export and you can see the finished video:
With the same approach,you can also replace different scenes and subjects to generate all kinds of fun Earth transition videos。
For example, a panda eating bamboo.
A fox in the desert.
05. Some sharing
In the past, this kind of shot required expensive effects technologies like VFX. Now, at zero cost, pets can cosplay rebellious teenagers.
I can only say that AI has really mastered the word "absurd." When technology allows everyone to be a director, the mental state of the whole world indeed begins to collectively run wild—after all, who wouldn't want to see their own pet give the universe the middle finger?
Overall, from the explosive popularity of these pet AI videos, I can see that the AI industry is developing in a more open and more diverse direction.
It gives ordinary people more opportunities to show themselves, and also makes the entire content creation field livelier. But at the same time, the industry also needs to continuously break through and innovate in order to go further.
Original link:Tens of millions of people watch a pet make an "international gesture"; overseas AI videos are getting more and more unhinged.