Help Center
AI Baby in a Penguin Suit Just Hit 10K Likes
2026.04.02 · 238 readersWhat's up — Carat crew here. 🤗
An AI character series just crossed 10K likes, a video AI team now makes images too, and you can walk through AI worlds in real time. Five stories today.
An AI character series just crossed 10K likes, a video AI team now makes images too, and you can walk through AI worlds in real time. Five stories today.
🔥 AI Baby in a Penguin Suit Just Hit 10K Likes
Creator @PinodiArt built an AI character called 'Gugu Gaga' — a baby in a penguin outfit. The series has crossed 10K cumulative likes on X.
The storyline is pure comfort content: bringing an umbrella on a rainy day, bath time at home, warm milk in bed together. Comments are flooded with "too cute" and "I'm officially a Gugu Gaga fan."
The draw? Original character storytelling. Every video features the same character with consistent expressions and movements that feel like real animation.
The secret sauce here is character consistency. The same penguin baby shows up in every clip, and that's how you build a fanbase. You can create your own character videos on Carat using Kling's consistency feature.
📌 Three Things to Know Today
1️⃣ A Video AI Team Now Makes Images Too?
Alibaba's video AI team just dropped Wan2.7-Image — a separate model from the Qwen image AI already available on Carat. Same company, different team.
What creators are eyeing: auto color palette extraction. Feed it a reference image, and it pulls HEX codes to match the exact color scheme. Fine-grained face control and multilingual text rendering are also in. On X, some are already comparing the output to Nano Banana Pro.
2️⃣ Have You Ever Walked Inside an AI Video?
PixVerse just opened R1, its real-time world model, to everyone. Traditional AI video works like this: type a prompt, wait, get a fixed-length clip. R1 flips that. The AI world reacts in real time, and you move through it.
This update adds Shared Worlds (multiple people in the same AI environment) and custom avatars made from your photos.
OpenArt has something similar called 'Worlds,' but they're different beasts. OpenArt Worlds creates a 3D environment where you capture still images from different angles — a static tool. R1 is a dynamic experience where the video responds live. AI worlds are splitting into two lanes: 'photo set' vs 'interactive game.'
3️⃣ 'Make Me a 2004 Party Photo' — And It Did
A Reddit post asking ChatGPT to generate "a 2004 college party photo taken on a flip phone" went massive. The result nails it: noise, low resolution, red-eye flash — the whole package. Replies range from "I thought this was from an actual album" to "pure early-2000s vibes."
Recreating the look of a specific camera era with AI is becoming a thing. You can try this on Carat with ChatGPT Image.
🧪 Prompt Tip of the Day
Turn a Subway Platform Into a Movie Scene
Upload your photo with this prompt and Nano Banana 2 turns it into a cinematic shot at a subway platform with a speeding train behind you.
Create a realistic cinematic photo of the person using the face from the uploaded image. She is standing on a subway platform while a train passes behind her at high speed, creating motion blur. The woman is wearing a chic outfit. Warm cinematic lighting, shallow depth of field, 35mm film grain.
The key ingredients: motion blur gives the background train a sense of speed, while shallow depth of field keeps only you in focus. Add 35mm film grain and you get that real film camera texture.
Try this on Carat — upload your photo with Nano Banana 2 and paste the prompt.
See you tomorrow with more AI creative picks. 😊
See you tomorrow with more AI creative picks. 😊