I’ve seen these cozy game ads so many times while scrolling through videos — soft morning light, gentle music, harvesting something that doesn’t exist in real life. Usually, they are building or simulation games, the kind you can lose an afternoon to.
One day, in a moment of curiosity, I thought: what if I could make one myself?
From Idea to Concept
The spark came from a simple thought: what if breakfast could grow like a flower? I imagined a small meadow at dawn, tiny flowers that looked like fried eggs, a cozy cottage in the distance, and a little wicker basket waiting to collect the harvest.
I wasn’t trying to tell a full story. I just wanted to hint at a playable loop: gather → return → cook, with the feeling of a gentle, magical morning.
My Initial Storyboard
Here’s a summary of the shots I planned for the video:
Time | Scene Description | Camera / Action | Sound |
0-2s | Misty meadow with dewdrops | Slow push-in, sunlight piercing the fog | Birds, gentle breeze |
2-4s | Fried-egg flower blooming, yolk trembling like the sun, petals curling | Macro, shallow depth of field | Soft "pop" sound |
4-7s | Hand picks fried-egg flower into a basket with toast mushrooms and bacon ferns | First-person POV, gentle motion | Rustle of picking |
7-10s | Walking toward the cottage past a milky-white stream | Follow shot, reveal world | Footsteps, flowing water |
10-13s | Inside cottage, placing flower into skillet, turning into perfect breakfast | Overhead to eye-level, warm light | Sizzle, cozy music |
13-15s | Black screen with slogan: "Today, plant a breakfast" + Game logo + Download button | Freeze-frame | Wind chime |
During iteration, I decided to remove the macro bloom shot (2-4s) and the walking-through-world shot (7-10s). They looked beautiful but slowed down the pacing and distracted from the ad’s main loop.
Anchoring the World with a Key Frame
Before generating the video, I created a single image to anchor the aesthetic and mood.
Prompt | Result |
|---|---|
|  |
This frame set the tone: soft, warm, slightly magical, with a coherent color palette. It also helped Kling 3.0 understand the “world” I wanted to build across multiple shots.
Generating the Video with Kling 3.0
For the final video, I crafted prompts for each shot, maintaining continuity and the cozy atmosphere:
Shot | Prompt Highlights |
1 | Morning meadow, soft mist, cozy cottage in distance, fried-egg flowers, small wicker basket, oil painting texture, Ghibli-inspired, golden hour light, peaceful, slightly magical |
2 | First-person POV picking fried-egg flower, basket visible, soft morning light, shallow focus, oil painting texture, Ghibli style, warm and tactile |
3 | Interior cozy kitchen, placing fried-egg flower into cast-iron skillet, transforming into breakfast, basket on rustic table, warm fire and daylight, copper pots, oil painting texture, Ghibli style |
4 | Close-up of skillet with fried-egg flower, rustic table, morning light, calm magical realism, oil painting texture, Ghibli style |
5 | UI elements appear: hand-lettered slogan "Today, plant a breakfast," rounded download button with wood-grain texture, compliance text, integrated organically into scene |
A few challenges came up:
Some prompts felt ambitious. I had too many overlapping styles: impasto brushstrokes, cinematic elements, and realistic food. It sometimes confused the model.
The last frame with UI didn’t come out exactly as I hoped. Kling 3.0 currently doesn’t allow adding a separate tail frame image, though I actually had a static image I was happy with. It’s a small regret — the ad would have been perfect with it — but it’s part of the experience.
Reflections
Even with these tweaks, the video already feels like a tiny game: pick, gather, cook, repeat.
I didn’t intend to make a full game. I just followed a single whimsical idea a bit further than usual. And in doing so, I discovered that AI tools like Kling 3.0 don’t just generate visuals — they let you explore what a world could feel like, even in 15 seconds.
Maybe that’s all a cozy game ad really needs.
If you want, I can also help make a version with embedded screenshots and prompts visually formatted so the article itself looks “interactive” and easier to read on a blog, keeping your prompts and shots inline.
Do you want me to do that next?
