Fruits’ Journey
Ai Short Film/Animation (personal project)

Duration: 3 weeks

Program:
Image Generation: Chat GPT
Video Generation: Figma Weave

ChatGPT • Runway Gen-4.5 • Kling AI • After Effects • Premiere Pro
Postworks: Adobe After Effects 2025


Workflow: https://www.linkedin.com/feed/update/urn:li:activity:7451027823810609152/

Watch the full video



                   

Concept:  A cinematic AI short film that follows a group of fresh fruits on an unexpected adventure through New York City.

For this project, I took a deep dive into AI workflows
and explored how these tools can expand storytelling and visual creativity.




Creative Process


Chat GPT
1.   Generate the character sheets and visuals for character consistency.


Chat GPT




2.   Generate the first-frame visuals
for each shot.



Google Gemini (Nano Banana Pro) via Figma Weave
3. Bring and Enhance visuals before video generation.


Runway Gen - 4.5
Kling Ai

via Figma Weave

4. Generate videos using custom prompts



Adobe Premiere Pro5. Assemble and touch final shots
with SFX



Creative Challenges


From experimenting with motion to refining character movement and timing,
every step has been a challenging but valuable learning process.




👤 The Character Consistency Challenge

Problem: Since there are multiple characters appear, Runway and Kiling changes their faces every single shot.

Solution: Build a strong character sheet in Chat GPT and find strong reference images. Also, write more precise prompts by using prompt enhancer.



🛠️ The Tech Stack Challenge

Problem: Creating a highly stylized world often causes generative video models to frequently melt complex textures or break down during heavy structural motion.

Solution: I combined generative AI with a traditional compositing workflow. I isolated stable background plates, separated key visual elements into layers, and rebuilt the scene in Adobe After Effects. This hybrid approach allowed me to preserve the richness of the stylized environment while maintaining clean motion, visual stability, and greater creative control throughout the animation.
🤖Tool Challenge

Problem: Finding the best AI workflow to preserve my creative vision while maintaining character consistency across multiple AI tools.

Solution: Finding the best AI model wasn’t the solution but building a workflow that combines the strengths of different AI models was. Research and test each AI models and design a workflow that leveraged each model’s strengths. For this project, I primarily used Runway Gen-4.5 and Kling AI, assigning each tool a specific role:

  • Runway Gen-4.5: Used for cinematic wide shots, camera movement, and maintaining character identity across scenes.

  • Kling AI: Used for close-up shots, expressive character performances, and realistic motion such as facial expressions, body movement, hair, and cloth physics.

Rather than relying on a single model, I treated each AI as a specialized creative tool.






Back