Project Story
What Inspired Me
I’ve always enjoyed short-form video content like Instagram Reels and edits, but I found the actual editing process frustrating. Good video editing takes a lot of time and effort, and template-based tools often feel like pure luck — they either work perfectly or fail completely, with very little control in between.
Beyond that, I was motivated by the idea that everyone should be able to feel like the main character of their own story. A strong edit isn’t just about visuals or effects; it’s about narrative, emotion, and intention. Most existing AI editors focus only on producing an output, not on understanding or explaining the story behind it. I wanted to change that.
How I Built the Project
I built MIMIC as an intent-driven video editing system. Instead of relying on fixed templates, the system takes human creative intent and translates it into structured editorial decisions.
At a high level, the process works as follows:
- Interpret the user’s creative intent
- Convert that intent into a deterministic editorial plan
- Execute the edit under real-world constraints
- Generate a transparent explanation for why each decision was made
The focus was on building a system that doesn’t just edit, but also thinks about why an edit works.
What I Learned
This project taught me a lot about turning subjective creativity into structured systems. I learned how difficult it is to encode narrative, pacing, and emotion in a way that still feels human.
I also learned the importance of explainability. An AI system becomes far more useful when users can understand its reasoning, rather than treating it as a black box.
Challenges I Faced
One major challenge was avoiding randomness. Many AI editors depend heavily on probabilistic behavior or rigid templates, which can disconnect the result from the creator’s intent. Designing a system that feels intentional and consistent was difficult.
Another challenge was scope and time. I had to balance ambition with clarity, making sure the project clearly demonstrated its core idea without becoming overly complex.
Final Thoughts
MIMIC is about making creative tools more human-centered. The goal is not to replace creativity, but to support it — helping people tell meaningful stories while understanding why their edits work.
Built With
- css
- fastapi
- ffmpeg
- google-gemini-3-flash-(multimodal-vision-&-reasoning)
- librosa-(audio-analysis)
- next.js
- pydantic
- python
- react
- tailwind
- typescript
- websockets
Log in or sign up for Devpost to join the conversation.