MITS/MI2S - Multimodal Image2story

• 

  MITS home

• 

  Input API keys

• 

  Input image, writing style, etc

• 

  Result

Inspiration

As a teenager, I really wanted to create my own stories. Fantasy stories, like most teenagers. I often read on Wattpad about fantasy stories. But I'm really bad at writing. I only understand things through visuals, I can make a big picture of something, but have difficulty giving details of something. This also applies to the writing I write. I just wanted to amaze the person I loved when in the past with a story I made up, but that will never happen. That's why I tried to make this application, at least I can make writing/stories through continuous pictures. I think it doesn't have a big impact but I just want to make my teenage dream come true. Maybe I'll try sending this application to her too. I hope she likes it.

What it does

What this application does is simple, it produces a story based on continuous images and input categories such as writing style (novel, short-story, ballad, etc), story theme (heroic story, thriller, etc), and so on. The story will continue for each image the user enters.

How we built it

How i built it was I code directly in Github, that's why there's lot commit in my project reposity. Hahaha, sorry, my laptop suck and i actually never code locally.

Challenges we ran into

Here's my story how i ran challenge from this project: I did all the code alone, except for the video creation part where my friend helped me. I also borrowed my friend's laptop to record and I also asked him to edit it. Another challenge was that I had to deal with a type of error that Generative-AI doesn't understand and there were no related answers on StackOverFlow or the Streamlit forums, this only lasted 2 days luckily. Apart from that, I also had difficulty creating complex applications the first time, but I was able to overcome this by using complete documentation and using an OOP development orientation. There is one more thing, I can't put objects to be placed outside the main file because I don't know and understand how to do it.

Accomplishments that we're proud of

The achievement I can be proud of is that I was able to create this application alone. That's also because I just want to make my dream come true since I was a teenager. Apart from that, there aren't any yet. It's possible that after the other features I've created I'll be proud of it.

What we learned

Things I learned: 1) Code documentation is important 2) If the code complexity increases, consider using an OOP or functional approach. For my code, I use OOP. 3) Don't use too much technology if you code alone, especially since I actually can't integrate databases and have never used Google Cloud Platform services.

What's next for MITS/MI2S - Multimodal Image2story

I want to make research publication (with help of my friend, i also get inspired by other publication) and continue to develop this app as last project of Software Engineering Course. Also i use 'Attribution-NonCommercial-ShareAlike 4.0 International' so anyone can develop it more than i did but cannot be commercialized.

Built With

• google-generativeai
• python
• streamlit