TeleComic

Inspiration

TeleComic was inspired by the desire to provide the average person, who might not have either writing or artistic expertise with an innovative solution to design their own comics using AI

What it does

TeleComic aims to enrich and visualize the user's ideas in the form of an AI generated comic book.

How we built it

We use a wide variety of programing languages and tools, such as MongoDB, Python, JavaScript, Flask, GitHub, NextJS, OpenAI image generation model (Dall-e) and Davinci-003

Challenges we ran into

At first the biggest problem was getting the correct type of images. We tested many different prompts and keywords, that heavily varied the consistency between the different comic panels. Ultimately, we ended up with a very consistent and characteristic image descriptor.

Accomplishments that we're proud of

From our initial tests, we thought that the generated images would look much worse than they do now. With rigorous testing, we managed to bring their quality to industry standard, and higher.

What we learned

We learned a lot about prompting the AI, in other words, getting it to give us what we are looking for. Other than that, we figured out how to work with APIs better and connect them to our server.

What's next for TeleComic?

We're looking forward to letting people express themselves is creative ways and to expand our horizons. As for future improvements: one thing that that could definitely be improved is the consistency of side characters, which is now only limited by the OpenAI token limit.