About the Project

When AI became trendy and I was experimenting with software like Stable Diffusion, the idea for a website similar to Google Images came to mind. Instead of a generic image search, I chose anime as the basis. The planned workflow is simple:

1. A user uploads an anime episode.
2. AniPic verifies if it's an anime.
3. AniPic breaks the episode into frames.
4. Each frame is individually categorized by AI.
5. The categorized frames are then searchable in the search engine.

The core of this project is the categorization AI. I began collecting data and, if memory serves, ended up with about 4TB of episodes. I broke these into frames using a script. However, I noticed I had produced too many redundant frames due to a faulty FFMPEG configuration.

As of today (status: October 26, 2023), heaps of these frames are lying unused on my server. I plan to complete the project at some point. The next step would be the categorization of the frames, either manually or with an existing AI like ChatGPT, for example.

I've thoroughly planned the project and have also considered legal aspects, as the use of anime frames is tricky from a licensing standpoint. The project could be useful both for enthusiasts and content creators looking for specific images for their videos.

One idea for monetization is introducing premium features. However, I doubt the revenue would cover the expenses for categorization, etc. But if enough capital is available, it could become a fascinating project for the anime community.

Planning

The planning for the project is quite comprehensive and includes flowcharts that I've created with Draw.io:

For server communication: