Media automation, reimagined.

The old repository is simply bloated, no way to fix.

We simply create a new one instead. Write some scripts that may step into the old repository, copy some files, back and forth.

Subrepositories shall just be ignored, added to .gitignore, with some scripts that anyone can clone them back into relative places.

You may change something crucial in subrepositories, and that is important. We may apply patch over original repository.

Ask the ChatGPT on how to manage a giant repository? Had better not to commit multiple files, multiple subrepositories, big binary files, refuse to do so when such problem detected, right before things go sour, and fix it yourself.

I may integrate that warning feature into git_atomic_commit.

The time is vital for our project. If one feature is taking too long to execute, you must rewrite it or replace it, by whatever means.

Quality is equally important.

One possible application for cplex is memory management, especially for cuda memory. Can also be applied to task management, decision making, etc. Must be based on previous data and adjusted frequently.

We aim to apply what is missing in the old pyjom repository: the scoring, tracking and monitoring framework, the semantic search engine, content creation agents, critics that can make existing video more attractive and many more.

Most importantly, we aim to better assist human creators, for better prototyping, more confident trial-and-errors and researching with solid proofs, so machines can be more useful than ever, eventually integrated into the core of our content automation system.