I've always been fascinated with Generative media, if you search for Midjourney on my X profile you'd see a lot of my experiments with it there. Now a lot of AI tools for generating videos have been released - Veo2, Sora, Kling, Runway Gen 2 etc.
The problem
I've also seen a lot of people trying to create AI generated movies but they lack one key thing - "Creative Storytelling". There are two problems with generative media.
Fragmentation - People have to use multiple models from image generation, to upscaling models then finally to video models to create a lot of things. With models like Veo2 it's gotten shorter but its still not available globally as at the time of writing this.
Prompt Based Models - Most these models are still prompt based and for storytelling to thrive you need to give back power to storytellers. People should be able to tell a story and the models do the visualization.
The solution
I decided to fix all that by building Director. A desktop app that gives you all the models in one place and gives you room to tell your story and watch the models visualize it for you. This app is built on Electron JS using Cursor AI. I also interviewed a couple of screenwriters and looked through a lot of screenwriting software to make sure the UX of this is on track.
AI Safety Features (Experimental & Future Plans)
To address potential misuse and misinformation concerns, I'm researching and working on implementing several security measures before the software's release:
First, I'm developing an experimental verification system where each generated video receives a unique hash identifier - similar to how NFTs are minted. This hash will enable easy tracking of the video's origin, with all records stored in a public ledger.
Additionally, I'm exploring two technical safeguards:
Irreversible steganographic hashing, which embeds permanent, invisible markers within the video
A proprietary file format that requires a specialized media player for playback
The custom player integration would allow social media platforms to automatically identify and label AI-generated content, promoting transparency and authenticity.
Follow my thread on X for more updates on this.