Hi HN,
ClipCert is a personal project I built to explore a simple idea: Can we use cryptographic signing (not AI) to prove whether a video is authentic?
How it works: You digitally sign a video. Later, anyone can verify that video using your username (linked to your public key). The system gives a match percentage, showing how closely the submitted video matches what was originally signed.
It’s not detection, it’s verification!
Right now this is just a proof-of-concept (scaled down i.e. 10s max videos, .mp4 only, etc.).
I’d love your feedback — on the concept, tech, risks, and use cases. Could this add real-world value?
Feel free to test!
Thanks — happy to answer anything.
Stu
JohnFen•6mo ago
Am I understanding this right? That the feature with this that doesn't already exist with standard signing tools is the ability to quantify how much the video differs from its original form? That's an interesting idea.
I wonder about its practical utility, though. The nature of the differences matters more than the quantity. For example, a video being cropped, can be a substantial physical change without really affecting the meaning of the video. Likewise, it's easy to imagine minor physical changes that can hugely affect the meaning of the video.
stuvinton•6mo ago
Thanks for your engagement, yes your understand is right "..ability to quantify how much the video differs from its original form" is a really good way of putting it.
You are also right about the limitations; if the aspect ratio changed this would affect the verification process and would result in a much lower and likely a 0% score, I could build this to be more forgiving but I wanted it to be more sensitive, and then it would rely on someone seeing that the 0% match would be a good indicator that there was a crop or something.
stuvinton•6mo ago
How it works: You digitally sign a video. Later, anyone can verify that video using your username (linked to your public key). The system gives a match percentage, showing how closely the submitted video matches what was originally signed.
It’s not detection, it’s verification!
Right now this is just a proof-of-concept (scaled down i.e. 10s max videos, .mp4 only, etc.).
I’d love your feedback — on the concept, tech, risks, and use cases. Could this add real-world value?
Feel free to test!
Thanks — happy to answer anything. Stu
JohnFen•6mo ago
I wonder about its practical utility, though. The nature of the differences matters more than the quantity. For example, a video being cropped, can be a substantial physical change without really affecting the meaning of the video. Likewise, it's easy to imagine minor physical changes that can hugely affect the meaning of the video.
stuvinton•6mo ago