Youtube to Tiktok/Shorts Auto Reframer
Author:
Real discussions and feedbacks of Youtube to Tiktok/Shorts Auto Reframer
🔗 Join the Discussion on Forums
App Description
If anyone heard of Opus Clip I tried to replicate that just as a personal project, with pretty good results but definitely not perfected yet.
When I started I was using GPU but the costs were insane so I tried to optimize everything for CPU and I accomplished that on a 5$/month server I can process about 700 hours of videos a month (about 20,000 clips). In comparison that’s a 99.88% savings compared to Opus Clips (0.10$ per minute vs $0.000119 per minute),
You can check it out here
Keep in mind I was focusing mostly on the auto reframe system with computer vision, face detection, speaker detection, pose detection, motion detection, scene detection and more, all while using a low power CPU which was quite a challenge.
The other part such as finding the “viral parts” or split it up in good continuous segments is implemented but the system transcribes the video and tries to split up the video into clips of 1-2 minutes in continuity but I’m using a really bad transcription model and LLM at the moment so the results when it comes to that are just not good at all but it’s just for the demo so that will be worked on in the future.
I’m not offering it as a product but I just wanted to showcase it for now. I would love to make it open source if I know there would be collaborators to perfect the entire workflow and system so we don’t have to rely on these high paid services. I’m open to hearing from any developers interested!
I said Youtube but in reality you can use Youtube, Twitch, and 7000+ other websites and it’ll work since it uses yt-dpl in the backend.
Project Overview
This project is an innovative attempt to replicate the functionality of Opus Clip, focusing on auto-reframing videos for platforms like TikTok and Shorts using computer vision technologies. It stands out for its cost efficiency, processing videos at a fraction of the cost of competitors. However, the project is still in development, with areas like transcription and segment continuity needing improvement. The creator is considering making it open source to foster collaboration and further development.
Links
🌐 Website: https://social.looksty.com
Features & Benefits
✅ Cost-efficient
✅ Uses advanced computer vision technologies
✅ Potential for open-source collaboration
Areas for Improvement
🔄 Still in development
🔄 Needs improvement in transcription and segment continuity