FFMPerative VS staxrip

Hello HN!

We made an update to our most popular open source project ActionAI: https://github.com/smellslikeml/ActionAI

It's a toolkit to help anyone train and deploy custom action recognition models on the edge. More here: https://www.hackster.io/actionai/actionai-custom-tracking-mu...

We made a CLI so that it's even easier to install and deploy. This update uses Intel's dlstreamer with a gvapython element to run the classifier.

If you're interested in other kinds of model customizations, check out our efforts at https://remyx.ai.

Would love to hear your thoughts.

In the spirit of making FFmpeg and video processing easier, we open-sourced a chat-based agent to assist video production - https://github.com/remyxai/FFMPerative

If you're looking for a quick way to edit/process video, we open sourced FFMPerative, a chat tool for video editing. It's easy to install and use and if you're running Ubuntu, you can install with apt-get install ffmperative

We open-sourced a command line tool for easy video editing. FFMPerative allows you to compose video edits in natural language. Simply describe your edits and it will handle the rest.

We use GPT-4/Vicuna as a video director, planning a sequence of video edits when provided with the necessary context about the video clips. Video-LLaMA and Whisper allow us to extract more context through video understanding and transcripts. With FFMPerative, we execute the described edit plan and put together a video.

Repo: https://github.com/remyxai/FFMPerative

Its literally a feature of all the underlying encoders. And you can do it even with exotic ffmpeg/vapoursynth filter chains. So I can't imagine why.

TBH I would just recommend Staxrip if you are on Windows: https://github.com/staxrip/staxrip

There is a linux equivalent, I can't remember what its called.

Or maybe some av1an GUI for Mac/Linux. All of these things support a target file size with many more features.

Since ffmpeg CLI still makes me pull ny hair out, I am going to plug vapoursynth:

https://www.vapoursynth.com/

Its Pythonic video filters... But also so much more: https://vsdb.top/

And Staxrip, which makes such good use of ffmpeg, vapoursynth, and dozens of other encoders and tools that I reboot from linux to Windows just to use it: https://github.com/staxrip/staxrip

https://github.com/staxrip/staxrip GUI But av1 outdated a bit which could be replaced/rebuild https://github.com/m-ab-s/media-autobuild_suite

2-4: I use StaxRip to encode the video because it includes the QTGMC deinterlacer, which is from my research a very good deinterlacer that produces less artifacts in the video than other deinterlacers like the one included in HandBrake.

But I actually got through it and found a great solution for myself using StaxRip and an updated QSVEnc by Rigaya. Replace StaxRip's outdated QSVEnc with the updated Rigaya QSVEnc.