I am wondering is it possible to use the API to send a video to an Open AI model to analyse it. What I mean by that is I want, for example, to send a video of a football match and ask the model to analyse it and tell me why this team conceded, why did the other team score, what could have been made better, etc. Like give tips based on video footage. I understand that probably I have to train it before. Is it even possible?
Hi Lache. According to the last time checked it can do this via
the video gpt vision. though this is not available for gpts yet
so you would have to use the api. But yes it works and is amazing.
Currently you can only process still frames, so you could split the video up into frames and then process them that way, but video as an input is not yet a feature.
I am taking a different approach, uploading videos to YouTube and using some open sources like youtube-dl to obtain the transcript of the video. Then, from that transcript file, we can analyze the content.