Hi guys,
With GPT4o Vision capabilities, will it be able to provide real-time feedback on people’s form during exercises?
Also when will it be released via the API and how much do you expect it to cost to use?
That may be a helpful use case! We won’t know until we get to try it. There’s literally no way to judge its accuracy and helpfulness in these areas until it’s actually out in the wild. Just make sure to have a tripod handy at that point lol.
No clue. Especially not with the real-time video streaming; out of all of the capabilities announced, that is very likely going to come last. I would suspect though it could rack up in price rather quickly.
Not for the moment at least, from the live stream, it seems like Gpt4o only takes a screenshot (photo) right when given a prompt (voice command) and analyzes it fast, this is done smoothly while keeping the camera open as to give the impression that it’s real time feed, you can notice this behaviour clearly when “Sky” was asked to describe the emotion of the prompter (can’t remember his name) but the rear camera was active at first, it said " looks like a piece of wood/table" ? but right when switched the camera and reasked, it answered correctly.
I think it makes sense for it to work like that as analyzing a feed in 30-60 fps would cost huge processing power.
Saili
would be nice if chatgpt app could use smartglass camera instead of phone camera for the live video feature. this would it make it much more useful for those with impaired eyesight. for instance if you still can see but need magnifier etc for reading a book it would be more practical to use it with smart glasses camera.