[mabedan]

Quite sure it's done in order to speed up the camera app performance and reduce the time to first photo time. The camera module requires some tenths of a second to boot up and it makes sense to start that process at the earliest indication of user's interaction. In this case, a touch-down is a good indication, even if user ends up swiping instead of touch-up. The same thing happens in the lock screen, if you hold your finger on the lock screen and move 1 pixel to the left, the camera module starts up even if you don't finish your swipe to camera gesture.

[dgoldstein0]

Wouldn't surprise me either. I know a guy who worked at Apple on iOS perf and the one time he was telling me about it years ago, it was "camera app doesn't start fast enough, so we reworked memory management". Apple really cares about the camera.

[gregoriol]

We should sue Apple for this: their Camera app gets an unfair advantage here compared to third-party camera apps.

[isodev]

Yup, all the gimmicks I have to do in my app to distract users from the camera loading...

[rangestransform]

No thanks, the time from locked to first capture is already too long on my 15 pro

[jgrahamc]

Yeah, makes total sense why they'd do it, but in my case it was increasing "alert fatigue" (why is my camera on?) and so I moved it.

[dfgfddfg]

I bet this is in the new version 26. That version is so garbage and I regret updating. 95% of the time, when I open the phone, it doesnt unlock my phone with face and I have to enter PIN. Sometimes I cant take photos also. In the browser, when I touch the address field nothing happens and I can go on and on and on. Just leave the shit as is, people. Its like if I have a screwdriver in my workshot and every other month, when I come back to use it, you change some bullshit, so I have to operate it slightly different. Fuck that

[reddalo]

No, I confirm that this camera behavior also happens on iOS 16. But I agree that iOS and macOS 26 are the worst thing Apple made in a long time.

[t0mas88]

Also happens on iOS 18

[andai]

I think ChatGPT has a similar feature. I was amazed how the reply starts coming in literally the moment I press enter. As far as I can tell that is only possible if all the previous tokens I submitted have already been processed. So when I actually submit the message, it only needs to update the inner state by one more token.

i.e. I think it's sending my message to the server continuously, and updating the GPU state with each token (chunk of text) that comes in.

Or maybe their set up is just that good and doesn't actually need any tricks or optimizations? Either way that's very impressive.

[port3000]

The 'flash' / no or low-thinking versions of those models are crazy fast. We often receive full response (not just first token) in less than 1 second via API.

[ibaikov]

Support systems often do this - they stream message and agents already see what you are typing. I know a few banking apps that do this.

[altbdoor]

> I think it's sending my message to the server continuously

It is, at least I see it for the first message when starting a new chat. If you open the network tools and type, you can see the text being sent to the servers on every character.

Source, from spending too much time analysing the network calls in ChatGPT to keep using mini models in a free account.

[throwawaymobule]

IIRC, apple has a patent from years ago for keeping a camera module in a semi-active mode when the phone isn't entirely idle to make starting it faster.