1.22 × 1.22 μm

0.7 × 1.4 μm

I lack a clear opinion on using multiplied color filter arrays in general. I don't know exactly how big of a part is played by marketing, production process and physics. But I certainly want to make it obvious when physically the photosites are split. There's definitely a non-infinitely small area occupied by photosite walls. Therefore, if I were confident that a smaller individual pixel is something to worry about, I'd point at the fact that even if to exclude photosite walls, you're getting a slightly smaller single pixel area (0.98 μm²) on iPhone 14 Pro UWA sensor than you'd get on IMX766, a sensor with 1.0 × 1.0 μm pixels at twice the physical count. No judgement, but if you find physical pixel pitch as something to praise in sensor design, Apple's aux cams now aren't a place where you should look, they're Dual Pixel, aka 2PD, aka 100% Focus Pixels.
And who knows, maybe the main cam sensor pixels follow GN-like design too, splitting them further in two?

Nevermind, they completely skipped mentioning the actual tele, while slide I'm responding to refers to the crop of the main cam, only their site has any bits of info. So no, the actual tele sensor isn't Dual Pixel nor 1.22 μm pixel pitch. The specs suggest that it's simply the re-used previous iPhone 13 Pro Max module. The claim of 2x improvement is hard to verify, as it's unknown what it actually refers to and how it'd be measured.

Despite the seemingly absent sharpening on 14 Pro raws, the data inside them is still pre-demosaiced and denoised in a questionable way. The algorithm used clearly wants to result in absence of chroma noise, replacing it with luma that seems to offer grainier result than it could. As earlier, it's hard to say a bad word on dynamic range itself, but there's clearly no "revolution" when it comes to offering a truly raw raw format some might have expected.

Considering what Apple could easily do with their tight control over hardware and software, it's somewhat disappointing that the post processing parts which could be improved in an update simply won't be applicable to any previously shot raw photos. Way too much here is set in stone. You're getting a raw that won't be able to ever offer you way better denoising or way better demosaicing, and likely not just that. 10 bits per color channel on a stacked photo also doesn't sound too much futuristic, especially given the high probably of a higher bit depth stream coming out of sensor.

First capacitive FOD! 🥰
Credit: @ra9998vr

Despite my doubts on their remosaicing/demosaicing/superres algos, that's a surprising honesty. "Almost" in reference of 2x zoom, while by cropping that sensor to 12.5MP, they could easily claim even 4x (yes, it'd look bad without really great code handling the processing).

Best part? It's already there on your Snapdragon phone, there's no need to team up with anyone:
org.codeaurora.qcamera3.iso_exp_priority.use_iso_exp_priority

https://www.youtube.com/watch?v=4uRwN_XTfo4

You shouldn't trust sketchy app sources, but if to you I'm a particularly trustable sketchy source, here's the upcoming LOS 20 camera app I compiled just now. This build will likely just install on T.

Slight context:
The app is CameraX based. And as per the main developer, it's not planned to be extended by Camera2 interop. Current position is that resolutions listable by getHighResolutionOutputSizes will not be available, until a required change ships to CameraX. Same goes for EDGE_MODE, NOISE_REDUCTION_MODE, SHADING_MODE, manual controls and RAW format, as these are Camera2 API-only features.

For reference:
https://github.com/LineageOS/android_packages_apps_Aperture

Here's what will happen if you connect HP Elite/Pro x3Butler into a compatible dock with video output.

My new film shot entirely on my phone

Had to feature this <3