Solid overview, but it underweights what's arguably the single biggest determinant of whether a phone camera produces a "good" photo: ISP tuning. The post frames software processing as a short list of named features (HDR, Night Sight, AI moon shots), when in reality almost every pixel has already been through dozens of tuned pipeline stages before any of those "features" run.
The ISP pipeline that turns raw Bayer data into a JPEG includes black level correction, lens shading correction, defect pixel correction, demosaic, auto white balance, color correction matrix, tone mapping, multi-stage noise reduction (raw domain, luma/chroma in YUV, temporal across frames), sharpening, local contrast, and multi-frame fusion. Each stage has dozens of parameters and most of them are scene-dependent.
"Tuning" means months in a light booth shooting color charts at every CCT from roughly 2300K to 7500K, then more months outdoors capturing skin tones across ethnicities, foliage, sky at different times of day, neon, candles, fluorescent shop lighting, mixed lighting. Every sensor + lens module gets its own calibration. Then you tune the perceptual layer on top: skin tone preservation, sky and foliage segmentation feeding AWB, AE metering weights with face priors, highlight roll-off, the chroma vs luma noise tradeoff across frequency bands.
This is why two phones with the same Sony IMX sensor look completely different. Sony ships a reference tuning. Apple and Google throw it away. Pixel phones famously have small sensors and beat phones with 2-3x the sensor area on real-world output, almost entirely on the strength of ISP tuning and computational photography stacked on top.
The headcount is genuinely large. Apple and Google each have camera organizations in the hundreds, with a substantial fraction doing nothing but tuning: color scientists, ISP tuning engineers, perceptual quality engineers running blind A/B against competitor output. It's why nobody else -- not Samsung, not Xiaomi, not the smaller players -- quite matches them even when they buy the same or better sensors. (Disclosure: I run a camera company and we live this problem.)
The sensor size advice in the article is correct but easy to misread. A small sensor with great tuning regularly beats a big sensor with mediocre tuning, and the gap is bigger than people expect.
I get this all the time. Especially since GPT5, generating an image starts a massive chain where it confirms what you want, and asks you to say yes, and you say yes, and then it confirms again, and this can go on for 5-6 times. Then if you swear at it, it refuses to continue. It is insane. Fuck you, OpenAI
honestly i find it confusing. it tells me to make a move, but i don't know if i'm playing black or white. sometimes when i click on a point, it puts the opposite color on a different point.
"Meta Ray-Ban Display glasses are designed to help you look up and stay present. With a quick glance at the in-lens display, you can accomplish everyday tasks—like checking messages, previewing photos, and collaborating with visual Meta AI prompts — all without needing to pull out your phone. It’s technology that keeps you tuned in to the world around you, not distracted from it."