File: Screenshot 2025-01-29 160405.png (36 KB, 955x628)

36 KB PNG

>>4402396 (OP)
this, in the video he says 500 is cleaner than 3-400, not 200
true for most cameras too, just at various different values
takeway is if you're minmaxing, to be aware of the breakpoints and stick to those base values as often as possible
all the more reason to avoid auto-iso too

One stop of ISO noise difference is imperceptible unless pixel peeping uncorrected raws in flat gray areas. The only time you need to worry about this is if you're underexposing to protect highlights. If your ISO is any higher than base ISO #1 when doing this, and you have to push exposure past where base ISO #2 would be, it'll look like dogshit compared to doing the same exposure push on a single gain stage sensor.

Pixel well capacity matters a lot more in everyday use. Smaller pixels retain colors less effectively with underexposure/high ISOs even though more, smaller pixels retain detail better leading to many a pixel peeper to claim high res cameras are not worse in low light (the total death of the reds and noticeable chroma noise blotches that are a bitch to edit out say otherwise). If you care about clean images, buy a camera with fewer, larger pixels and don't zoom in 200%.

>>4402398
1: ISO setting should be visible in auto ISO all the time or with a half press. changing whichever exposure parameter (ss/a/metering comp) you're willing to compromise on and using half press AEL allows indirect ISO control in auto ISO. Anyone coping about this probably has a zf (worst auto ISO controls on earth).
2: If you're shooting at base ISOs, you're protecting highlight data but deeper shadow tonality and color pushed up from underexposure actually looks slightly different than getting it right in camera for the same reason shooting at a higher ISO in camera can throw away highlight data, it's not really the exact same thing. The SNR tests are just extremely close to it being the same thing. It's best practice to compromise a little so you don't have to push the files too much. Especially on canons, and any sony raws that were shot with shading correction left on.

>>4402415
>auto ISO
>just override it
true

File: Screenshot 2025-01-29 222456.png (46 KB, 1476x778)

46 KB PNG

>>4402398
well I have no reason not to just shoot at 200 iso then

>>4402415
Is 7.2 micron considered big?

File: maxresdefault.jpg (113 KB, 1280x720)

113 KB JPG

>>4402396 (OP)
got damn did this nigga get on the ozempic or somethin? he's lookin wasted away.

>>4402423
it looks like she has a hitler mustache in the thumbnail

>>4402396 (OP)

likely fujifilm superccd

>>4402420
Does anyone even make cameras with pixels larger than 6um anymore?

>>4402445
I have absolutely zero clue. It's a 33mp ccd sensor from 2006ish.

>>4402396 (OP)
>>4402398
>>4402415
just got this video rec'd - https://www.youtube.com/watch?v=4oOWWb8dp04 [Embed]
it all makes sense but i never thought about these things this way.

>>4402451
cameramakers could solve this problem by making iso just a field in the exif rather than baking it into the raw files, and using digital gain for jpegs
then total dynamic range would be a flat line (two flat lines really), though of course you would get more headroom the higher your ISO was
but they won't because dpreview would bitch about "fake iso"

>>4402451
honestly this video is based. I'm OP and I had no idea what the fuck matt was talking about, but this guy is describing the sensor in the same way the detector in mass specs are described and I do have experience with that and how they often had dual sensors (light based specs might be the same too, honestly but I only had experience with those in college, not professionally) where in mass specs, or at least the one I used, one detector basically counted the physical number of ions hitting the sensor and the other detector was for when there were too many ions to count so it needed to convert the ions hitting into a current and then measure the current through a resister or whatever. the thing that fucking sucked about this is you had to make sure the 2 sensors agreed because they could be independently calibrated and it was fucking annoying if your data range needed to use both sensors