I remember when I had a 35mm point and shoot as a kid, anything 800 and up looked too grainy for my liking. Maybe that was just cheap film (was a kid after all), but being able to shoot at 6400 on digital and have it look ok is impressive to me when I think back to film.

I don't think its about ISO anymore. With all the denoise AI algos you can get acceptable image quality out of basically anything.

>>4326755
It's improved a lot, but its not going to improve forever.
Because light is made of those pesky photons there is a hard limit how small the exposure per pixel can be without being horribly noisy. And while AI can clean up the result to a degree at some point it will either fail or just hallucinate detail that is not there..

>>4327905
So they should focus on dynamic range is what you are saying?

>>4328404
That'd help if the reason one part of your image got so little light is because you had to balance it with another part of your image being too bright. But it'll do fuck all if everything's dark because there simply enough light. At that point it's down to aperture and shutter speed, both of which will affect the image beyond just the exposure. (Plus the former making for larger, heavier and more expensive lenses.)

High ISO performance has been and always will be a gimmick
Marginal improvements in the last decade if anything, but prior to that there was a large and noticeable improvement

If the Pixii is a sign of things to come, we're due for another leap in ISO performance with better built in ("ai") noise processing and (((computational photography)))

>>4328463
>built in ("ai") noise processing and (((computational photography)))
Be prepared to throw pictures out the window in court cases unless made by a (((trust certified))) (((always internet connected via))) (((chip to microsoft cloud))), (((microsoft verified))) camera.

The future earth is gay.

>>4327905
Sensor quantum efficiency is not at 100% yet. There ARE higher DR, more sensitive sensor technologies developed but patent trolls like applel keep buying them to never use them. They are buying them to keep them out of the competitions hands and claim that their computational photography is the way forward for dynamic range.

>>4328463
No, we are fucking not. And by WE I mean people who already know this is intentional
"Your photography is AI manipulated, therefore it will not be considered as real. Fucking fake news spreading right wing terrorist."
You know what's a sign of things to come? Sony literally does not allow you to use the C2PA feature on your camera unless you are press.

>>4328465
>everything that doesn't come through the military associated sensors is fake news
already happening in gaza btw, everything else is "hamas says"

If they can use technology to create this dichotomy wherever they want, it's joever for the free press because normies implicitly trust science and technology. A computer does not lie (except when someone makes it lie, but they can't think that far ahead)

>>4328464
brings to mind a Judge Judy episode from 20+ years ago where she wouldn't admit the guy's photos as evidence because he didn't have a date back camera / there was no timestamp lel

>>4328473
Just wait for "show me the negatives with dates" and suddenly you'll wish you bought up more snapshitty autofocus SLRs and froze more iso 1600 color film

>>4328464
>court cases
worse, court of public opinion, should you catch anything that doesn't go with the propaganda

imagine your police misconduct photos being accused of being AI and thrown out before court even convenes because corporate globerica was nice enough to free you of the shackles of... noise.
>that never happened
>the police did not shoot their dog execution style and kick their 12 year old in the head
>why are you posting fake photos
>are you a BLM protester
>probably trans too

>>4328475
hopefully they allow push processing in court or I'll be SOL

>>4326755
very minor improvements between 2000s sensors and mid 2000s, almost none since then.

there is noise improvements sooc because of cheap parlor tricks like removing luma noise in raws and improved software denoise in camera jpeg engines, otherwise there is little difference between old and new in terms of dynamic range

>>4328481
the dual gain 24mp sensor nikon uses now mogs everything they had before unless your measure of DR is "shoot everything at base ISO and PUUUUSH" (once you cross the DG level, pushing base ISO looks worse than getting it right in camera).
panasonic's g9ii reads out from both gain circuits in sequence, which is a nice trick for exposures under 2x the readout speed

dual gain models in general and leica's fancy vignetting reducing microlenses were the last great push. everything past that besides more sensor size has been patent trolled and otherwise lawyered out of the consumer still camera industry's reach - and i think this might be a part of why nikon bought RED, since that would have included their supplier contracts with semiconductor producers and RED has 17 stop DR 16 bit raw signal chains on some of their cameras.

>>4328481
>Little difference
Unless you shoot canon

Worst sensors in the industry (STILL. the forced shadow NR is definitely hiding something very ugly since they dont admit to it in marketing materials but every independent reviewer to investigate it has found smearing in the shadows that other cameras don't have - except for the pentax K1-II.)

>>4328473
I'm not american AND a zoomer so I haven't seen that episode, but damn. What a cucked world
>>4328478
Gay earth. Everyone is going to run around with smart glasses and live stream everything they do, from plowing a field to plowing her field, to twitch so they'll always have hard evidence in the court of opinion. Ranked, competitive sex and speed-runs are going to be a thing in my lifetime.

Captcha: M2XDM

File: canon+814.jpg (224 KB, 2000x1336)

224 KB JPG

>>4328488
It's gotta be on film to be real, so everyone will be banging their girlfriend with this bad boy rolling

I am used to never shooting above 800, and I am still trying to adjust to it, as nowadays it's as clear as 100 of the old days.

File: Quantum-Efficiency_image1(...).png (159 KB, 768x440)

159 KB PNG

>>4328465
Quantum efficiency of highest end scientific sensor is up to 95% depending of wavelength.
Commercial camera sensors are not quite as efficient but still there is not that much to improve before one is effectively counting individual photons.

>>4328490
god I wish, that would be so kino. Are there any onlyfans accounts that use film btw?

>>4328490
>It's gotta be on film to be real
Anon, have you ever heard of a thing called a film duplicator? One could still create AI generated content, expose it frame by frame onto a film reel and develop the seemingly "legitimate" in front of a court audience and poof, "real proof" has been forged.
>so everyone will be banging their girlfriend with this bad boy rolling
It would be GOAT kino honestly, would do it personally as well, and love the idea that we ditch computers all together and go /dune/ due to me being a CS dumdum who hates technology more and more each passing day.

But remember, always: Future Earth is (unfortunately) gay

>>4328609
>Are there any onlyfans accounts that use film btw?
OF is quick bucks, those whores don't have the money or incentives to film on real filmstock.

>>4328691
Effort required to fake digital: own computer
Effort required to fake low resolution film: own computer and high end projector, modify a film duplicator, run your own darkroom
So maybe, we shouldn't be shooting 8mm or 16mm. Just take a snap on a pentax 67.

Effort required to fake high resolution film: modify a rare and expensive 6x7 film duplicator and spend months looking for someone that will provide you with a projector capable of OUTRESOLVING it. The sampling rate of the fake must be higher than the sampling rate of the film. To my knowledge 8k projectors are hideously expensive and still do not outresolve MF film. You would need a 16k projector and i'm not sure those even exist. Maybe an 8k projector could outdo 645 with really shitty color stock.

So now we're snapping on medium format and it can only be faked by the state, and the state doesn't have to fake film (anymore). All they have to do is twist the arms of major brands to limit C2PA to employers buying their highest end cameras only, and allow (FORCED) AI processing to trickle into everything else. Which is where we're going. It's much easier to manipulate the people via the economy, and generates more tax dollars than it spends.

File: Sony-IMX249_Mako-G234_Abs(...).jpg (325 KB, 2107x1165)

325 KB JPG

>>4328601
this. That other anon thinks 100% QE is a big deal but it isn't and being locked away by Apple? They are not, and have never been at the forefront of imaging sensor tech.

Here is the QE graph for IMX294, an older m43 sensor used in the BMPCC4K and GH5s. QE is over 50% for normal sensors and their CFAs. (a bit higher than 50 for blue and the all important green and a bit under for red). So over 50 photons out of every 100 are getting captured. Getting the rest of them sounds like it would be a huge improvement, but that works out to less than one stop.

Less than two stops in some red/blue heavy situations if they can get perfect QE out of Foveon. The improvements in ISO were huge in the 10 years from say, 1998-2008, then moderate from 2008-2018, and marginal from then on as would be expected even with a linear progression in QE.

[EXIF data available. Click here to show/hide.]

Camera-Specific Properties:
Image-Specific Properties:
Image Orientation	Top, Left-Hand

>>4328490
>f/1.4 constant aperture across 7.5-60mm and close focusing
What the fuck? I thought old lenses were shit. It doesn't even look that stupidly huge. Am I missing something?

>>4328727
Smaller than 1/2.3 and not meant for pixel peeping so the lens doesn't have to be super sharp and corrected. Youtube samples look ok though