• Members 140 posts
    April 17, 2023, 1:06 p.m.

    Definitely, if you are shooting JPEG, the ISO setting is essential. We agree on this, yes?

    If you are shooting RAW, the ISO setting is necessary unless your sensor is ISO invariant. I would say that the definition of an ISO invariant sensor is one in which you can get the exact same result using you RAW editor’s Brightness or “Exposure” slider as you can by you can by adjusting ISO in the camera.

    I tried using Canon’s DPPro software to adjust the brightness in my RAW images to equal the effect of the camera’s ISO setting, but it did not work. I took fixed manual exposure shots in daylight, using a tripod and adjusting ISO. The ISO 800+ shots were blown out. Unrecoverable.

    You may be correct in theory; I have no idea. But in practice, on the Canon platform, the ISO setting definitely matters for both JPEG and RAW.

  • Members 140 posts
    April 17, 2023, 1:15 p.m.

    This is a situation where ISO doesn’t matter, because presumably you control your lighting for an ISO 100 anyway, as well as the shutter and Aperture you want, correct?

    In my view, the photographer still needs to account for the ISO of 100, even though it’s being held constant. It seems like we should just move off the ISO discussion and talk about the paper and the lighting setup, which is much more interesting.

  • Members 976 posts
    April 17, 2023, 1:21 p.m.

    Or they can account for the outcome.

  • April 17, 2023, 2:05 p.m.

    Probably good advice back then.

    David

  • Members 976 posts
    April 17, 2023, 2:35 p.m.

    Yes, and nothing like

  • April 17, 2023, 3:22 p.m.

    Which in any case wasn't ever a possibility - based on the incorrect idea that an image sensor behaves like an RF diode detector.

  • Members 50 posts
    April 17, 2023, 3:26 p.m.

    Some people (not me, fgs!) are doing OOC JPEG output. They like to have a certain lightness on their images, depending on the scene. For them, it is important to understand how ISO, F-stop and shutter speed interact. Call it a triangle or anything, but you need to understand the three things. I prefer a more practical approach.

  • April 17, 2023, 3:30 p.m.

    Let me step in where angels fear to tread. My background is in Music, Audio, and Electronics from dc to microwaves.

    As I see it, one can increase the number of photons entering the camera by

    a) opening the aperture wider OR
    b) opening the shutter for longer.

    If these are done to excess, a) has also has the effect of reducing depth of field, b) has the consequence of blurring a moving image. These two concomitant effects set the maximum number of photons that the photographer allows to enter the camera during a given shot.

    The photons are passed on to the sensor and if there are not enough of them to fill it, the resultant photo will be “too dark”. Given that we have optimised the settings of time and aperture, as above, the only way to brighten the image is to apply analog gain before the sensor, or digital gain after the sensor.** As I understand it, increasing the ISO setting will add one or both of these types of gain. An analog amplifier is going to add some noise and this may show in the photo, if it is higher than the noise coming in through the lens. Whether digital multiplication adds noise, I do not know, but it presumably amplifies the quantization “noise” of the sensor.

    ** conversely, we have clipping if too many photons hit the sensor...

    I have put my head in the lion’s mouth. Please do not chew it off, but I would welcome to have any errors of understanding or logic pointed out.

    Many thanks!

    David

  • Members 1737 posts
    April 17, 2023, 3:35 p.m.

    c) making the lighting brighter

  • Members 1737 posts
    April 17, 2023, 3:36 p.m.

    "fill it"? Needs explication.

  • Members 457 posts
    April 17, 2023, 3:37 p.m.

    The noise source is not an "analog amplifier," though, through brightening, the noise is made more visible. Raising ISO (without changing exposure) lowers the noise.

  • Members 1737 posts
    April 17, 2023, 3:38 p.m.

    Analog amplification in practice in modern cameras does not add to the input-referred read noise, and may make that smaller.

  • Members 509 posts
    April 17, 2023, 3:44 p.m.

    Hey, can I request a new forum feature? It would be called "Sparing my blushes". It's function would be to search all posts for a certain phrase and issue an automatic correction addendum to each post.

    Thanks

    Dave

    p.s.

    I found a very useful article on DPR from back in 2007 all about ETTR and ISO technicalities and strategies. Bizarrely, it seems to support what people are saying here. Most disconcerting 😁

  • Members 457 posts
    April 17, 2023, 4:07 p.m.

    IMO, the best thing that can happen to me on the forum is when I have made a mistake (I am trying to avoid it, though). When others have corrected me, I have learned something, and the time spent on forums is not entirely wasted :).

    Mistakes and failures (in business) are opportunities to learn, not a reason to feel humiliated.

  • April 17, 2023, 4:19 p.m.

    No.I assume the lighting is a given.

    I take it that the sensor can only take a certain number of photons before overloading -- at that point it is "full" (like a bucket).

    I am referring to kTB noise, or even excess noise, which is present in any analog amplifier. This sets the s/n ratio and raising the signal level must also raise this noise and make it visible. I dont understand your last sentence in the context of pre-sensor analog amplification.

    How so? (TANSTAAFL) In audio systems we have "ambient noise", which is in the room, and usually this is higher than the noise added by the transducer and its amplifier. But it seems to me that ambient (Brownian?) noise in a photographic scene is much lower. So any any visible noise must come from the amplifier directly following the light collector (lens system).

    [The quoting system is not playing ball with me!]

    David

  • Members 1737 posts
    April 17, 2023, 4:23 p.m.

    I think many folks assume the lighting is a given. But looking for ways to make the scene brighter is very effective, and should be front of mind for all photographers, even beginners.

    Have you calculated input referred read noise vs ISO curves? Failing that, have you looked at the curves that others have calculated?

    Analog gain reduces the input-referred effects of noise sources further down the pixel pipeline than the programmable gain amplifier.

  • April 17, 2023, 4:24 p.m.

    There is no direct relation between how 'full' the sensor is and how light or dark the final image looks. In fact, most cameras don't use their sensors to anything like the photoelectron capacity that is available, because the engineers wanted to avoid the non-linearity that can come as the potential well approaches full capacity.

  • Members 1737 posts
    April 17, 2023, 4:27 p.m.

    Bolding mine. WRT the bolded text, that's not true in any camera system for which I've calculated photon transfer curves.