• Members 240 posts
    May 11, 2023, 7:02 p.m.

    I understand that you are suggesting that a low light events photographer does not benefit from lowering their ISO from 12,800 to 6400 where possible.

  • Members 976 posts
    May 11, 2023, 7:02 p.m.

    I answered it many times: I either preset everything (sport arena), or I use manual exposure + limited ISO auto.

  • Members 240 posts
    May 11, 2023, 7:02 p.m.
  • Members 976 posts
    May 11, 2023, 7:04 p.m.

    No, and Bill Claff's data isn't the reason why why DR falls as ISO climbs.

    Iilah, mayby; I'm not. But I try - part of it is remaining silent or asking questions when I'm not sure - and that happens a lot. And I never ask questions prefacing the question with "whatever the answer is my way of doing things is right".

    I prefer a column.

  • Members 240 posts
    May 11, 2023, 7:09 p.m.

    And there we have it. Limited ISO Auto. I agree. Therefore, in those environments, ETTR is not irrelevant whatsoever. When shooting M mode with Auto ISO then using +EC of one stop is legitimate. It halfs (LIMITS) your ISO. With the flick of the EC dial I can half my ISO from 12800 to 6400 if lighting allows. The benefit of this approach is that I dont have to set a maximum auto ISO. I'm in control, with the benefit that I can still grab any shot that suddenly fall outside these lighting parameters.

  • Members 976 posts
    May 11, 2023, 7:12 p.m.

    Which camera?

  • May 11, 2023, 7:14 p.m.

    It's a matter of what the right concepts are. As I said at the beginning of the post, you can apply the right words to the wrong concepts and what you get is a mess. You don't see anything that disagrees with what you wrote because you have adopted a mental model and your brain takes whatever you see and bends it into that model. Let's see...

    'Under' compared to what?

    It's not the condition of being 'under' that determines SNR, it's simply the size of the exposure. This is why the concept of 'underexposure' can be so misleading.

    For a photo, from the largest exposure you actually use. For a potential photo, the largest exposure is determined by the clipping point.

    The 'increased noise' cause by low exposure doesn't affect the noise floor. The ISO control might.

    No. DR and SNR are loosely connected.

    Everything we say and write is 'just words'. Lightness has a standardised meaning. It's up to you to decide whether to use it or not, but using specialised vocabulary words in ways outside that defined within the specialisation is a recipe for confused thinking. 'Brightness' is more colloquial. The important point I was making is that the type of phenomenon at input and output of a camera is completely different. Thinking they are related just by 'gain' or 'multiplication' is a category error.

  • Removed user
    May 11, 2023, 7:15 p.m.

    Got me puzzled with that one, Iliah. I'm trying to imagine setting an aperture in the commonly-used Aperture Priority but watching the shutter speed going up when ISO is increased - which does not "keep the exposure" ... on my simple cameras anyway.

  • Members 240 posts
    May 11, 2023, 7:23 p.m.

    Sony A9II. I have mine set to aperture rear command dial, shutter front command dial. The EC dial is directly to the right of the rear command dial.

    I'm in M mode, auto ISO, no upper limit set.

    I shoot Raw. I use Zebras. I set the Zebras to "custom 109+"

    My ONLY concern initially is setting shutter speed (slowest I can get away with for my motion blur requirements) and aperture (brightest I can get away with for sufficient dof).

    My goal is to NAIL THE DECISIVE MOMENT!

    I can then alter shutter, aperture or EC with my thumb and forefinger without taking my eye off the EVF, dialling in my settings as the shoot and/or lighting evolves.

    The zebras will give me an indication if I can improve raw file quality by altering my parameters on the fly without taking my eye off the EVF.

    For example, a simple one stop +EV will half my ISO. Or I can slow my shutter, or open my aperture. (The latter two options being the ones that put more light on the sensor). But halfing my ISO will give me a bit more DR in post.

    BUT, that's IF time allows and it doesnt detract from the primary goal of NAILING THE SHOT.

    Now, if anyone has a better way of doing it, for fast moving dynamic situations in the dark without missing a decisive moment then I'm all ears.

    Scientific best practice is great, nailing the shot is even more important for me (and especially for my clients).

  • Members 509 posts
    May 11, 2023, 7:27 p.m.

    Well, I'm not going to dispute with Iliah, but I'll try and explain why I said that.

    Assuming the goal is ETTR at base ISO. If I don't have enough light to achieve that, I can resort to raising the ISO. But we know that doing that doesn't magically make the sensor more sensitive. In effect, what it does is underexpose compared to an ideal ETTR exposure and then lie to you to make you feel good about it.

  • Members 509 posts
    May 11, 2023, 7:32 p.m.

    Perhaps you are an expert in the correct terminology and fussy about it. I'm not, and like most people I use it loosely and inconsistently because I have a list of about 5 synonyms for these words and they all mean the same thing to me. If we are going to communicate using a standardised language to avoid muddle, that is good but I suggest we need to set out those terms as a glossary. You can't just assume that your correspondent is familiar with physics and engineering language. Most of us are photographers who know little of engineering. Be more tolerant.

    Correct exposure in my world is ETTR. Any exposure that is less than ETTR is underexposure. Using an exposure that is less than ETTR results in increased noise and reduced DR compared to ETTR. Increasing ISO in order to try and ETTR is just lying to yourself.

    Now, you have said several times that I have a mental model that I won't let go of. Why don't you set out what your perception of my mental model is and we can decide whether it is my concepts that are fault or if I'm just careless with words,

  • Members 240 posts
    May 11, 2023, 7:39 p.m.

    I think after several hundred posts on ETTR at base ISO across several threads at this place plus thousands at the other place I think we ALL get that.

    Now we are discussing a tiny little aspect that some folk may find vaguely useful - Namely, relating it all back to actual real world photography and best practices based on real world genres shot.....

  • Members 83 posts
    May 11, 2023, 7:44 p.m.

    Thanks @JimKasson for the information at LensRentals. Well done.

    One thing I am still a little puzzled about is how the gamma curve (or curves) impact the values near clipping and how this might be predicted from a raw histogram. A Franklin's Gull was flying over my house when my camera was ready for a perched bird and I am old and slow. When I make what Canon DPP calls "gamma" adjustments, the shape of the curve changes, the input limits change (but less than is possible with rawtherapee), and the toe and shoulder of the curve cause levels to be stretched or compressed.

    (some command lines executed on my Debian machine and some on my iMac)

    image.png histogram in Canon DPP with cursor over pixel that has red value of 255.

    image.png histogram produced by
    /home/jrm/src/libraw2021dec18/LibRaw-master/bin/dcraw_emu -v -T -6 IMG_4055.CR3
    and
    /opt/local/bin/gmic IMG_4055.CR3.tiff keep[0] display_histogram 800,600,65536,0,65535,0,"cut(i,0,65535)" normalize[-1] 0,255 output[-1] IMG_4055a_hist.png ; echo "gmic histogram"
    [gmic]-0./ Start G'MIC interpreter.
    [gmic]-0./ Input all frames of TIFF file 'IMG_4055.CR3.tiff' at position 0 (1 image 8191x5463x1x3).
    [gmic]-1./ Keep image [0] (1 image left).
    [gmic]-1./ Render 800x600 channel-by-channel histogram of image [0], with 65536 clusters, minimum value 0 and maximum value 65535.
    [gmic]-1./ Normalize image [0] in range [0,255], with constant-case ratio 0.
    [gmic]-1./ Output image [0] as png file 'IMG_4055a_hist.png' (1 image 800x600x1x3).
    [gmic]-1./ End G'MIC interpreter.
    gmic histogram

    image.png histogram produced by changing the input range in Canon DPP to +2.00 and exporting 16 bit tiff and
    /opt/local/bin/gmic IMG_4055h.TIF keep[0] display_histogram 800,600,65536,0,65535,0,"cut(i,0,65535)" normalize[-1] 0,255 output[-1] IMG_4055h_hist.ppng ; echo "gmic histogram"
    [gmic]-0./ Start G'MIC interpreter.
    [gmic]-0./ Input all frames of TIFF file 'IMG_4055h.TIF' at position 0 (2 images [0] = 8192x5464x1x3, [1] = 252x168x1x3).
    [gmic]-2./ Keep image [0] (1 image left).
    [gmic]-1./ Render 800x600 channel-by-channel histogram of image [0], with 65536 clusters, minimum value 0 and maximum value 65535.
    [gmic]-1./ Normalize image [0] in range [0,255], with constant-case ratio 0.
    [gmic]-1./ Output image [0] as png file 'IMG_4055h_hist.png' (1 image 800x600x1x3).
    [gmic]-1./ End G'MIC interpreter.
    gmic histogram

    Franklin's Gull (Leucophaeus pipixcan) in Norman, Oklahoma, United States on May 8, 2023

    [EXIF:Image] Exposure Time : 1/800
    [EXIF:Image] F Number : 9.0
    [EXIF:Image] ISO : 400
    [EXIF:Camera] Focal Length : 560.0 mm
    [MakerNotes:Camera] Camera Temperature : 41 C
    [MakerNotes:Camera] Lens Model : EF100-400mm f/4.5-5.6L IS II USM +1.4x III
    [MakerNotes:Camera] Measured RGGB : 403 1024 1024 779
    [MakerNotes:Camera] WB RGGB Levels Measured : 2097 1024 1024 1783
    [MakerNotes:Camera] Color Temp Measured : 5807
    [MakerNotes:Camera] WB RGGB Levels Daylight : 1975 1024 1024 1886
    [MakerNotes:Camera] Color Temp Daylight : 5200
    [MakerNotes:Camera] Normal White Level : 13535
    [MakerNotes:Camera] Specular White Level : 14888
    [MakerNotes:Camera] Linearity Upper Margin : 12735
    [MakerNotes:Image] Measured EV : 14.12
    [MakerNotes:Image] Measured EV 2 : 31
    [MakerNotes:Image] Focus Distance Upper : 164.65 m
    [MakerNotes:Image] Focus Distance Lower : 81.91 m
    [MakerNotes:Time] Time Zone : -05:00
    [MakerNotes:Time] Time Zone City : Chicago
    [MakerNotes:Time] Daylight Savings : On
    [MakerNotes:Time] Time Stamp : 2023:05:08 19:17:21.86
    [CanonVRD:Image] Gamma Contrast : 0
    [CanonVRD:Image] Gamma Color Tone : 0
    [CanonVRD:Image] Gamma Saturation : 1
    [CanonVRD:Image] Gamma Unsharp Mask Strength : 0.7
    [CanonVRD:Image] Gamma Unsharp Mask Fineness : 1.7
    [CanonVRD:Image] Gamma Unsharp Mask Threshold : 2
    [CanonVRD:Image] Gamma Sharpness Strength : 4
    [CanonVRD:Image] Gamma Shadow : 1
    [CanonVRD:Image] Gamma Highlight : -1
    [CanonVRD:Image] Gamma Black Point : +0.000
    [CanonVRD:Image] Gamma White Point : +0.050
    [CanonVRD:Image] Gamma Mid Point : +0.000
    [CanonVRD:Image] Gamma Curve Output Range : 0 16383
    [CanonVRD:Image] DLO Setting : 47
    [CanonVRD:Image] Crop X : 3317
    [CanonVRD:Image] Crop Y : 1626
    [CanonVRD:Image] Crop Width : 2400
    [CanonVRD:Image] Crop Height : 1600

    image.png

    PNG, 72.5 KB, uploaded by JohnMoyer on May 11, 2023.

    image.png

    PNG, 65.2 KB, uploaded by JohnMoyer on May 11, 2023.

    image.png

    PNG, 67.2 KB, uploaded by JohnMoyer on May 11, 2023.

  • Members 509 posts
    May 11, 2023, 7:44 p.m.

    Before all those posts, I had no real understanding of the exposure/ISO thing. I adopted the majority view and changed my understanding, I thought to that of Jim, Iliah, Bob and others. Well, Iliah and Bob said I'm wrong, and as far as i can see I'm still adopting the understanding of the majority from those threads, not the dissenters. And indeed, Jim gave his seal of approval to my description just a few days ago, so I'm curious as to why all of a sudden I'm considered to protecting an incorrect mental model.TBH is still seems like it's a dispute over words...

  • Members 240 posts
    May 11, 2023, 7:50 p.m.

    There's been so many posts that one must sort the wheat from the chaff. Or, better still, go out and shoot. Of the three folk you mentioned, two of them have also divulged that they use Auto ISO in the real world, Jim is the only one that uses fixed ISO AFAIK ;-)

    In the UK we have fuel consumption figures for cars. They all seem to perform most frugally at 56mph.

    But the speed limit is 20mph in city centres and 70mph on motorways.

    It's handy to know the optimum but it's as much use as a chocolate fireguard in the real world.........

  • Members 976 posts
    May 11, 2023, 7:52 p.m.

    No SNR benefit after ISO 1600 that I could measure, only clipping. Can be kept at ISO 640. That, if you shoot raw.

  • May 11, 2023, 7:56 p.m.

    What's the significance of that?

  • Members 509 posts
    May 11, 2023, 7:56 p.m.

    Previously, I used base ISO only. Then one day it dawned on me that a noisy shot that is sharp is better than a smooth shot that has unwanted motion blur or camera shake, so I switched to using auto-ISO. Then because of the mega threads on ISO and exposure, I changed my mind again and decided to keep the camera on base ISO and ETTR. When the light is too low to pull that off, I'm just going to set whatever aperture and shutter speed I need and ignore correct exposure. Just underexpose if that is the only option and brighten/lighten/raise midtones (whatever the official terminology is) in post. Jim said that might sometimes not be perfectly optimal as a strategy but it will do. I was happy.

    Now Iliah and Bob have cast doubt on this, so I am once again worried.