Exposing to the Left

Discussion in 'Open Discussion' started by Lcrunyon, Feb 21, 2015.

  1. Lcrunyon

    Lcrunyon Mu-43 All-Pro Subscribing Member

    Jun 4, 2014
    Hi all,

    I have a technique question for everyone. If I intentionally under-expose my shot - not enough to clip the shadows - and increase exposure in post, would this be a viable technique to reduce noise (by avoiding having to raise ISO too high), or would it make no difference in the end?

    I'm shooting some birds in my backyard and it's snowing a lot, so conditions are such that my ISO needs to be higher than I like.
  2. MadMarco

    MadMarco Mu-43 Veteran

    Oct 30, 2014
    Guildford, England
    I've tried this in both directions and in the end is made no real difference to the noise levels.

    This goes as far as shooting lower ISO to under expose, then boost in post processing. Comparing this to higher ISO correctly exposed and the difference is negligible and in some cases the under exposed image is actually worse. The only way to improve noise is to lower the shutter speed or increase the aperture to allow lower ISO.

    Definitely a case of no free lunch!
    • Like Like x 1
  3. Lcrunyon

    Lcrunyon Mu-43 All-Pro Subscribing Member

    Jun 4, 2014
    Gotcha, and thanks!

    These little guys move really fast, and to freeze the action I need a shutter speed of at least 1250. Aperture is already as low as I can go. Thought it was worth asking, though.
  4. Lcrunyon

    Lcrunyon Mu-43 All-Pro Subscribing Member

    Jun 4, 2014
    Just to follow up, I just processed some under-exposed pictures I shot before you answered my post, and I think you are absolutely right. The under-exposed images are no better, if not worse, than those I shot with the correct ISO. Thanks again.
    • Like Like x 1
  5. MadMarco

    MadMarco Mu-43 Veteran

    Oct 30, 2014
    Guildford, England
    No problem!

    I always like to test these things for myself, at the very least it helps prevent you from messing up important once in a lifetime shots.
  6. Rudy

    Rudy Mu-43 Top Veteran

    Jan 24, 2013
    Oakland, CA
    When you underexpose and increase the brightness in post you are operating on the digitized data. 8 bit per channel for JPEG and 12 bits for raw.
    Changing the ISO number on the other hand affects analog gain inside the camera before the data is being digitized. If there is too much gain (higher ISO) then the sensor noise will be noticeable in the digitized image.
    The final outcome however is not the same. In the first case you generate gaps in the histogram and in the latter you get a broader noise distribution.
    A good way to illustrate this is to take an image of a grey scale gradient, a light falling off on a wall would work.
    If you severely underexpose and then fix the exposure in post you will see steps where there should be a smooth transition.
    If you use the higher ISO the image will still be noisy, but there won't be any noticeable steps.
    • Like Like x 1
    • Informative Informative x 1
  7. David A

    David A Mu-43 All-Pro

    Sep 30, 2011
    Brisbane, Australia
    What is "under-exposure"? If you're shooting a snow scene so there's a lot of white in the frame and you're using the normal whole sensor averaging exposure meter mode, then the meter is going to recommend an exposure that's going to darken down the snow so it's actually going to recommend over-exposure. If there's not much in the frame other than snow and the bird then you can probably "under-expose" by 2 stops and end up with a shot that doesn't require you to increase exposure in post if you're using averaging meter mode. In other words, you won't really be under-exposing, you'll be correctly exposing for the bird if you "under-expose" by 2 stops.

    Meters assume that the area you're metering has a "normal" distribution of shadows, highlights, and mid tones. Not all scenes have that and a scene with lots of snow has a higher than normal proportion of highlights and a lower than normal proportion of mid and shadow tones so the recommended exposure you get is going to be wrong if you use a whole frame average. Correcting for that meter error isn't "under-exposure", it's correct exposure. Meters don't guarantee correct exposure regardless of the scene in front of them so what you're thinking of as "under-exposure" may very well not be under-exposure and it could very well be the right exposure.
  8. agentlossing

    agentlossing Mu-43 Hall of Famer Subscribing Member

    Jun 26, 2013
    Andrew Lossing
    I think it's simpler than that in most people's minds - underexposure is left of the metering's recommended exposure. Oversimplification? Yeah, but it works. I've always felt that spot metering was the only way that the pure mechanics of light metering really made sense to me.

    Sent from my Nexus 6 using Mu-43 mobile app
  9. Turbofrog

    Turbofrog Mu-43 Hall of Famer

    Mar 21, 2014
    Good post, I agree with you in principle, but you have the explanation backwards in your snow scenario. The meter thinks that a normal medium grey tone distribution is correct, when in fact there is a lot more white, and so it will recommend underexposure, not over.

    I regularly leave my exposure compensation at +2EV when shooting in the winter time.
    • Agree Agree x 1
  10. The following is a copy of a post I made a while back when discussing the concept of "ISOless" sensors. This type of sensor means that you can virtually replicate a higher ISO by underexposing at a lower ISO and then pushing the raw file (the method being suggested by the OP), but there is no advantage in technical image quality to be had by doing so. Note that the test below was done using an E-M5 and the results may be worse for Micro 4/3 cameras that don't use the Sony sensor. Even here the image shot at the higher ISO still looks better than the one shot at a lower ISO and pushed.


    Not every modern sensor is ISOless, and I think that the technology itself should be regarded as merely another way of designing a sensor as opposed to a better way of designing a sensor. The only real advantage I see to ISOless sensors is the ability to recover stupidly underexposed images. With Sony being the main proponent of this technology, I did a quick test with my E-M5 which uses a sensor of Sony origin.

    E-M5, 25mm, f/4, 1/13sec, ISO 200


    E-M5, 25mm, f/4, 1/13sec, ISO 200, (the same file as above but pushed approx 4 stops)


    E-M5, 25mm, f/4, 1/13sec, ISO 3200


    A comparison at 100% reveals that more work would be required to attempt to equalise the images, but the results are close enough to indicate to me that the Sony m4/3 sensor is near to being an ISOless sensor.

  11. b_rubenstein

    b_rubenstein Mu-43 All-Pro

    Mar 20, 2012
    Melbourne, FL
    Starting point exposure is based upon the amount of light falling on the subject. This is measured with an incident light meter. There is no need to interpret the meter reading based upon the reflectivity of the subject. (Meters do not get fooled, people using meters don't understand what the meter reading means.)

    The whole business of exposing to the left/right may have had some use in the early days of digital cameras 15 years ago when in camera processors, processing algorithms, sensors and electronics were in their commercial infancy.
  12. budeny

    budeny Mu-43 All-Pro

    Mar 4, 2014
    Boulder, CO
    What about exposing to the right while keeping ISO the same? From my expirience, sensor captures more information and and image is taking post-processing with better results.
  13. Turbofrog

    Turbofrog Mu-43 Hall of Famer

    Mar 21, 2014
    Obviously assuming you still have a sufficient shutter speed to ensure there's no blur and you are not clipping any critical (non-specular) highlights, exposing to the right with the same ISO will always result in better quality. More light, more information.
  14. David A

    David A Mu-43 All-Pro

    Sep 30, 2011
    Brisbane, Australia
    You're right that incident light meters don't need interpretation but most people don't use an incident light meter. In fact most people don't even have an incident light meter. They use the meter in their camera instead and those meters are not incident light meters. In any event, if you're shooting a landscape and the area you're photographing is 10 miles away, the light may well have changed by the time you get back to your shooting position after taking the incident light meter reading. Incident light meters are great for things like studio work but not practical for some other sorts of work. If you can't move from your shooting position to your subject, place the incident light meter appropriately and take your reading, then get back to your shooting position to take your photo without the light changing, an incident light meter is not the meter to choose.

    And while you don't have to interpret an incident light meter reading based on the reflectivity of the subject, where you place an incident light meter to take your reading is important. Placing it in shadow will ensure your reading is unreliable. You need to place it somewhere where the light falling on the subject can fall unobstructed on the meter's sensor or your reading will be inaccurate.

    The meters in cameras are reflected light meters and they do need interpretation. Their starting point is not the light falling on the subject but the light reflected from it.

    So, in the end it doesn't matter whether you use an incident light meter or a reflected light meter, you need to know how to use the meter. With an incident light meter you need to know where and how to place it in order to get a reliable measurement. With a reflected light meter you need to know how the reflectance characteristics of your subject affect the meter response and compensate for that when necessary. There is no meter that will guarantee correct exposure recommendations every time regardless of how you use it. Use any meter correctly and you'll get a reliable exposure reading, use any meter incorrectly and you won't.
    • Like Like x 1
    • Agree Agree x 1
  15. Lcrunyon

    Lcrunyon Mu-43 All-Pro Subscribing Member

    Jun 4, 2014
    Sorry if I didn't explain myself well, but my situation was different. I was in manual mode, and so in full control of the exposure. It was getting dark due to the weather, and my meter was telling me to increase exposure a lot by one means or another. Compensating for getting the snow nice and bright would have been another issue entirely. I didn't want to lower my shutter speed anymore (and aperture was already wide open) so that I could capture the birds' movement. The only thing I could do was increase ISO, but I didn't want the noise from doing that.
    Last edited: Feb 25, 2015
  16. LovinTheEP2

    LovinTheEP2 Mu-43 Top Veteran

    Feb 15, 2011
    I do not feel exposing the left makes much sense for trying to get loss noise shadows.

    By under exposing your decreasing dramatically the amount of light hitting the sensor which will cause exactly what you do not, loss of photons which will cause noise. So by pumping up an image with fewer photons you'll wind up with an image close to a properly exposed image. The properly exposed image will probably have more accurate color too.

    Getting clean shadows is more about proper exposure, sharp glass, Fastish exposure to avoid motion blur, avoiding strong backlit around shadows.

    If you really want clean shadows, then image stacking multiple exposures is the way to go.
  17. pellicle

    pellicle Mu-43 Hall of Famer

    Feb 10, 2010
    Kiillarney, OzTrailEYa
    it seems to me that what you are demonstrating here is that gain in photoshop VS gain in sensor amplification results in the same. If you put a well exposed shot of 200ISO in there for the 100% it may lookbetter ... perhaps I missed it
  18. That's correct. The original post was from a test to compare the in-camera ISO 3200 vs do-it-yourself ISO 3200 to see if the E-M5 has a so-called "ISOless" sensor. In this instance I've re-posted it here to show the OP that deliberately underexposing at a lower ISO and then pushing the exposure during processing won't result in less noise.
  19. jyc860923

    jyc860923 Mu-43 Hall of Famer Subscribing Member

    Feb 28, 2012
    Shenyang, China
    I'm looking at Luckypenguin's test, the ISO200 pushed +4 vs ISO3200, and I think it's obvious the ISO3200 wins with better DR and noise. Does that mean native sensitivity ALWAYS outperforms digitised push? And if so, can I assume that ETTR at higher ISO still gives an edge over normal exposed image? Just want to know even though I don't like ETTR most of the time for the harsh results in highlight.
  20. m4/3boy

    m4/3boy Mu-43 Veteran

    Jul 21, 2013
    Underexposure is never a good idea; you are throwing away data and asking for mediocre results.
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.