Choosing the Dark Side

Canon vs Nikon. They never end. Lucky Sony, Olympus, Pentax and other underdogs for not having to deal with the stupid arguments. But this post isn’t about brand wars.

Expose to the right (ETTR) is a common advice in digital photography. It simply means, try to make sure that you expose your shot with bias towards the right end of the histogram. Make it as bright as practically possible without blowing out the highlights. If you understand how digital photos are stored, this makes sense. You want to maximize every bit of those 12-14 bits.

There is danger in blindly following this advice since the linear profile of digital camera sensors is not very forgiving. Once you clip past a certain limit, no data is stored in the photograph. This is characterized by blown highlights. Unfortunately, it is a lot easier to blow the highlights than lose the shadows.
There is something I discovered just a few months ago that I would like to share with you: It is better to underexpose than expose to the right. Not just underexpose but severely underexpose especially if the dynamic range of the scene is too wide.
Have a look at this photo because I quite pushed the camera beyond its limits when I took the shot:

Very dark isn’t it? The exposure was ISO 400, f8, 30 seconds after +2.5 stops of exposure compensation from the metered reading. That’s pushing the sensor a bit too much. I could have opened up to f5.6 but my cheap lens is very soft at that aperture. Going ISO 800, on the other hand, will only introduce more noise.
Now have a look at the same photo after post processing:

That’s a world of difference! I just pushed the exposure by +1.35 stops and then pulled some of the shadows with fill light. I have managed to extract details in the shadows while preserving the highlights. There’s more: peep all you want but there is barely a trace of luminance or chroma noise even after brightening the shadows. Amazing!!!

The photo was captured with a Pentax K5. It’s really amazing how modern sensors have improved. I would expect the same performance in the Nikon D7000 and Sony A55 because all of them use the same Sony sensor (surprise?!!!).

This is not the only instance where I managed to salvage a seemingly hopeless exposure. I do a lot of HDR work when the scene is too contrasty and I normally bracket at -2,0,+2. Many times, I was able to scrap the HDR because I was able to extract enough information from the -2 frame. Single exposure shots are still way cleaner than HDR so I always try to pull the shadows if I can.

Experiment with your own camera and see how much you can extract from a severely underexposed image. Make sure you shoot RAW.

So who’s coming with me to the dark side?

My new iPad blogging software ruined the original post. Lesson for me: sticking to one buggy software is sometimes better than switching software.


5 thoughts on “Choosing the Dark Side”

  1. Interesting – I’ve come to much the same conclusion for much of the type of shooting I do.
    I shoot long exposures at night, in the city. As you say, you can’t recover blown hig lights, but you can often pull a LOT out of the shadows.

    In fact I recently created a composite image from the same RAW, showing the almost black frame before processing and much higher detail image after processing.

  2. Hello, this was an interesting article, but there is IMO a flaw in your reasoning and it seems you missed a critical experiment and jumped to conclusions….

    >> ” I could have opened up to f5.6 but my cheap lens is very soft at that aperture. Going ISO 800, on the other hand, will only introduce more noise.”

    Pushing the exposure in post production from an underexposed shot, also introduces more noise. You did that and saw that the amount of noise was very acceptable with the K-5 shot.

    It seems that you simply *assumed* that pushing ISO (to 800 or more, depending on conditions) would end in MORE noise than was produced by the post-prod exposure push to achieve the desired result.

    But did you try using both methods for the same shot, and then compared the resulting noise in actuality…..? Maybe you’ll be surprised..

    Because in the end, in theory, shooting @higher ISO with proper exposure, or underexposing (with lower ISO) then pushing the light in post-prod, give actually…. more or less the SAME result. Because in both cases, ceteris paribus, the sensor received the same amount of light! 😉

    1. You are correct in saying that pushing the ISO in camera and pushing it in post would yield the same effect. Well, almost. In-camera ISO push will push everything including highlights. Software such as Lightroom is smart enough not to push the highlights further if they start clipping so the image becomes brighter but does not clip. LR kinda simulates the analog behaviour of film.

      Thanks for the comment.

      1. Hello again Mateo. ☺

        I’m not sure you got my point.

        “Pushing” ISO in-camera (actually setting the camera to higher a iSO reference, like going from 400 to 800) doesn’t push the levels of light. It actually lowers them. For a given light measure/metering, you have LESS light sucked by the sensor if you take your shot @800 rather than @400.

        Which means that for a given measure/metering, proper exposition @800 is exactly the same as underexposing by 1 stop @400. You get the exact same amount of light in the sensor, and the same result after processing the RAW file (ceteris paribus).

        The so-called “sensitivity” (ISO) is merely a reference for proper post-processing. It’s not like you’re using different film stock.. 😉

        The RAW converter (in camera or third-party) pushes the levels accordingly to the ISO reference of the shot, that’s all.

        Also, as long as the exposure is correct, there are no clipped highlights… what i called proper exposure excludes overexposure.

        This said, I certainly agree that the K-5 / K-5 II / K-5 IIs family allows sever undexposure without ruining shots. I love these cameras.

        Take care ☺


      2. I understand what you mean. The problem with the ISO push is that it forces the camera to underexpose (in the real sense of exposure) which introduces more noise. Note that I was using aperture priority mode hence the exposure compensation. ISO 800 would have resulted in a shorter shutter speed of 15s which underexposes it even more. ISO is not part of exposure so the goal is to optimise aperture and shutter speed then “tune” the ISO or push in post.

        In that shot where shutter speed was maxed out at 30s and aperture is optimal at f/8, pushing to ISO 800 will only clip the white patches in the sky uncontrollably. A different approach would have been to maintain the same shutter speed and aperture but shoot at base ISO 100 and that would have resulted in the same exposure except that I would not see anything in my LCD or even in the computer until I start pushing.

        I hope that clarifies.

        Edit: edited to clarify underexposure as a result of ISO push.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s