View Full Version : Getting the exposure right

02-06-2017, 10:46 PM
One of the important factors in creating a high quality image is getting the exposure right in camera. Why is getting the exposure right in the camera important? Highlights that are fully blown (clipped) in RAW can never be recovered. Images that are underexposed may be corrected in post processing, but it requires careful processing and noise reduction. With small-sensors cameras such as the Canon 7D series underexposing more than one stop usually results in a poor quality image. EOS-1DX (MKII) and the EOS 5D MKIV, on the other hand can handle several stops of underexposure relatively easily. Nevertheless, images with extreme exposure adjustment may not be eligible to enter certain competitions or eligable for publication where the original RAW is used. In any case, it is best to get the exposure right in the camera, which also saves time in post processing.
But what is the best way to set your exposure? There seems to be some confusion on the internet forums about this. Some photographers rely on and recommend using the histogram. Histogram works just fine for landscape photos but it is not a reliable method for bird photography. This is because the subject, i.e. the bird often occupies a tiny percentage of the frame. Even when a bird is reasonably large in the frame to make it a keeper, it contains a small fraction of the total pixels in the image. The image shown below is a full-frame RAW capture with the 20.1 Mega-pixel EOS-1D X Mark II. The golden eye is descent size in the frame. I selected the golden eye in Photoshop, the pixel count is 70,211. Compared to the full-frame (20.1 Million pixels) this is roughly 0.3% of the total pixel count! The histogram displayed on the rear LCD of the camera, on the other hand, is calculated from all the 20.1 million pixels in the frame.


Golden eye. EOS 1D x Mark II, EF 400mm f/4 DO IS II + EF Extender 2X III. ISO 800. F/8 at a/3200sec. Hand held. Manual exposure. The bird is large enough in the frame, but it contains <1% of the pixels in the frame!


14-Bit RAW and 8-Bit RGB histograms for the image above in Canon DPP 4.5. The histogram displayed on the camera’s rear LCD is similar to the lower one. The bird is the tiny “hump” near the right border, it is hardly visible. The large peak is NOT the bird but the background.

The screenshot above shows the histogram for this photo in DPP4.5. It is obvious that the histogram displayed is dominated by the background. The bird is the tiny hump on the far right, it is almost invisible. Therefore if you were to peg the background peak to the right 3rd of the histogram the whites will be totally blown. Furthermore, in images where the bird has similar colors to the background it is impossible to distinguish the bird and the background in the histogram and tell what is what. The goal is to expose properly for the bird, not the background.
So what is the best way to tell the correct exposure for avian photographs? I find that the most reliable method for checking exposure on the rear LCD is to rely on the highlight warning, aka “blinkies”. When exposure warning is on, the areas with an RGB value of 255 will blink, clearly showing the hot spots. Because the image on the camera’s rear LCD is a 8-Bit compressed JPEG file, a blinking area may be recoverable using the data that is in the 14-Bit RAW file. The extent to which you can recover the highlights using the RAW, depends on the camera model. With high dynamic range sensors such as the one in EOS-1DX Mark II you can often recover a solid blinking area. With small-sensor cameras like the EOS-7D series, only small blinking patches will be recoverable. I usually increase my shutter speed until I don’t see any solid blinking area to be on the safe side. With both EOS-1D X and EOS-5D series you can easily increase the exposure on the dark tones, but if a bright area is clipped there is not much you can do about it. In the image above I did have two small blinking patches but the highlight detail was preserved in the RAW. This is as far as I am comfortable pushing it.

click here (http://arihazeghiphotography.com/blog/wp-content/uploads/2017/02/blinkies-1.gif) to see the blinkies as displayed on the EOS-1D X Mark II rear LCD
Below is the image after processing in DPP4.5. I used a highlight adjustment of -3 to bring out slightly more detail in the whites.


For bird photography, I also prefer to use manual mode at all times, especially for in-flight shots. As explained above, because the bird occupies a tiny percentage of the frame, the camera’s meter will set the exposure for the background. This means the bird will be blown when it flies against a dark BG, and it will be severely underexposed when it lies against a bright BG. With manual exposure set for the bird, the exposure remains locked as you pan with the bird which means all frames will be exposed properly so long as the bird stays in the light.

I hope this quick tutorial will help you setting the exposure in the field.

Krishna Prasad kotti
02-06-2017, 11:13 PM
Quite an interesting and Informative Post. Thank you

02-07-2017, 01:58 AM
Thanks Krishna, we are revamping this forum, hopefully there will be more material and articles soon

Gerald Kelberg
02-07-2017, 05:02 AM
Hi Ari, thanks for the article - it is a well presented reminder of an important principle.

I am having difficulty reconciling your calculation that the bird represents 0.3% of the pixel count as it is about 16% of the image. Is this because there are fewer pixels making up the whites of the bird? I'm surprised that the difference is so extreme. But I guess that is your point!

You're the scientist, not me!

Best wishes, Gerald

02-07-2017, 11:14 AM
Hi Gerald, thanks for kind words,
Here the bird is not 16% of the frame, it is only 0.3%. In order to see exactly how many pixels are on the bird, you can't simply draw a box around it, you need to use the lasso tool in photoshop to select it, the info tab will tell you the pixel count. our eyes often deceive us :S3:


Arthur Morris
03-05-2017, 03:25 PM
Arash, Are you saying that the bird accounts for less then 1% of the pixels in the image?

03-09-2017, 02:32 AM
yes Artie that's correct. In the original it is less than 1% of the total pixels of the frame. Human eye is good in perceiving length but not area, as a classic example, if you double the radius of a circle, its area increases 4 folds.

And Peter is right, in the small version it is 5% because I slightly cropped the image.

Joseph Przybyla
03-25-2017, 04:13 PM
Arash, thank you for the expert information.

A question, do you use the screen showing the histograms for the luminance and the individual red, blue, and green channels or just the blinkies? If just the blinkies would that show a blown out color channel?

03-28-2017, 02:27 AM
Hi Joe,

I just use the blinkies. Blinkies is for the entire image (i.e. luminance only) but with experience you can tell whether a channel is blown beyond saturation or not. I try to be conservative and keep blinking areas very sparse, shadow recovery is not an issue usually.