Month: August 2016

Today’s Question: What about the new ISO-invariant cameras? Does you answer [about optimal night exposures in yesterday’s edition of the Ask Tim Grey eNewsletter] apply to them too?

Tim’s Quick Answer: In general I would still say that increasing ISO in the camera is preferred over adjusting in post-processing, even with a sensor that has been labeled as being “ISO Invariant”.

More Detail: The term ISO Invariant in this context refers to a sensor where you can achieve the same results by increasing the ISO in the camera, or under-exposing the image and then brightening in post-processing. That doesn’t mean you’ll have a result with low noise levels necessarily, but rather that the results will be the same with the two approaches.

It is important to keep in mind that, as I pointed out in the article “ISO Illustrated” in the December 2013 issue of Pixology magazine, raising the ISO setting really represents underexposing a photo (perhaps severely) and amplifying the signal recorded by the image sensor in an effort to compensate.

In other words, raising the ISO setting can be thought of as brightening a photo in much the same way that dragging the Exposure slider in Adobe Camera Raw or Lightroom will brighten the photo.

In other words, in the context of ISO the real question is whether the camera can do a better job of amplifying the signal recorded by the sensor or if software in post-processing is able to do a better job.

In general I have found that the camera does a better job of amplifying the signal compared to using software after the capture. This makes sense considering the camera has the benefit of analog data to work with from the image sensor, rather than digital values in the RAW capture file after the capture.

Some cameras perform better than others, of course, both in terms of baseline noise thresholds as well as amplification quality. What I would say in general though is that based on what I’ve seen and have been able to test, there isn’t a clear advantage to ignoring the ISO setting in the camera, even with an “ISO Invariant” sensor.

As such, my recommendation is still to expose properly in the camera, even if that involves increasing the ISO setting to achieve the desired overall settings. That still represents underexposing the photo and using ISO to brighten in the camera, but I have found that this provides a superior result in most cases.

Today’s Question: When shooting at night, is it better to shoot for a “proper” exposure at a high ISO, say 3200, or underexpose a couple stops with a lower ISO, then compensate in processing with increased exposure?

Tim’s Quick Answer: You will get the best results by achieving the brightest exposure possible without clipping highlight detail while at the same time using the minimum possible ISO setting. If you need to achieve a shorter exposure duration, generally speaking you are better off increasing the ISO setting rather than creating an underexposure.

More Detail: When you raise the ISO setting, in a way you can think of your result as being underexposed based on, for example, a shorter exposure duration. The resulting image is then brightened up through the use of amplification of the signals recorded by the image sensor.

With the various cameras I have had the opportunity to test, the results consistently show that it is better to let the camera brighten the image through a higher ISO setting than to apply brightening to the image after the capture. In other words, the in-camera amplification of the signal recorded by the image sensor yields higher quality than applying the same change in brightness with software after the capture.

So, if at all possible I would use the lowest ISO setting available to minimize noise, and create an exposure that is as bright as possible without clipping highlight detail (or only clipping the brightest areas, such as illuminated lights). If I needed a faster shutter speed (shorter exposure duration) for any reason, I would raise the ISO setting in order to achieve that goal, because this will generally provide the best final image quality compared to underexposing the scene and then brightening the image later.

Today’s Question: Is there any way to optimize a JPEG for a black and white photo? Since there’s no color information, can more gray tonalities be squeezed into fewer megabytes?

Tim’s Quick Answer: While it is certainly possible to produce a black and white JPEG image, this is not something I recommend due to the relatively high risk of posterization (the loss of smooth gradations) in such an image.

More Detail: JPEG images do not support high-bit data, meaning you can only have 8-bit per channel information available for a JPEG image. For full-color images that translates to more than 16.7 million possible color values. However, for a black and white image, having only 8-bits for what is then a single channel means there are only a maximum of 256 shades of gray available for a black and white (grayscale) image.

With only 256 shades of gray available, it can be very difficult to have (or maintain) smooth gradations of tonal value. For example, it is very common to see a banded appearance in a sky rather than a smooth gradation with a black and white image in the 8-bit per channel mode.

When strong adjustments are applied to an 8-bit per channel black and white image, the loss of smooth gradations is compounded. Note that the limitations of 8-bit per channel black and white images apply even if you are working with a color original. Even with a Black & White adjustment layer in Photoshop, working on an RGB image, for example, the final image can only contain up to 256 shades of gray, even though there is more information available in the source image.

Because of these factors, I highly recommend working only in the 16-bit per channel mode for black and white images. That, in turn, means JPEG images should generally be avoided in terms of the source image you optimize for a black and white photo. Instead, only a 16-bit per channel source image should be used as the basis of a black and white interpretation of a photo. You can then certainly save the final result (after all adjustments have been applied) as a JPEG image for purposes of sharing the photo, and still retain relatively high image quality for that final output.