Noise and Blown Highlights Aren't the End of the World

If there are two things that photographers work really hard to avoid in their images, they are noise and blown highlights. It is drilled into us from day one that these are to be avoided as much as possible. But are they really the end of the world? Maybe not. 

Coming to you from Blake Rudis with f64 Academy, this great video discusses the topic of noise and blown highlights in photos and why they are not always the end of the world. Sure, the more we can avoid these things, the better, but that being said, if you can't avoid them, then to get the best quality image, you should embrace them as much as necessary. For example, I often see newer photographers do keeping their shutter speed too slow for the action they are photographing or focal length they are using in an attempt to keep the ISO as low as possible. It is much better to have a noisy but sharp image than a photo that is blurry from too slow a shutter speed. You can always apply a bit of noise reduction, but once a photo is blurry, there is rarely much you can do to save it. Check out the video above for the full rundown from Rudis. 

Alex Cooke's picture

Alex Cooke is a Cleveland-based portrait, events, and landscape photographer. He holds an M.S. in Applied Mathematics and a doctorate in Music Composition. He is also an avid equestrian.

Log in or register to post comments
4 Comments

Thanks for this!

That I doubt a bit: "better than anything else". 'darktable' with its 32bit fp processing (LR 16 bit integer) and scene-referred workflow in RGB colour space is to be beaten first.

Which camera has more than 14 bit raw data so that 32 bit plays a major role?

The two bit width are not related in any way. The 14 bit represent the amount of (light) data (of each pixel). The 32 bit fp the precision of the calculation. I quickly searched the web and found this page which explains it fairly well: http://admvfx.com/knowledge-base/floating-point-vs-integer/

That said, processing RAW data is normally done in several steps. Each step calculated with 16 bit integer leads to more and more loss of information, much much more compared to 32 fp operations.

An important fact to consider: Sensors deliver much more data in the brighter part of the image and that is the data we do not want to loose (to reconstruct the highlights).

But that is not the only advanced feature of darktable: The scenic workflow is in a RGB colour space and not in Lab. You'll find the information in the web if like to know more about it.