About Clipping blacks and blowing highlights: an attempt to bring together art, science and discipline

Introduction

Clipping in photography is well known and whereas sometimes done on purpose, it mostly comes as a non-desirable effect, because of poor exposure (worst case) or at least reaching the limits of the sensor range (best case). By clipping, I mean both blowing highlights and clipping blacks. The topic has been debated countless times in different forums and blogs.

As a summary, some people believe it does not really matter as long as the photo is great and other advocate why and how to avoid it. Other rightfully point out it is sometimes better not to fix it whereas other explain in detail how to do it the right way.

This is a classic case of different opinions in photography between those who do not want to consider something else than the purely artistic result and the scientists obsessed by being consistent with some physical principles. As usual too, both are right and wrong at the same time. Indeed, what matters in photography is the result, the emotions a photography can carry, and whether you like it. Period. Clipping, no clipping, who cares. At the same time, it is true to say that blowing your sensor which can no longer deliver any information but “I am blown” (white burned) or “I am blind” (black clipped) is not really what someone can call good practise, to say the least.

I am trying in this post to find a way to make all these opinions somewhat aligned, in a very much Swiss-like consensus way.

How to detect it and how to fix it

There are also plenty of information about the topic. I would recommend reading:

[1] How to Avoid Burned-Out Highlights
[2] Stop Doing This to Your Photo’s Highlights
[3] What is Clipping in Photography and How to Fix It!
[4] Restore Those Clipped Channels
[5] 6 Ways to Reduce Blown Out Highlights in Your Outdoor Photography
[6] Highlight Clipping in Adobe Photoshop Camera Raw (and Why You Should Care)¨
[7] What Is Clipping and How To Fix It
[8] Blowing Highlights And Clipping Blacks: The Rule Behind Lost Details

“Physical” and “visual” clipping

Most of the people know well the “physical” clipping: when the sensor is blown. Technically speaking, it means the pixels of a given channel (R, G, B) or its luminance (Based on the square root of R, G, B weighted according to the human eye characteristics) is at its maximum value (typically 255 for 8 bits JPG) or its minimum (0).

But it is also important to remember that what matters is the “visual” clipping: the pixels that are almost blown or clipped also matter because (at least for JPG images), there can be no way to really fix them properly and get information from the clipped regions of an image.


Example: a JPG image of a very high contrast scene.

Let’s have a look at the clipped pixels highlighted in blue for the blacks and red for the highlights in the image below. First, one could argue that using JPG for such conditions is not the best idea, RAW would have been by far a better choice but without going to start another countless debate RAW vs. JPG, the image has been poorly exposed as there is no clipped pixels in black (they would be in coloured in blue in the image below) whereas they are quite a few blown ones (in red below). So, basically, it says the image should have been significantly less exposed..


Same image with highlights in red, blacks in blue (none in this case)

But whereas the number of actuals blown pixels in red is not so significant, the number of visually clipped pixels is at an unacceptable level. It is making the image ugly whereas it was an interesting one. There are almost blown from a physical perspective, but for our eye, they are just blown… you can try to reduce highlights or exposure, there is basically no information recorded into the sunny mountains part of the photo. The image will stay poor. So, what matters is not the truly clipped pixels but those which look clipped. Using Lightroom or other software tool is not enough even if, again, you can’t do much to fix it when you shoot JPG. That’s a good transition to the next point.

Clipping is not the same animal when you shoot RAW or JPG

I believe it does make sense to differentiate JPG from RAW images when it comes to clipping. For RAW images, with modern sensors, clipping images is rare. Or you really do it on purpose. Or you have no idea how to use your camera’s exposure systems! The below example shows how tolerant sensors are now to clipping:

Very high contrast image with my son in an hotel room, completely in the shade. The skyline behind him is of course much brighter. This is really an extreme case and with a good but really mainstream full-frame sensor (Nikon D750 in this case), there is almost no clipping shooting RAW.

I know it is not so simple and you can clip some parts of a photo despite your goodwill and expertise whilst shooting RAW. My point, however, is to say it is rarely a problem and it is easy to identify and to anticipate as it will only concern extremely high contrasted images.

When it comes to JPG, this is a totally different story. It can be easy to clip parts of an image and it can be difficult to fix it as we have seen above. What matters is first to know quite well how to detect that the image will have some clipping. Second, you need to know whether it is a problem for your image or it is not. There is no good answer to this (from my perspective, though, it will very often be a problem). One approach would be of course to shoot RAW anytime there is a risk of clipping, just to have more latitude in the process, but it is not always possible or desirable. At least you know what to do. So, it looks important to understand the causes and the consequences of clipping and how RAW can fix it while bringing the usual inconvenient of shooting RAW (processing time, file size, buffer limits, …). If you don’t shoot RAW, you normally have reasons for this choice. This is a good transition to the next point: this is where good and bad clipping matters as well in your decision.

The Good clipping and the bad clipping

The bad clipping is the one you should not get. Just expose better your image by underexposing it when you have bright parts or underexposing it when you have potentially too many black clipped pixels.

The good clipping is just inevitable. Below an example:

When we analyse the image below, we can see we have both blacks clipped and highlights burnt. In red the burned pixels, in pink the “visually” clipped ones. In dark green the black clipped ones and in light green the “visually” clipped.

Having significant both red and green zone just say you are going beyond the capabilities of your sensor. Just buy a better one with a higher dynamic range… or use an artificial way (flash, umbrella, filters …) to decrease the contrast, which of course is not always possible or desirable depending on the kind of pictures you are shooting.

Conversely, if you aim at having high (or low) keys images, the result will be clipped, fair enough but the pre-processed images – before you start to work at them, the RAW images or the JPG out of the camera should not be clipped. And to illustrate this, a cute gallery that I like of high keys-on-purpose images:

Knot
Gallery on Flickr of white and high key images

I like these images but I would not bet they were clipped out of the camera.

Good principle: clipping is bad

Long story short, it will be difficult to convince me clipping is not bad. Indeed, if you are looking to shoot high or low keys image or if you want to stylize your images, that’s more a post process thing. If you know what to do, you can argue “I clip on purpose” but most of the time, clipping is just bad. Your sensor doesn’t provide information any longer but a very black and white approach of the reality. What you will do in the post process is a different discussion, when you shoot, and you anticipate clipping, unless knowing exactly why, you should just take whatever it takes to limit it (thanks to under/over exposing or bracketing) or avoid it (same actions + RAW + stacking/HDR).

Conclusion and summary

Let’s start by another example. From my perspective, this image below is poorly exposed, over clipped in a white grey ugly sky:

The city of Mopti, Mali

The light was terrible, due to some haze caused by hot air. This image looks ugly to me whereas Mopti is such a dramatic city and I have tried to post process it, there was no way to fix it (I was travelling, was short on time, and I did not see a way to avoid clipping). Light is bad, it is what it is.

My point: clipping is (very often) bad even if you can’t avoid it. They may be some counter examples (try to shoot an image of a polar bear in the artic without clipping the snow…) but they demand to have at least understood how to produce a pleasant image and taking counter measure to reduce the visual impact (shooting RAW, shot only when there are some shadows to produce some darker zones, …).

Tokyo Japan, RAW image underexposed by 1.5 EV, no final clipping whilst the original image before post-process looked challenging with both under (the sphere) and over exposed parts (backlight windows).

When you always shoot the same kind of picture, you know what you are doing. You don’t really need the following conclusion as you have no problem to deliver images you are familiar with. But it is also good in life to try new things. And it is good to remember some good principles because when you shoot new subjects, in a new way, in new places, you will have many reasons to fail delivering great images. It is good to remember some basic principles. Beyond all the discussions and remarks, I like to remember something easy not to forget:

Shoot whatever you like, but clipping is bad.

The more you know how to detect it, avoid it or at least manage it, the better. It is not a fight between art and science, it is about discipline.