Post-processing images and why it's important to shoot raw

April 3, 2014

One question people ask me constantly is if my images really looked that way when I took them or if they are 'retouched'. This is not especially easy for me to answer (it's probably easier to explain it in this article in fact). If I truthfully respond that they are not retouched people get the wrong idea of seeing a 1:1 depiction of what the camera sensor captured, and when I truthfully tell them that they are post-processed they immediately think of those heavily corrected fashion or beauty product images. Normally people lack the understanding what post-processing really means and why it's a very important part of the photography work flow, so I'd like to start by getting some misconceptions out of the way.

What post-processing means

In 2014 more than 99% of all images are taken with a digital device. In this case digital means that the image recording process works basically as follows: A measurand (light / a stream of photons) is detected by a sensor that produces an electrical output - the sensors task is basically creating a distinct relation between the input and the output signal. The output signal is then conditioned (e.g. amplified or filtered) before it is digitized into concrete quantification steps so it can be saved as digital data.

Before the digitization the whole process is strictly analog so the funny thing is that when we talk about digital cameras we are actually talking about analog cameras – just the post-processing and the storage of the data is digital.

After the A/D conversion some sort of processing is always employed, this can be for instance removal of defective pixels or - to a certain degree - noise reduction. Due to the fact that this already happens after the actual capture it can be understood as the first instance of post-processing. At this point the image is a so called digital negative which is still far from being ready to be viewed or printed and a variety of further steps have to follow in the processing pipeline.

This diagram shows the main processing steps of digital images. 1. Sensor data, 2. Demosaicing and creation of RGB channels, 3. Raw image, 4. Finished image after post-processing. The outlined steps 3 and 4 can be done manually on the computer if a raw work flow is employed.

The user of a raw format capable camera basically has two options how to continue:

When you shoot JPG

Shooting jpg is the modern equivalent to letting your film getting developed in the photo shop. It's doubtless convenient because someone else is doing the work for you but you have basically no control over the outcome. Nowadays the camera develops your pictures to a positive format such as jpg in a fraction of a second so you don't have to wait for a couple of days to see them. If you don't like the result you can change some camera settings and retake the picture, if the situation allows it of course.

When you shoot RAW

Shooting raw is the modern equivalent to developing your film yourself in a darkroom, where the darkroom is replaced by a computer program and you don't have to deal with health threatening chemicals any more. Of course this is much more work than just letting the camera present you with a finished, viewable image, but it also gives you control on the appearance of the image and the opportunity to correct flaws of your technique or equipment at least to a certain degree. Since I manually post-process all of my images, shooting raw is mandatory for me.

I'd like to quickly sum that up: By choosing jpg format you can decide that the camera should do the remaining processing steps from your captured negative to the viewable positive by itself and by choosing raw the image developing software on your computer will do them under your guidance. What is however impossible is to not post-process the images since sensor raw data is simply not viewable.

The difference between post-processing and retouching

The difference between post-processing and retouching is pretty simple to explain. If you change the appearance of an image by emphasizing elements that were already there in the unprocessed negative but probably hardly visible or looking different that you would like them to look you are post-processing. If you change the content of the image by adding or removing elements or intentionally changing the form of your subject (I'm looking at you liquify tool) you are retouching. I however do not consider some processing steps that would technically fall into the retouching category as retouching, such as cloning out sensor dust or other minor blemishes, applying lens corrections, creating panoramic images, HDR or focus stacking.

The most common post-processing steps and their influence on the photograph

The first processing step in my imaging pipeline is the lens correction. Lenses introduce aberrations such as distortion, vignetting and chromatic aberrations, which of course degrade an accurate reproduction of the photographed scene. Distortion correction requires a recalculation of the image and reduces resolution. For this reason it should only be applied when there are straight lines visible where the distortion becomes obvious. Vignetting can sometimes even be a desired component of the image which directs the viewer's look to the brighter area in the center of the frame.

Next are various exposure adjustments. Sometimes images are over- or underexposed either deliberately (e.g. ETTR) or because of user/metering error. It is also desirable most of the time to adapt the gradation and the contrast, which means to brighten/darken parts of the frame depending on their initial lightness.

One of the great advantages when shooting raw is the complete freedom to choose the white balance during post-processing. I always use automatic white balance in camera and never hassle with adjusting the WB setting in the field which lets me concentrate on the important things that can't be adapted later.

Finally some noise reduction / sharpening can be applied.

In the images below I have put together some examples before and after post-processing. Hover your mouse over them to see the final results (or tap on them if you're using a mobile device).

Please note that even the 'unprocessed' pictures below have probably been cropped and lens corrections have been applied to achieve better comparability.

Here the typical adjustments have been made. Mainly contrast, clarity and color were a little intensified and a bit of vignetting has been added.
This is a very nice example why processing images should be mandatory. The building was lit very brightly so the contrast to the dark foreground and sky is quite huge. In post, the dark and the bright regions could still be saved and so this image with maximum dynamic range was possible. Please note that all the information you can see in the final picture is there right from the moment where it was taken - there was just some post-processing needed to properly accentuate it.
When I was on my way home from a long and sunny day of hiking and shooting, the sky started turning dark, giving it a somehow menacing mood, and it began to snow. The camera didn't quite catch the dramatic situation that was present during the shot so I mainly increased the contrast between the white snow and the gray sky which also let the falling snowflakes show more clearly.
A dark object in front of a bright background is a challenging situation for every sensor. Using at least one stop of exposure compensation to the right (brighter) is absolutely required for recording the dark feathers of the eagle but still – the raw image leaves much to be desired. In post I lightened the underside of the bird to make all the detail of the plumage visible.
This image is a prime example of overexposure due to user error. It was taken just moments after the image I showed above. I had set the exposure compensation to +1 1/3 EV which was appropriate while the eagle was still circling in the sky, but when he suddenly decided to take a break I simply didn't have time to adjust the settings accordingly. The evaluative metering algorithm bumped up the ISO speed due to the suddenly darker background which led to a massive overexposure. By darkening the photo (which I already thought was lost after reviewing it on the camera screen) by nearly two stops, it could thankfully be saved.
Especially when dealing with very high contrast situations like the one shown here, getting the most out of your sensor is indispensable. The detail in general and especially the small droplets were emphasized by adding a little fill light to the head of this male Goosander, while the bright feathers were darkened a bit. Additionally the saturation was increased and a bit of vignetting was added.

But it's also possible to process jpg, right?

Yes, of course you can always open your finished jpg image in a graphics editing program and apply the same changes that you would while developing a raw file, but you probably won't achieve the same result for the following reasons.

A raw file is in principle the accumulation of all the information collected by the sensor during the exposure, stored with lossless compression and full color depth, which results in a 20-30 MB file size. Most modern DSLRs have a color depth of 14 bit which means they digitize the electrical output signal from the sensor with 214 (=16384) concrete quantification steps. This maybe sounds a bit complicated on first glance but is actually quite easy to understand if you think about it like this: Every pixel consists of tree color values – red, green and blue. A single pixel is not simply red, green or blue but it contains something of every three components. This 'something' is defined by a numeric value between 0 and 16383 (which are 16384 or 214 possibilities) so for example a pixel could consist of RGB {14687, 578, 8857}.

A jpg file on the other hand is a 1-3 MB file with lossy compression and a color depth of 28 (=256) concrete quantification steps. This means every pixel has RGB values between 0 and 255 (which are 256 or 28 possibilities). The 14 bit color RGB {14687, 578, 8857} remapped into 8 bit would be RGB {229.484375, 9.046875, 138.390625} - the problem here is that bits can only be integers so it would be approximated to RGB {229, 9, 138}, which means that the approximation that has to be done for creating a jpg file leads to a slightly different color than the original raw image contained.

Additionally color changes cannot be resolved with the same level of detail so if the contrast of an image or of a part thereof is increased, posterization can occur. A pixel with 8 bit color depth can display 16,777,216 (=2563 or 256*256*256) different colors while a 14 bit pixel can display 4,398,046,511,104 (=163843) colors which is 262,144 times more than the jpg can.

Let's use the following shot of an urban landscape as an example for jpg processing. The sky above the city turned out very bright with very constant brightness and only very little differences in color while the city is in comparison really dark. In post-production the sky has been darkened to be able to see the detail in the clouds and to bring back the blue sky color that you could see with the bare eye but appears to be rather white on the image, and to lighten the dark buildings in the frame. I tried to apply the desired processing on the jpg and on the raw image simultaneously so that the difference can be observed.

This picture comes straight out of camera with no external processing - apart from lens corrections - applied. I tried to use an exposure setting to get the most out of this difficult lighting situation. For best possible shadow noise performance the exposure was set to the maximum brightness that keeps the sky from blowing out – a small area around the sun excluded.
This is the raw image after post-processing. As you can see the recovery of the overexposed sky turned out pretty well – giving it back its color, and also the detail in the dark regions of the image is preserved. The whole dynamic range the sensor can deliver is visible here.
Here you can see what happens when you try to make the same adaptations to a jpg file. The color of the sky couldn't be reproduced and turned into an ugly grayish tone, the buildings lost a lot of contrast and detail and the overall image looks somehow washed out and dull. When the image was remapped from 14 bit to 8 bit simply too much information was lost for a successful recovery of the dark and light regions.

The differences are already quite visible in these web sized illustrations and when we take a closer look they get even more apparent.

This is a 100% crop from the center of the frame. The raw image on the left shows a nice color gradient in the heavily darkened sky which was made possible by the 14 bit color depth, where the jpg image shows severe posterization (large regions with the same color and harsh transitions between colors). Due to the fact that post-processing jpg is destructive, some artifacts from multiple saving can also be observed around the clock tower.

The above photograph is also an interesting example for the application of post-processing in general. Some people may say that this image is heavily or even over-processed which is a completely understandable statement, but if truth be told the processed image is much closer to reality than its out-of-camera counterpart. For the human eye the above scene did not include an exceptionally great dynamic range - the bright sky and the dark buildings could be seen easily and with great detail and full color (probably due to the fact that the brain also heavily post-processes what our eyes see). Although the raw image obviously includes all this information, the standard settings for interpreting the data are not able to recreate the originally captured scene appropriately.


What I'd like for you to take away from this article is the understanding that post-processing is a natural and indispensable part of digital photography and it does not have to mean falsifying or distorting. In contrary, it can be used by the photographer as a valuable tool to recreate the original scene or sometimes even more importantly the mood, where the camera fails by itself. Cameras don't lie but it's important to keep in mind that they don't tell the truth either.

It's also essential to realize that proper post-processing can only be achieved by shooting raw and making corrections manually on the computer. Of course this means more time and work has to be invested in every single photo, but the results will definitely be worth it. Think about it this way: If a picture isn't worth five minutes of post-processing work then it's most likely also not worth the disk space it takes or the paper it's printed on. For a truly great image every minute trying to get the most out of it is a minute well spent.

Shooting jpg and trying to 'get it right in camera' is a quite unrealistic approach in my opinion because it's based on the incorrect assumption that reality can be exactly reproduced by a rasterized, two dimensional array of integer values and therefore doesn't have to be interpreted by someone that understands the message the picture is supposed to convey.

After capturing an image, for me there are only two options how to proceed – either deleting or post-processing it. If you want my advice – always shoot raw and make a habit of spending some time on post-processing your images, the results will speak for themselves.

Back to top