The short answer is that there's no such thing as "should" in photography. Reality is that your camera (usually) gives you two output formats (sometimes three), and there is really no reason why you must forbid yourself to use one or the other. I hope that takes care of the "should" part. Now, which one do you want or need is up to you and your particular shooting style. In this post, I will try to enumerate the key points regarding the raw file format, and what it brings to the table.
This article is perfect for you if you are new to raw files and would like to know more.
What is a raw file?
A raw file is not an image format. (It is also not a religion, even though it's hard not to get that impression when you look at what's going on on the Web.) You may see an image when you browse through raw files in the image or file browser, but that's not the contents of the raw file. Rather, it's a Jpeg preview generated by the camera, embedded in the raw file as a preview.
Raw files contain the raw image sensor data (that's why they are called 'raw'). They also contain the Jpeg preview of the photo, as well as some metadata about the exposure and similar, just like the EXIF information in Jpeg files.
Because the raw files contain sensor-specific data, each camera manufacturer has their own raw file format: Nikon has NEF, Canon has CR2, Sony has ARW, Fuji has RAF, Olympus has ORF, and so on. There is also a raw file format called DNG (short for 'digital negative') that was developed by Adobe as an open standard, and is used by several camera and phone makers like Ricoh, Samsung, Sigma, Leica, etc.
The raw data is literally raw data, so it not possible to consume it without converting it into a visible image first.
Also because the raw data is just raw sensor data, some types of camera output cannot be recorded as raw. For example, images that are composed out of multiple shots (HDR, panorama, etc.), or images that have special filters applied to them, or images using aspect ratios other than the aspect ratio of the sensor (e.g., square), digital zoom, etc.
What do you do with raw files?
To do anything useful with the raw files, you need specialized software which we call "raw developers" or "raw image editors." The most commonly used software that serve this purpose are Adobe Lightroom and Capture One. But there are many others, such as DxO Photo Lab, Luminar AI, Darktable...
The software is used to 'develop' raw files (this is a term that comes from film photography, where paper prints were developed in a lab using light and chemicals). The process is similar to regular photo editing, but limited converting the raw file into a normal image file rather than complex photo retouching. For the most part, they do not replace tools like Photoshop.
Unlike the pixels in your image files, the camera sensor does not literally record each pixel with full color information. A pixel on the sensor (it's not really a pixel, but let's call it that for simplicity) records the light intensity of only one color, either green, blue, or red, and the values of the adjacent pixels are combined to build the pixel's final color. This process is called 'demosaicing', and it's one of the first things a raw developer does with your raw file.
Most camera makers use the same kind of color pattern on their sensors but there are also sensors that don't (e.g., X-Trans or Foveon X3). Therefore different demosaicing strategies may be needed, and this is why, unlike standard image formats like Jpeg, raw developers do not necessarily support every raw file out there.
When a raw file is first opened, the result is usually very similar to a Jpeg image that the camera would produce under the same conditions. This default appearance is not defined by the camera manufacturer. Instead it is a more or less accurate simulation by the software. In some cases, the camera's Jpeg look can be impossible to faithfully recreate (but you would be shooting Jpeg if that's what you want anyway). The next step would then be to adjust different parameters in order to achieve the desired look. Common adjustments include:
- Exposure and white balance
- Crop, rotation, and perspective correction
- Highlight and shadow recovery
- Brightness, contrast and tone curve adjustments
- Noise removal
- White balance and color tuning
There is nothing in the list that stands out as unique to raw development. All of these operations can be performed in traditional image editors as well. The main differences are the degree to which the parameters can be tweaked and the quality of the result. This is because the raw file contains more image data than could possibly be represented in a Jpeg or a TIFF file. Typically, raw files will record color intensity information in 12- or 14-bit resolution as opposed to 8-bit in Jpeg. This gives 16 to 64 times as many shades of color in each channel in raw files compared to Jpeg. When you work with raw files, you will usually notice that the file contains much more than what's visible on the screen as you play with the sliders.
Highlight was quite hot on the image shown above. Attempting to do highlight recovery worked quite well in the raw format (right), resulting in nicely saturated colors in the sky. In contrast, attempt to recover highlights in the Jpeg file (left) results in greyish tones in the sky, and overall darker image.
Although raw developers may look like normal photo editing software, one of the key differences is that they perform the so-called 'non-destructive' image editing. Rather than physically modifying the pixels in your image, the software only shows the preview with all the settings applied. The original image is intact (because the original is not an image, but just sensor data). You can therefore modify settings on the fly, turn some edit on and off, etc. If I don't like the highlight recovery, I can tone it down after I've preformed a bunch of other operations. I can also quickly toggle it on and off to see the before and after. Because of this behavior, it is virtually impossible to irrecoverably ruin an image such that you would need to start over. If you are familiar with Photoshop or similar more advanced image editors, this is similar to the adjustment layer feature.
The non-destructive edits are similar to recipes. In fact, some raw developers support saving individual recipes as files so that you can load them later. Most raw developers support copying adjustments from one file to another, or saving them as presets. This can greatly speed up the workflow when working on a batch of similar images.
Note that not all image manipulations benefit from the raw file format. For example, fixing the white balance and minor simple color tweaks can usually be performed on Jpeg files with no noticeable degradation of the output quality.
How do you expose for raw development?
For the most part, shooting for raw development is not much different than shooting 'straight' (I use this term for shooting Jpegs only). However, to maximize your chances of getting great results from your raw file, you should pay attention to what your camera is capturing as opposed to what your eyes see on the LCD or through the viewfinder. The preview images you see in the camera are the same thing as Jpegs, to the extent of the display's rendering capability, and they don't tell the whole story.
To develop our raw files effectively, we need to grab as much image data as we can. When it comes to shooting raw, histogram is your best friend. It gives us a good indication of the light we were able to net.
If your camera supports it, enable the histogram in your viewfinder (many mirrorless cameras do). Some cameras do not show live histograms, but instead you have to display the image and check the histogram there. In most cameras, histograms are calculated based on the jpeg rendition of the image. To get most out of it, you will need to select the settings that yield the highest dynamic range. Usually, this is some flatter picture profile ("Flat", "Neutral", etc.) and enable features that extend the dynamic range (e.g., Active D Lighting). Having said that, you can also just shoot with the standard picture profile as you will always have more wiggle room than the histogram suggests.
Simply put, a histogram shows the amount of pixels at different exposure levels.
On the far left we have the pixels that would be completely black, and on the far right we have pixels that would be completely white. The center of the histogram represents our midtones. I will not go into the details of how to read histograms, as that's not that important for this post.
The above histogram is of a picture that is more or less correctly exposed. There is no clipping in the shadows or highlights. Clipping means that the light was too weak or too strong for the image sensor to record useful data at the selected exposure settings.
Here is what the histogram looks like when we have clipped highlights.
The histogram looks cut off at the right edge. We see that there is plenty of room in the shadows, so what we can do here is to reduce the exposure (close the aperture, increase the shutter speed, decrease the ISO) to bring the histogram close to the ideal.
Let's take a look at the histogram of an exposure that is clipping in the shadows.
Similarly to the highlight clipping, again we see a histogram that is cut off at one end, this time the left end. Again, there is some room on the right, in the highlights, so we can increase the exposure a bit (open up the aperture, decrease shutter speed, increase ISO) to prevent clipping.
What if the histogram clips on both ends, though?
Regardless of how flexible the raw files are, there is a physical limit to the range of exposures the image sensor in the camera can record. Even the best cameras can only record around 14 to 15 stops between the darkest and the lightest point in the image. An average digital camera will shoot with around 10 stops of dynamic range. Therefore, in most cases, the range of brightness in the scene will normally far surpass the camera's capability if you are not careful with the composition and/or lighting.
When this happens, we have no wiggle room to recover both the shadow and the highlights. We have to either change the composition to exclude the parts of the scene with extreme exposure values, or use graduated ND filters to physically modify the brightness levels hitting the sensor, or sacrifice either the shadows or the highlights.
In my photographic style, I intentionally set the light up to increase the depth of the shadow areas. This means that I'm usually looking at histograms that are along the lines of the last examples. I sacrifice shadows to keep details in the highlights. Here's what that looks like:
As you can see, sacrificing the shadow detail is not necessarily a bad thing.
There is a technique called "expose to the right" (ETTR), a rule-of-thumb exposure strategy. With ETTR, we are supposed to expose the image at highest exposure value that does not clip the highlights. This moves the pixels on the histogram to the right, hence the name. Usually, this is done by enabling exposure warning in the camera's settings, and then increasing the exposure (decrease the shutter speed, increase the ISO, open the aperture) until we get a warning for the highlights, and then dialing back the exposure just enough to make the warning go away. This generally results in a lighter image than a correctly exposed one. The idea with this approach is to reduce the noise in the darker parts of the image. It gives you images that you absolutely must edit later, and reduces the margin of error in the highlight. With today's cameras and their relatively great low-light performance, I would suggest to expose normally and adjust exposure just enough to avoid clipping if any. This works well for today's cameras.
Why doesn't everyone just shoot raw?
Not everything is automatically better when shooting raw. Consider the following comparison:
If you look closely at the details on the roof and in the background, you will see that the raw file (right) lacks definition compared to the Jpeg (left). The camera-processed Jpeg looks a bit clearer. What would happen if we adjusted our settings in the raw file, though?
With a few tweaks, the developed raw file (right) now looks much better than the Jpeg (left).
Although the results are better, it involved some work.
Some people are perfectly happy with what the camera does for them. This is the work they did not have to do, and besides, they paid for the in-camera processing capability when they bought the camera, so why wouldn't they let the camera do the money's worth of work for them? They prefer the look of the camera Jpeg over what raw developers offer (and that can sometimes be difficult to replicate faithfully in the raw developers).
Another reason people may not want to shoot raw is that the software used for raw development is either very expensive, not very good, or not very easy to use (or any combination of those). For example, Capture One costs hundreds of dollars per license, and even their annual subscription price isn't something you can file under casual expense (in two years it will easily add up to a price of a brand new lens!). Many camera makers supply their own raw developers, but those tend to have less features than the professional software. Finally, open-source software like RawTherapee and Darktable, while being free, aren't exactly easy to use. While using expensive software is perfectly acceptable for a professional, not every photographer makes money from photography.
The cost of software is not the only cost when dealing with raw files. You will need faster cards if you want to use high-speed bursts, and you may want larger cards, too, if your camera has (too) many megapixels. The storage space in your computer also becomes a real issue with raw files. For example, the Fujifilm raw files on my 26Mpx X-T3 takes up four times as much space compared to a Jpeg in normal quality.
Finally, raw development is not a silver bullet. As you saw earlier, there are things that cameras do well. After all, camera manufacturers have intimate knowledge of the sensor and lenses, and are able to provide superior correction of some of the known (to them) flaws of their systems. Distortion correction, chromatic aberrations, light falloff (vignette), and similar flaws can be dealt with in the camera more effectively. In some cases these fixes are applied to raw files as well, but not in all cameras.
Why would some people only shoot raw?
There are photographers who swear by raw. They are enamored by the flexibility raw development gives them. I see two main reasons such people use raw files.
One is simply lack of skill. They frequently end up with compositions that are far beyond the capabilities of the Jpeg format, and they compensate for that by shooting raw and fixing things later. I've also done this until I decided I would rather learn how to expose correctly and went Jpeg-only.
Another type of raw shooter is a photographer that needs some safety margin due to the nature of their work. They may shoot under conditions that are beyond their control, and they shoot raw to increase the margin of error. Some professionals prioritize reaction time over precision: they set a ballpark exposure in manual mode and never bother to change it during a shoot. This saves battery life and allows them to worry about composition and moments, rather than the camera settings. Once they are done with the shoot, they adjust the exposure in the raw development phase, usually in bulk.
There are, of course, more reasons. Sometimes the camera Jpeg may not be pleasing to the photographer (e.g., older generation Sony cameras). Using raw means the the image can be completely adapted the the photographer's tastes without degrading the quality of the final image. Older cameras (especially older DSLRs) may not be very good at metering, and constantly miss the correct exposure. In such cases, photographers may find shooting raw to be simpler than constantly correcting the camera.
As you can see, raw files offer a great deal of flexibility for the final image. They may be used to recover extremely exposed part of the image, improve image sharpness, clean up noise, tweak colors, and more. If you are working in a particularly unpredictable environment, you can use the raw format to secure some margin of error. On the other hand, raw development is not for everyone, particularly for those photographers that are time-constrained or cost-conscious.