The truth is that we are always shooting raw… even using the simplest compact cameras or mobile phones with camera. The difference is what happens inside the camera just after shooting.
Shooting raw means that the camera register in the card the data directly captured by the sensor without any kind of processing of that data. So the raw data contains the values just after the analog-digital conversion (to the bit depth or that A/D converter, 10, 12 or 14 bits as we already seen in the sensor article).
Every digital camera has a microprocessor that, as well as controlling the general camera functions, is specialized in image processing. In fact, many new features incorporated to the cameras in the last years (face and smile detection, in camera tone mapping (d-lighting and similar), etc.) are improvements achieved by the constantly increasing power of cameras image processors.
So, when we do not shot raw, the camera uses that image processor to save it as a JPEG (jpg extension file). This image processing in the camera covers some stages (once obtained the digital data and not necessary in this order, although it may be the logical one):
- Black point subtraction (as we seen in the sensor article, even in the total absence of light,the sensor returns greater than zero values). If it is not subtracted from the raw data the image looks dull. In that raw shot took with my Canon A-540 compact camera it is very evident:
|Without subtracting the black point||After black point subtraction|
- White balance. White balancing consists in raising or lowering each of the three channels (red, green and blue) trying to neutralize the image grays. This is necessary because the different light sources produces different colors when the light they emits reflects in the neutral surfaces of the images (the grays). Our brains are able to do that adjustment in a very efficient manner (otherwise, outdoor everything would have a cyan tint, because of the blue light scattering produced by the sky) because they know what should look neutral (pure experience) but the camera is not so smart and we have to tell it what colors should be considered neutral, unless we want to introduce creative color shifts in the image. It is not the camera registering the colors wrong, it sees light as it is, but our brain expects another thing. On the other hand, the fact that the sensor has double green sensels in comparison with reds and blues, makes the camera much more sensitive to green light, so an white unbalanced image looks too green.
|Without white balance||With camera automatic white balance||With dcraw automatic white balance|
- Interpolation. As we show in the sensor article, the raw data needs to be interpolated before presenting a complete image:
|Without interpolation||Interpolated in dcraw with AHD||Interpolated in dcraw
with PPG and postprocessed
|Interpolated and processed
by the camera
- In some cameras, tone mapping. As well as happens with white balance, our brain is very efficient doing tone mapping. In very contrasted situations, it is able to adapt very fast and give as a good register from shadows to highlights. Some camera tries to do something similar, raising the shadows and lowering the higher lights.
- In some cameras, lens distortion and vignetting correction.
- In some cameras, cropping.
- Output space color conversion and output gamma correction (typically reflex cameras let the user choose between sRGB and AdobeRGB, although compact cameras allow sRGB only). The purpose of this step is to prepare the image for visualization in a render device with limited color rendering. Specifically sRGB was designed for CRT monitors. As for the gamma correction, the need for it is caused by the non lineal response of computer monitors. The shadows are raised using an exponential function, that produces a non-linear image that looks linear to us in fact. Using the above image:
|sRGB color space, gamma 1.0 (without gamma)||sRGB color space, gamma sRGB|
- Saturation increase.
- Contrast increase.
- Sharpening increase.
- Image stile and artistic filters applying, generally the includes at least some saturation, contrast and sharpening preadjustments.
The final result is a JPEG image processed in a record time and with a good quality (even more in the reflex cameras and high end compact cameras, that have more powerful image processors), although the quality varies quite a lot from some cameras to anothers.
So, why should we want the raw file for if the camera is able to do a good quality processing? I believe the answer is already in the question: some of us do not settle in a “good quality processing” and want the best possible one.
And, what advantages and disadvantages raw shooting has? That is for another article in this series.
-  Unfortunately, some cameras are already processing the raw data. I say unfortunately because that corruptes the spirit of shooting raw, that, as we will see, implies having the full control of the process in order to achieve the maximum possible quality from our Camera. For example, Nikon cameras are subtracting the black point from the raw data, that makes later noise reduction more difficult; and some Sony cameras are filtering the green channels when using high ISO values, at least with the default parameters. ↩
-  Using the great CHDK, I have a pending article about it, in order to be able to shoot raw. ↩
-  It is more notorious than in a reflex camera because the black point, 31, is closer to the saturation point, 1023 (it uses a 10 bits A/D converter) than it will be in a reflex camera. ↩
-  The two greeen channels are usually processed alone at this stage. ↩
-  As well as being interpolated, the camera image processor has done some other image adjustments. It is very evident the noise reduction, the contrast increase and the sharpening. ↩
-  For instance, Olympus reflex cameras correct lens vignetting. ↩