The proponents of camera raw point to the information that is recorded in the camera raw format, claiming that 14- or 16-bits of information in the sensor should allow much more flexibility in post-processing than the mere 8-bits of information per pixel in a JPEG image. On the surface, it would appear that a camera raw image should have 3-4 stops of additional lighting information that is discarded when the 8-bit JPEG is created. If this is so, then it should be possible to recreate the same image if the exposure is within 3 or 4 stops of a proper exposure. Today I’m going to put this assertion to the test.
All images shown here were captured using RAW+JPEG simultaneously. The JPEG images were saved at the highest possible quality. The pictures were all taken hand-held, so there may be slight perspective or focus differences between them that should be ignored for this comparison.
In this experiment, I took this picture with proper exposure settings (Fig 1). To my eye, the raw image seems a little richer in color and the gradations in tone seem somewhat smoother, but both exposures are good and, if there were no side-by-side comparison, either would certainly be acceptable. It is interesting to note that to get these two images to appear nearly identical, I had to boost the Fill Light on the raw image by 37.
In this next series of photos (Fig 2) – all in camera raw format – I reduced the exposure compensation in the camera to force an underexposed image. Afterwards, I went into Lightroom and boosted the exposure by the same amount. If the original sensor information is retained with an additional 3 stops of exposure, then I should be able to pull each of these underexposures up to reproduce the baseline image. Interestingly, this wasn’t the case. The more underexposed the image was, the more I had to reduce the contrast to retain the same light/dark balance. The Fill Light parameter stayed about the same until I got the the most underexposed image, where I had to increase it to 47 – although that was probably because I had maximized the amount of contrast reduction that Lightroom allowed. You will also notice that the colors become more washed out as the photo was increasingly underexposed. There was no compensation that I was able to apply in Lightroom that would recapture the richness of color at a 3-stop underexposure that the original held.
Looking at the same series of photos, but this time in JPEG format (Fig 3). In this series I also had to compensate by reducing the contrast the more the image was underexposed, but not as much as in the raw images. Also, the richness of the colors did not alter as much. It is very interesting to note that the properly exposed JPEG image actually got the colors wrong (white balance settings did not change). The properly exposed image appears much more yellow than the underexposed images, and the actual color was indeed closer to the underexposed images than the properly exposed one.
Now, looking at the overexposures (Fig 4). This series of camera raw images starts with the baseline image on the left and goes to a maximum of 3-stops of overexposure. As I reduced the exposure back to normal, I had to decrease the Fill Light and increase the Brightness to compensate. The first stop of overexposure can be compensated for almost completely. There is very little difference between the colors in the two images. By the second stop of overexposure though, you can clearly see some color changes although the Recovery slider helped quite a bit, and by the third stop the highlights are completely blown out and nothing could bring them back.
In the JPEG images (Fig 5), no consistent control seemed to come into play as the images were more and more overexposed. In some I had to increase the brightness; in others, I had to reduce the contrast, in others I had to reduce both brightness and contrast. The result was that the first stop of overexposure could be readily compensated for. In fact, the colors looked better. Go figure. By the second stop of overexposure, the colors had clearly changed for the worse, and the third stop of overexposure was completely blown out beyond repair.
So, it is clear that camera raw is no protection against over- or under-exposure. Now let’s compare how camera raw compares to the equivalent JPEG images.
We’ve already seen how the baseline images are well exposed, although I feel that the camera raw image produced the better color and tone, it wasn’t a big difference. Here we can see the lower exposures with the camera raw on top and the JPEG equivalents on the bottom (Fig 6). I’m seeing a richness of color in the raw images that is missing in the JPEG images. Apart from that, it appears that every one of these was able to be corrected using Lightroom back to a proper exposure, albeit with some loss of vibrancy across the board.
Now, on to the corrected overexposed images (Fig 7). I’m seeing a lot more difference here. At one stop of overexposure, there is little difference in the corrected images, but when you get up to 2-stops of overexposure, the difference is significant, and neither image is really salvageable when you get to 3-stops of overexposure.
Now here’s something interesting: when I converted both of the maximum overexposed images to Black & White (Fig 8.), the JPEG image converted into a better looking image. The blown-out areas could be pulled back so that they were closer to the JPEG image, but it the areas that were blown-out in the JPEG image were more equally blown-out, whereas appears that there was some additional headroom in the raw image. Lightroom could correct the raw image so that it looked similar to the JPEG image by pulling down the yellows, oranges, and greens(!).
There’s one more aspect that I need to share with you in doing this experiment, and that is the issue of noise.
Noise typically becomes an issue when you are boosting the information that comes from the sensor. This is what ISO does – it amplifies the signal coming from the image sensor much like turning up the volume on the radio. On a radio, increasing the volume also increases the static or noise that is underlying the sound. You don’t really notice it at all until it is loud enough to hear. In much the same way ISO – and boosting the exposure setting in Lightroom – does the same thing for images. Suddenly, the minute variations due to noise in the sensor system are made more prominent and you can see them as a grainy texture in the photograph. This can be addressed in Lightroom through their noise reduction filters, but it is interesting to note that there is much more noise in the camera raw images than in the JPEG images that were boosted by the same amount (Fig 9). Once again, I think this is due to the camera raw having more information to pull from, but it is nowhere near the amount that the raw proponents would have you believe.
Based on this analysis, I once again find in favor of camera raw, but the advantage is not a big one. In most cases I find that JPEG images can be corrected almost as much as camera raw, although admittedly with some loss of color richness. Camera raw seems to handle overexposure better than JPEG, but only to a certain extent. Looking at these images side-by-side makes it easier to see the differences, but in many, if not most, cases these differences are not large enough to make a significant difference. Keep in mind that people won’t have these two views to compare against; they will only have your single image and that image will likely be just as compelling, just as good whether you use JPEG or camera raw. Now, that being said, I will also add that it appears that camera raw does a better job in retaining tone and color than JPEG does. For this reason, if I wanted to have the best possible image, I would use camera raw.
In this final comparison, I have an image of some pyracantha berries (Fig 10). Again, these images both resulted from a single click of the camera. I can’t say that I understand the reason that camera raw processes variations in tone better than JPEG, but this image illustrates the difference clearly. Both of these images were post-processed in Lightroom, but which one holds more appeal? I’ll leave the answer to the viewer for now.