Sylvanite
Member
In the giving and taking of digital photography advice, I frequently see (or hear) someone assert that shooting in "RAW" mode is necessary for good results. Unfortunately, I also see (and hear) a lot of misconceptions as to why. I thought I'd describe how "Camera Raw" photos and photo editing actually work - and hopefully reduce some of the misinformation out there.
Let's begin with a description of what a digital camera does when it creates a digital image from an exposure.
Some real advantages of editing RAW images:
Some fallacies about RAW images:
I hope that helps,
Eric
Let's begin with a description of what a digital camera does when it creates a digital image from an exposure.
- First off, the camera's sensor records the amount of light (as seen through red, green, and blue filters) that falls on each picture element (or "pixel").
- We see colors as we expect them, but the camera sensor sees them as tinted by the ambient light. To correct that, the camera applies a formula for "white balance". That way, color tints appear more like what the eye sees.
- The camera also applies a "recipe", such as "landscape", "portrait", or "standard" to further adjust the image. The "landscape" recipe, for example, typically adjusts colors to accentuate blues and yellows (so our photos have a prettier sky). It also sharpens the image more to call out the details. "Portrait" recipe shifts the colors to better portray skin tones, and sharpens less. "Standard" recipe typically favors greens and reds more, and applies moderate sharpening. Many cameras will have other recipes as well, and some allow the photographer create custom ones.
- Then the camera converts the photo to the target file format, and stores the image. Information can be lost two ways during this stage. If the file format has less "color depth" than the sensor captured, then the image will lose some dynamic range. If the file format is compressed, then (depending on the compression scheme and the amount of compression) image quality may be reduced. A camera that stores images as 16-bit TIFF files typically suffers no loss, but a highly compressed JPEG image might show visible differences (in contrast and color depth). It's a trade-off between image fidelity and file size.
Some real advantages of editing RAW images:
- My current DSLR has a 14-bit color depth in each color channel. That is, it records over 16000 increments each in red, blue, and green. JPEG images only store 8-bits per channel (or 256 levels each). When converting from 14-bit color to 8-bit, you lose a lot of in-between shades. Some editing steps are better done before this conversion. If, for example, the exposure was slightly off, one could adjust brightness while more detail is present. One can also adjust shadow or highlight detail so that they persist when converted.
- With the right software, one can try out the camera's different recipes and see which one looks best.
- With greater color depth, sharpening algorithms do a better job and produce fewer artifacts. Therefore, judicious sharpening before converting to 8-bit color can yield better results.
- If one is going to make large prints from an image, Photoshop allows editing in 16-bit color depth, so no detail is lost at all. If the target media is web based, however, 8-bit conversion will be necessary at some point.
Some fallacies about RAW images:
- "Raw images have greater resolution". That is just plain false. My current camera, for example, can save both in RAW and JPG format in three sizes. The number of pixels (i.e. resolution) is exactly the same whether saved in RAW or JPG format.
- "RAW images permit additional editing controls". That isn't true either. In Photoshop (and Photoshop Elements), one can "Open As Raw", and all the RAW controls are present, even if the original image was in a different format.
- "RAW images can correct for improper exposure". That is only true to a limited extent. I've seen people claim they can correct for a 2-stop exposure error, but in my experience highlight detail is gone with a 1-stop overexposure - and no amount of post-processing is going to retrieve it. Trying to recover underexposed shadow detail introduces a lot of noise.
I hope that helps,
Eric
