Why go RAW?

 

Shooting in RAW format imposes several inconveniences like:

 

-          large file sizes

-          long lag between shots

-          requirement for post-processing

-          no support for RAW by photo labs

 

that make this format impractical in many situations, so why bother?

 

The full list of benefits will vary from camera to camera, but what’s stated here should be true for most Minolta DiMAGE cameras.

 

Colour depth

 

RAW is the only format in which the photos are recorded with 12 bits per pixel of colour information, for each of red, green and blue channels. JPEG and TIFF formats (as out of the camera) give you only 8 bits of colour depth. This means that in RAW we have up to 4096 distinct values of data in each channel, as compared to only 265 in the other formats. While the immediate difference is quite hard to notice, the post-processing can quickly lead to posterization, especially in very light shades. Having this extra depth available makes photos more resilient to post-processing.

 

For example if the original photo is too dark, you can make it much lighter without noticeable degradation of detail level. Making it 4 times brighter still gives you 1024 levels of detail, while on a 8-bit photo it will reduce down to 64. It’s true that noise is amplified together with the detail, but it looks much subtler when there’s enough colour depth.

 

Detail level

 

In a 5Mpixel camera it doesn’t matter whether your image is JPEG or RAW if all you’re going to do with it is display it on the web or print in small format – there’s more than enough detail to handle these situations. But if you consider heavy cropping of your images (for example to combat the zoom limitations) or large prints (bigger than A4), then all the detail you can get becomes very precious.

 

All photos are originally shot as RAW. If you select a different format, the RAW image is converted in-camera to the desired format. The in-camera interpolation algorithm is a compromise between reasonable quality of output and the time it takes to convert the RAW image into the final RGB map. It has to work in a limited amount of memory and has to be quick enough for the camera lag to be minimal, and it has to be good for all the images.

 

When you shoot in RAW, the camera stores the image as it was captured by the sensor, and all the conversion takes place in your computer after the image has been downloaded.  The difference is that at this stage you can afford the time to run less efficient but more sophisticated algorithms, and experiment to select one that gives you best results for a particular photo. Typically the algorithms differ in their ability to produce clean edges and handling of noise – depending what your photo contains, some will produce better results than others.

 

 

White balance

 

The “viewing-ready” formats are fully processed in-camera as soon as the photo is taken. If you shoot with a wrong white balance setting, or the auto doesn’t guess the right settings for you, your only recourse is post-processing with seldom a satisfying result, and very quickly leading to posterization (you have only 8 bits of colour data to play with).

 

RAW images are white-balance independent. The camera’s WB setting is stored with the image, but it does not affect its content. Consequently, you are able to play with the colour temperature until you are completely satisfied with the result, and in the end still have all 12 bits of colour data available to you.

 

Other benefits

 

What has been said about white balance, applies to colour space as well. The sensor has its own characteristic and it needs to be adjusted for proper display on other devices. If you are really particular about colour space accuracy, you could create a colour profile of your individual camera rather than relying on the series average embedded in your camera by the manufacturer – there are tools that can help you doing that. Such colour profile can be fed into a RAW converter, but I’m not aware of any camera that can download a colour profile (although some may be individually tuned in factory).

 

Similar kind of benefit is related to the ability to select gamma profile and image contrast before the interpolation is done.  It has significant impact on the noise level and uniformity of flat surfaces. Outside of RAW the gamma curve is fixed, and you have only one shot at the contrast setting, which is often too granular to be useful.