The JPEG - RAW Debate - part 4 of 1 2 3 4

by Mike McNamee Published


This image is the most convincing demonstration of the benefits of RAW that we can provide. As shot [1], the image was 2/10th of a stop underexposed and 300°K too cool, this amounting to about 7RGB points askew in Adobe RGB 1998 - a pretty decent exposure and as good as you might expect to achieve in less controlled conditions. The screen grab [2] shows that the subject's white shirt is clipped, indeed as shot, the RGB values were 255R; 252G; 255B. Reducing the exposure in RAW [3] reveals that there is in fact a certain amount of usable detail available - not very pretty, but useable with some care and masking. The same shot, adjusted as a JPEG [4], remained lifeless grey, regardless of how the curves were adjusted.


How does JPEG work?

The Joint Photographic Experts group, part of the International Standards Organisation (ISO), published a methodology for handling full tone range, colour images that has become the standard, universal format called after them - JPEG. In Windows it carries the extension .jpg or .JPG.

The methodology is one which compresses the data, ie it makes the file smaller whilst attempting to keep the quality up (within limits). In order to work well, the compressor first has to convert the file into something from which the luminance can be isolated from the colour - not something you can do with an RGB data set. The eye is tolerant of compression of colour but not of luminance detail and therefore the JPEG method beats the hell out of the colour but leaves the luminance well alone! That is why a greyscale image only compresses to about of the size of an RGB not the ¹⁄₃ that you might expect. Having put the luminance information to one side, the algorithm (as it is called) gets to work on the saturation and hue of the data. The colour data are divided up into blocks of 8x8 pixels (ie 64 in total).

The maths then looks at the bit depth in the colours and makes a guess at how much it can be reduced. This is variable (in Photoshop via a convenient Quality Slider) and if it is over-zealous you end up with visible JPEG blocking in the image. This is what is known as a lossy process and you decimate pixels which are not recovered when the file is reopened. However it is also a sophisticated process which, when done correctly, leaves an undetectable scar on the image - this is the basis for people saying JPEG does not produce detectable damage. It is there, you just cannot see it at normal magnifications - in addition the printing process tends to slur the colours together, further masking the imperfections.

The compression ratios are variable from an almost lossless effect to 100-fold. If you save a JPEG repeatedly using compression the pixels get a bit more decimation each time until it will show at normal viewing distances.

This is an underlying reason for some of the confusion which surrounds the JPEG-RAW debate. JPEG can be very good or very bad depending where you put the slider and how many times you do it. This leaves plenty of scope for the debaters on both sides of the argument to build their cases!

Please Note:
There is more than one page for this Article.
You are currently on page 4 Contact Mike McNamee

1st Published
last update 07/02/2018 11:58:07

More Digital Articles

The Societies of Photographers Convention and Trade Show at The Novotel London West, Hammersmith ...
You have 119 days until The Societies of Photographers Convention starting on Wednesday 22nd January 2020

20% off Canon Photography eBooks