All JPG Images Are Cobbled At Birth Everybody knows that JPG images are prone to compression artifacts, meaning every time an image is opened, altered in any way, and re-saved, file re-compression causes additional detail to be lost. But, did you know that if your JPG images are destined for print, those compression/pixelating issues may be the least of your problems?
There is a little-understood JPG gremlin lurking behind printed digital images causing them to print soft and indistinct in the shadow areas of images. Chances are, you’ve seen evidence of this gremlin in publications and simply accepted it as normal.
There are two contributing factors involved First, JPG formatting conforms every camera image into a pre-determined, one-size-fits-none “linear curve” (I think that’s an oxymoron), and it does so regardless of the original scene lighting. This is somewhat akin to voting in a country where there is a single candidate on the ballot. Your choices are slim and none.
Second, regardless of how much data your camera captures, JPG formatting severely compresses the image’s bit depth. Limited bit depth causes very dark areas of photos to virtually plug-up, appearing very dark and muddy. When there are not enough tones to separate, the detail in those areas vanishes.
Bit Depth. Bit depth is technotalk for editable levels of picture information. Typically, digital cameras capture at least 10 bits, or 4000 tones per color. JPG compression reduces those tones down to only 256 tones per color. This drastic reduction in tones means a drastic limitation to your editing and optimizing chores.
Summary: every image your camera captures is forced into a flatline curve appearance, leaving you with very little opportunity to change its appearance without visible damage. Let’s look at these issues individually.
Conformation Curve. Images captured by a digital camera are as unique as seashells; they’re all different. While some are just slightly different, others differ significantly. High-key images are composed largely of lighter tones containing very few dark tones (cat). “Normal” images have a pretty even dispersion of tones throughout the image (clouds). And low-key images contain mostly darker tones (lamp). You can see the difference, but JPG compression doesn’t even notice.
Linear curves have no influence on an image. It is what it is. Non-linear curves are used to alter the appearance of an image. With JPG, the same linear curve is applied to all images ecumenically. This won’t hurt, did it?
And then there are dozens of subgroups beneath these major types. This disparity of captured lighting conditions requires that individual assignment of tone curves be built to deliver detail from each image type. If you don’t adjust the middle tones, you will leave detail on the cutting room floor.
Since digital camera sensors capture massive amounts of image data (typically 4000 RGB colors inside each pixel), and each image captured can be interpreted in thousands of variations. That’s the good news. The not-so-good news is that JPEG files are initially displayed using a single interpretation of these massive data files. This irreverent generalization conforms every photo into an “average” tonal shape. Even worse, if not rescued and optimized immediately after the capture, this less-than-delicate interpretation of the image data almost always discards important detail from each photo. Hence the label “lossy (or lousy) compression.”
Of Photos and Waffles To clarify the previous statement, this data-reduction JPEG process throws away any excess data recorded by the camera by reducing the 4000 original tones (per RGB color) down to a skeletal 256 frame. This is like pouring five quarts of pancake batter into a waffle iron that can hold only 8 ounces, closing the lid, trimming the excess batter and throwing it away. What you get is a generic pre-shaped waffle photo from every capture. Yum! Now the somewhat-good news is that the pixels in that waffle photo CAN still be pushed around and shaped to some degree. 256 tones is still actually a lot of data; about two-plus tones for every percentage (1-99%). Not exactly elastic data, but quite manageable if the original camera JPEG is intelligently edited before it is resized and saved.
Overcoming the JPG Curse Herein lies one of the key secrets to success in the world of JPEG photo reproduction. Learn to control light. Edit the data for tone shape (Curves/Levels etc.) directly from a duplicate copy of the waffle photo data. Make at least one sacred duplicate of the original camera file and only perform edits on a duplicate. Unlike silver emulsion film from the old film days, you can duplicate digital data a zillion times without loss. While there is absolutely no comparison between JPEG and RAW image editing flexibility, there is still a whole lot of editable data in that camera JPG file.
Duplicate vs Copy Notice I said “duplicate,” not save and re-save as a copy. Duplicate the file before opening it in Photoshop or Lightroom. Original camera JPG images are every bit as detailed and sharp as 8-bit TIFF files. This (JPG) format was developed by a “joint group of photographic experts” tasked with making digital images load and transfer rapidly over the Internet. The intent of JPG compression is to reduce as much bit depth (the number of levels of color in each pixel) as possible. This serves the additional purpose of occupying as little disk real estate as possible. A noble goal. And one that does not lose detail IF the user understands the imaging process.
This imaging process is what I will be presenting in some detail in the following posts in this series. I will present the stages of (visual) image optimization required to produce dazzlingly detailed and surprisingly smooth JPG images. I inserted the work “visual” to differentiate my image optimization process from the JPG process of progressive display.
If you prepare images for print, presentation or are involved in the digital image food chain, I invite you to follow along. That’s the way I sees this. Let me hear from you. See you next time.