• Choose a category
• All
• Classic-Fiction

By Root 1957 0

to make an image look better will always result in there being some data loss. This is quite normal and an inevitable consequence of the image editing process. The steps on the page opposite illustrate what happens when you edit a photograph. You will notice that as you adjust the input levels and adjust the gamma (middle) input slider, you will end up stretching some of the levels further apart and gaps may start to appear in the histogram. More importantly, stretching the levels further apart can result in less well-defined tonal separation and therefore less detail in these regions of the image. This is particularly a problem at the shadow end of the scale because there are always fewer levels of usable tone information in the shadows compared with the highlights (see the section on Digital exposure on page 186). While moving the gamma slider causes the tones on one side to stretch, it cause the tones on the other side to compress more and these can appear as spikes in the histogram. This too can cause data loss, sometimes resulting in flatter tone separation (there is also an example of this shown in the section on Camera Raw Brightness adjustments on pages 188–189).

Interpreting an image

A digital image is nothing more than a lot of numbers and it is how those numbers are interpreted in Photoshop that creates the image you see on the display. We can use our eyes to make subjective judgements about how the picture looks, but we can also use the number information to provide useful and usable feedback. The Info panel is your friend. If you understand the numbers, it can help you see the fine detail that your eyes are not sharp enough to discern. Plus we have the Histogram panel, which is also an excellent teaching tool and makes everything that follows much easier to understand.

The histogram can therefore be used to provide visual feedback on the levels information in an image and indicate whether there is clipping at either end of the scale. However, does it really matter whether we obtain a perfectly smooth histogram or not? If you are preparing a photograph to go to a print press, you would be lucky to detect more than 50 levels of tonal separation from any single ink plate. Therefore, the loss of a few levels at the completed edit stage does not necessarily imply that you have too little digital tonal information from which to reproduce a full-tonal range image in print. Having said that if you begin with a bad-looking histogram, the image is only going to be in a worse state after it has been retouched. For this reason it is best start to out with the best quality scan or capture you can get.

Basic Levels editing and the histogram

This image editing example was carried out on an 8-bit RGB image, so it should come as no surprise that the histogram broke down as soon as I applied a simple Levels adjustment (in the following section we are going to look at the advantages of editing in 16-bits per channel mode).

1.

Here is an image that displays an evenly distributed range of tones in the accompanying Histogram panel view.

2.

If I apply a Levels image adjustment and drag the middle (gamma) input slider to the left, this lightens the image and the Histogram panel on the right shows the histogram display after the adjustment has been applied. To understand what has happened here, this histogram represents the newly mapped levels. The levels in the section to the left of the gamma slider have been stretched and the levels to the right of the gamma slider have now been compressed.

Bit depth

The bit depth refers to the maximum number of levels per channel that can be contained in a photograph. For example, a 24-bit RGB color image is made up of three 8-bit image channels, where each 8-bit channel can contain up to 256 levels of tone, while a 16-bit per channel image can contain up to 32,768 data points per color channel, because in truth, Photoshop's 16-bit depth is actually 15-bit +1 (see sidebar on page 307).

Understanding bit depth

To understand what the bit depth numbers mean, it is