Doing the Impossible with Smart Objects

Before CS3, there were plenty of things you simply could not do in Photoshop. With CS3 and the introduction of Smart Objects, a lot has become possible.

One (almost) obvious thing is that you can apply lossless filters. Just create a layer to a Smart Object and apply the filter. You can add loads of filters and remove or activate them, and of course edit them with Layer Masks.

Smart Object layer in Adobe Photoshop CS4

But not only that - you can now mix layers of different kinds. You can have one layer as a 16 bit image in LAB colour space, while you have another in 8 bit grey scale, and blend the two layers. You can even add RAW files using File > Place, so you have a RAW file layer, which will open for adjustments in Camera RAW every time you double click on it. This means that you can have an almost endless number of different image formats - all stored in the same file.

And if that is not enough, you can have one Smart Object contain another Smart Object that contains a third one and so on.

To have several instances of the same Smart Object, highlight the layer and press command-J. You can have as many instances of the same Smart Object as you like in a file. You can apply different kinds of layer styles and rotations and transformations of them. If you then update the Smart Object, for example by changing its colour, the modification will apply to all instances of that Smart Object in the file.

To replace the content of a Smart Object: Go to Layer > Smart Objects > Replace Contents....

You cannot edit the individual pixels in a Smart Object (unless you open it). However, you can make a copy of the Smart Object, for example with merging all visible layers to a new layer above the currently selected layer with Shift-option-command-E (or alt+ Layer > Merge Visible).

Master a few tools well instead of many tools poorly

One of the marvellous things with Photoshop is that you can achieve the same thing in so many ways. That is also one of the sad things. You may rush around, trying to learn as many of the different tools as possible, and you end up being able to do the same thing over and over again, without acquiring the skills to do anything really well. Two things spring to mind in particular.

Blend modes. Some people claim that a great way of lightening a picture is to add the picture to itself as a new layer and applying, for example, screen blend mode. Likewise, it is alleged that multiplying a picture with itself is an excellent way of darkening it. Well, those methods definitely work, but they are not very flexible. You can equally well achieve the same thing using curves or levels on an adjustment layer. Using an adjustment layer takes much less disk space, and it is much more flexible, as you can choose in which way you want to darken or lighten the picture. There are admittedly quite a lot of flexibility with a duplicated layer as well, if you use Blending options, but no matter which way you use, you need to do some work to get it exactly as you like it.

Another thing that may be overrated is LAB colour space. It can be used for several things, one of them being to get more colour into a picture as described here. However, you can also get more colour into a picture using the adjustment layer Vibrance or Hue/Saturation. You may not get exactly the effect you get using LAB, but the effects you can get with LAB are not necessarily better. And LAB is very destructive. You have to change colour space for the full image, and once you have gone back to RGB, there is no way to go back to LAB to undo or modify the changes you made.

LAB has other uses of course, and so do Blend modes. The point is simply that it sometimes is more efficient to look at how best to use the tools you have and to master them well, instead of trying to find a solution in other tools.

Using new tools is often fun. Using old tools is often efficient.

LAB colour

LAB is a colour space, like RGB or CMYK. RGB is used to describe colours on a computer screen. CMYK is often used to describe colours for printed material. LAB's main advantage is that it is very easy to adjust colours using it.

The secret trick is to keep all lightness information in L.

If you have an RGB picture and you use curves to adjust the values for R (red), you inevitably make parts of the picture lighter or darker, and you will probably have to adjust G (green) and B (blue) as well to get the lightness of the original back. In LAB you have no such problem.

The L value (channel) goes from 0 to 100 and indicates Lightness:

This is more or less a black and white version of the image.

The a channel is the axis green-magenta. "A" is admittedly not a good abbreviation for green magenta, but that's the way things are.

Negative values, from -128, indicate green while positive values, up to 128, indicate magenta. In the image above, the dark areas are negative, so they will turn out greenish. The light areas are positive, so they will be more red. Note that the shadow on the flower is almost entirely invisible. A shadow is lightness, not colour, so it should not be visible in the a-channel, and it is not.


The b channel is the axis blue-yellow.


Negative values, from -128, indicate blue and positive values, up to 128, are yellow.


If we add the three channels we get this picture:


A typical modification in the lab colour space is to emphasize the colours using curves, like this for the b cannel:


The values that seem to be cut off at the edges here are unlikely to actually be used in the picture. The extreme values of Lab describe colours that the human eye cannot see or that do not exist. The modification to the curve is basically just to make it steeper.

Below is a stone that at first may seem mostly grey, but looking a little closer one can see that parts of it have a shift towards red. With LAB colour adjustment, one can emphasise the red, so a casual onlooker will see it much quicker. At the same time the moss got a little more intense green, and the image becomes more interesting.

Before LAB colour adjustment.


After LAB colour adjustment.

Clearly, this is not the only way to increase the colours of a picture. Often, it is quicker to simply increase the saturation. But LAB adds a different kind of flexibility, which may be just what you need.