
Of all the new features in Photoshop CS3, those that stand out for me most are the ones that have been built around the new Align Content feature, engineered by Jeff Chien.
For example, you can use Align Content in Perspective mode to align group portraits and it is now possible to create really accurate Photo merge composites automatically.
When you put Photoshop CS3 Extended alongside the regular version of Photoshop CS3, we can let you know about what some of the extended features will allow you to do. Of these, the Image Stacks rendering is one of the most interesting new features in Photoshop CS3 Extended, better than Live Filters.
The new Stacks feature was engineered by Chris Cox and was originally designed as a tool for analytical work, where you could place a series of images together in alignment and apply a Stacks rendering to the layers and use this to process them in such a way that Photoshop will highlight the differences found between the layers, or as is shown below, blend the layers according to where there is a high frequency of recurring pixel values to display only the most commonly occurring pixel values. There are several key creative uses for Image Stacks.
For this first demo, has a series of six pictures where there was always at least one person walking through a scene. Using image stacks, you are able to automatically remove them from the shot.

Figure 1- Here is a sequence of photographs that were shot hand held over a time period of a minute or so. There were a lot of people walking in front of the fountain and I just made sure that I captured enough shots so that each portion of the picture had two or more frames where someone wasn’t in front of the camera.

Figure 2- I opened all of these photographs in Photoshop and went to the File Scripts menu to choose: Load Files into Stack. This opened the dialog shown here, where I chose Use: Open Files and checked the Attempt to Automaticallly Align Source Images and Create Smart Object after Loading Layers options.


Figure 3- Depending on how many pictures you have and how large they are, it may take a few minutes to process all the photographs. What you will end up with will be a new document with a Smart Object layer that contains all the previously open image documents as layers grouped within the smart object. If you double-click on the Smart Object layer you will see the full expanded list of layers contained in this Smart Object.


Figure 4- And now for the clever part. If you have the Smart Object open, make sure you close it again. You will want to start with the Smart Object selected (see the Layers palette top left). Go to the Layer menu and choose Smart Objects Image Stack Mode Median. Again, the stacks rendering may take a little while to complete. In the result shown here, the Median rendering managed to blend the layers such that nearly all of the people in the merged picture disappeared completely.

Figure 5- The image stack median rendering did a pretty good job of removing the people, but there were still a few ghost outlines left. Obviously some people were having too good a time in the sun to want to move around much. Plus there were a few bits of rubbish and artifacts around the edges of the picture where the frames had overlapped. By tidying up the final picture by adding a little bit of spotting on a new layer and added some masked curves adjustment layers to provide some dodging and burning to produce the final version shown here.
Tips for getting the best results
When you see this technique demonstrated it does at first look
quite magical, but there is a logical explanation for how the
process works. The technique relies mainly on the use of the
auto-align command to align a chosen set of sample images
together and place them within a grouped smart object. After
doing a little bit of experimentation I have found that if you
record at least five or six (or more) exposures, this should
provide enough separate images for Photoshop to process in order
to work out which pixels appear most frequently at any
particular spot in the picture and use the most commonly
occurring pixels only to produce the finished blend shown here.
Removing noise using multiple exposures
Here is a technique that makes use of the Stacks feature in
Photoshop CS3 to merge a set of identical exposures and obtain a
smoother-looking image.

Figure 6

Figure 7

Figure 8- I started by going to the File menu and choosing Scripts ➯ Load Files into Stack and selected a set of five images to open that had all been shot at identical exposures of a subject with the camera fixed to the tripod. These were photographs that had been shot at a high ISO setting using a long exposure in low light conditions. I checked the Align Source images and Create Smart Object options and clicked OK.

Figure 9- The selected images opened as a single image document grouped together as a single smart object. If I were to double-click on the smart object icon, this would open the smart object in a separate document window and allow me access to all the individual layers, which wasn’t necessary in this case, but would be if you wanted to edit any of the individual layers.

Figure 10

Figure 11- Back in the original Smart Object document, I went to the Layer menu and chose Image Stacks Image Stack Rendering Median. The processing may take a little while, depending on the size and number of layers, plus bit depth. Once completed, you will notice how the Smart Object layer has a ‘stacks’ icon indicating the smart object has been rendered using the stacks feature.

Figure 12

Figure 13- Here is a comparison showing a close-up view of a single exposure (Figure 12) and a rendered version (Figure 13) where five separate exposures were merged to produce a smoother, noise-free image. The Median rendering was used here because it analyzes the image content on all the layers and averages out the pixel values to use the most commonly occurring pixel values only, thereby eliminating nearly all of the noisy pixels that occur on each of the layers.
Median versus Mean
If you are processing a series of still life captures, then a
Mean stacks rendering can remove more noise than Median. For
example, if you were processing astronomy pictures, you would
want to use a Mean rendering.
The techniques shown here are fairly easy to accomplish. The Align Content feature is so good at recognizing areas of similarity and aligning images together as layers, that you can quite easily get away without having to use a tripod to shoot the pictures that you want to combine together. So anytime you are in a situation where you think it might be useful to remove people from a shot or you want to improve upon the image quality capture potential of a lowlight scene, just shoot a quick sequence of shots with the camera hand held, keeping it as still as possible.