When importing images, it is best to organize them in albums. Not only to keep things organized, but also because only images that are filed in albums (not “Unfiled”) can be stacked. Image stacking is a digital image processing technique which combines multiple images into one to obtain a single high quality image.
For example, you may want to combine a set of dark frames into a single master dark. Or you may want to calibrate, align and combine images into a single one.
If you have a set of images in an album that you wish to stack, select them, and then choose Stack ▸ Stack. This will create a single stacked image from all those source images. Usually the images you want to stack will be of the same type, dimensions, have the same exposure duration, are taken at nearly equal CCD temperatures, etc. Observatory therefore also has an alternative method of stacking: by selecting all images in an album, and then choosing Stack ▸ Auto-Stack, observatory will create multiple stacks, taking the image type, dimensions, etc into account.
When you stack images in the browser, they are replaced by a single image. You can recognize these from their stack badge:
What happened is that Observatory created a new image by averaging the corresponding pixels of your original images. If you switch to the Stack Inspector along the right side of Observatory’s window, you’ll see that the stack type is “Average”.
You can choose between the following stack types:
Each source image is added as a separate layer of the target stack image.
If you select the stack image, and expand the Layers panel, you’ll see that it actually consists of multiple layers, not one, and each of these layers is named after one of the original images.
Each corresponding pixel of every source image is added. Invalid or missing pixels are skipped.
The pixels of the source images are averaged. This is the default. Invalid or missing pixels are skipped.
The median of the source pixels is used for the target image. Invalid or missing pixels are skipped.
Similar to Average, but rejecting outlier values.
First the mean and the standard deviation of corresponding input pixels are computed. Then the value that is farthest from the mean is selected. If the difference between this value and the mean is less than or equal to the value of the threshold parameter times the standard deviation from the mean, then the mean is returned. If on the other hand the difference is greater, then this outer value is discarded from the set of input values and the mean of this new input range is returned.
When you select a stack, and choose Image ▸ Focus on Stack (
^⌘[), or click the Focus on Stack button in the Browser Focus Bar, the Browser will display the images that you used to create the stack. If you switch to the Browser List View, you’ll see a property that is unique to images that are part of a stack: Weight.
Initially the weight for all images comprising a stack is 100%. This means they take fully part in the creation of the stack image. Sometimes however you may want to give a higher priority to the sharpest images in the stack, and use less sharper images to reduce noise. All the above listed stack types take the weight into account when computing the stack image.
To adjust the weight, select all images comprising the stack (click the
… button), and choose one of the options under Stack ▸ Auto-Weight. Observatory will then compute a weight for each of the images. The choices are:
Computes the background value of each image. Assigns the highest weight to the images with the lowest background value.
Computes the average FWHM value for the stars in each image. Assigns the highest weight to the images with the smallest FWHM. In other words, the sharpest images receive the highest weight.
Computes the FWHM value of the brightest star in each image. Assigns the highest weight to the images with the smallest FWHM.
Computes the peak of the best fitting Gaussian profile for each star in each image. Assigns the highest weight to the images with the highest average peak.
Star Peak (brightest)
Computes the peak of the best fitting Gaussian profile of the brightest star in each image. Assigns the highest weight to those images with the highest peak.
For RGB images, it is important to select the channel for which you want this to be measured. If you don’t, then the values will be an average of the red, green and blue channels. Select the channel in the Channels panel.
You can also reject an image altogether, by choosing Stack ▸ Reject (
⌃⌘0), increase or decrease a weight by choosing Stack ▸ Promote (
⌃⌘=) or Stack ▸ Demote (
⌃⌘-), and reset it to 100 % by choosing Stack ▸ Accept (
While combining images, Observatory will also use each source image’s layer opacity as a weight.
You can add layer adjustments to the images comprising a stack. For example, you can calibrate each, double them in size, and then align them with each other. The stack image is automatically updated whenever you make changes to the stack. Most of these layer adjustments are also available to images that are not part of a stack, but some (e.g. align) are only available for images in a stack because they require a reference image. This reference image is call a Pick. When creating a stack, initially the Pick is the image with the earliest exposure time, but you can change it by choosing Stack ▸ Pick (
Because the stack image is managed by Observatory (it is automatically generated and updated), you cannot manipulate that image directly. To do that, you’ll need to select it and choose Image ▸ New ▸ Master. This will freeze the current state of your stack image into a new managed master. You would do this also when you create a master calibration frame through stacking, by creating and then moving that managed master to one of the calibration albums.
Note that although you can create a single stack for a set of images, because creating versions is light-weight (choose Image ▸ New ▸ Duplicate Version), for experimentation with the weights it may be useful to create multiple stacks for the same master images.