Processes, Objects and Scripts

In the simplest terms, processes are equivalent to tools and objects are typically image files. A script is a small program, often provided by a third party that can provide a utility function, combine existing tools into a more usable form and simplify batch processing. A script editor is also provided in PI to create your own, or modify existing ones. With a little knowledge you can read a script and work out what it does (that is if you have a few cloudy nights). As each tool is applied to an image, the “Process Console” tracks progress with a text-like record. This may appear redundant but it can be useful to scroll back to ensure everything worked on every file and that the various actions were performed according to plan.

Icons are minimized files or tools, floating on the desktop. In a program like Photoshop, the tools are anchored to the tool bars as small icons and a more detailed dialog is expanded when required. In PI, a process is selected from a menu and if the process (tool) is minimized to an icon on the desktop, it keeps its settings and can be renamed to something useful. Its icon form occupies very little space on the desktop and it can be recalled or stored for repeated use or arranged with others to remind you of a processing sequence. An image can be minimized to an icon too. It is handy to keep a partially processed image file close at hand for use later on, say for generating masks or alignment. These unique features allow one to build upon early successes.

Containers, Workspaces and Projects

For the more confident user, “Containers” is a posh word for a collection of image or tools. This is a software device to group together image files for batch processing and assemble a sequence of processes (tools) to apply to one or more images.

Workspaces are just a method to de-clutter the desktop and organize those images and tools that belong together and recall them as a group with a touch of a button. Files and tools are just dragged to the workspace buttons as required.

Projects are a lifesaver. They store all your work, tool settings, work in progress (and each image’s processing history) for a particular session. When you next turn on the computer, run PI and load the project, these are recalled and allow one to carry on with transparent continuity. The workspaces are preserved, and if you organize your work, implicitly record tool settings, workflow and image states. This is not only extremely powerful, a project also serves as a reference for subsequent images. Projects are unbelievably useful, as image processing is often an ongoing activity spanning several sessions.

Previews

Previews may seem obvious enough but PI has a unique slant. One can click on a make preview button and a small outline will appear on your image file. Pulling the edges re-sizes and moves it about. It has its own tab in the image too. You can apply a tool to the preview and get a good idea of what it is going to do. One first trick is to create multiple previews and quickly compare settings. The second is to combine several previews as a single entity, say, tiled around a central galaxy or nebula. This selection facilitates background or color calibration without being distracted by the higher luminance and color of the subject. Third, the preview has its own processing history. These settings can be copied for later use.

None of the above seem too strange or illogical in their own right. It is just, for a new user, little appears to be familiar or obvious. After a week or two, you will have it figured out and be on the learning curve of how to use the tools themselves. This is when the fun really starts.

PixInsight Workflows

A workflow is a defined processing sequence. You may have come across the term to describe the camera-to-print color management process in digital imaging. In astrophotography it is theoretically possible to adjust color, noise, sharpness and tonal distribution in any order. In practice, there is a preferred sequence that yields the best quality, or put another way, there are certain processing sequences that create issues that are difficult to fix later on. Here is the problem_ The initial images start off as a few bright points of light for the brightest stars and everything else appears featureless, or simply black on the screen. All the interesting stuff occupies a tiny tonal range in the shadows. There is nothing else to see in the image, or more importantly, judge processing options with. This detail magically appears with an applied curve, HistogramTransformation or some kind of image stretch. With PI or any other program, do not permanently stretch an image to see the faint details before correcting the basic faults in the image. It is OK to stretch an image preview or view with a temporary screen stretch, so long as the image data is unaltered. A stretched image magnifies all the good and bad points at the same time. Stretching functions change the tone distribution of an image, typified by a curved transfer function between the input and output pixel values, that boost the shadow contrast and compress the highlights. (Altering the end points, with a linear characteristic in-between still has a linear transfer function.) Images that have not been stretched are called linear and after stretching, non-linear. Certain processes are optimized to work on linear images and once an image is non-linear, there is no return. The reason for this is simple; the mathematical algorithms behind these processes are designed for the typical parameters of a linear image in terms of signal level, contrast and noise. For instance, most noise reduction algorithms are optimized for linear or non-linear images as they work on statistical variations that are affected by the image stretching process.

It is possible to classify many PI processes into linear and non-linear camps and in simple terms, the linear processes are applied first, before image stretching. The equivalent processes in the other specialist astrophotography programs generally exhibit similar behavior. Normal photo editing programs are less specific, as they are purposed for general imaging in a non-linear gamma space to improve noise, sharpness and the like. A typical PI workflow is shown in fig.5.

Pixinsight Processes

The following chapters step through a typical workflow and draw upon the alternative methods in Photoshop, Maxim DL, Nebulosity and PI. A table of the more common PI processes with a description of what they do and their practical application is shown in fig.6.

When you start using PixInsight, the individual settings in each tool will be just that, individual. Some tools have many controls, so many in fact that it is daunting. My approach is, if in doubt, leave them at their default setting. There are many Internet resources that offer practical examples and good starting points. Further experimentation will give you a feel for what works for your particular image conditions. The mouse-over help windows provide useful information for many tool settings.

After a while, especially with a common camera system, quite a few of the settings become second nature. PI still requires considerable catch-up on documentation, especially considering it is a commercial program. This quick appreciation gives an overview of the kind of tools that are at your disposal and unravels some of the acronyms. Enterprising users continually find new ways to use them for clever effect and in addition, new improved tools replace established ones from time to time. An active and friendly community continues to push the boundaries and share their insights.

There is no denying PI rewards patience and perseverance and it is not to everyone’s taste. It does offer some unique processing tools that make the most of painstakingly acquired images, including high bit-depth processing, multiscale processing to enhance and suppress detail and optimized mask generation. I only started to use PI after the original book’s market research identified the need for PI content. It was my intention to use PI to process a few images in the first light section, along with Nebulosity and Maxim DL. After trying it on a few images, however, I quickly noticed a significant jump in image quality, as a result of the complex mathematics that lie at its heart, and have to force myself to use anything else.

fig.5 A simplified PixInsight LRGB workflow from separate monochrome exposures through to the final image. There are many variations according to taste, for narrow band imaging, one-shot color imaging and also the nature of the deep sky object itself. The processes are generic and in many cases this workflow applies to other imaging software too. In PI, there is often a choice of tools to accomplish a similar function, each optimized for a certain situation. It is finding out what for, how much and when to, that is the fun part.

fig.6 Common PixInsight tool descriptions (continued on next page).