Author Topic: PixInsight Standard 01.00.01.219 Beta 3 Released  (Read 19660 times)

Offline Pleiades

  • Administrator
  • PixInsight Enthusiast
  • *****
  • Posts: 88
PixInsight Standard 01.00.01.219 Beta 3 Released
« on: 2006 September 13 03:55:02 »
Hello all,

The latest version of the PixInsight Standard core application, 1.0.1.219 Beta 3, is now available for download:

http://pleiades-astrophoto.com/download/STD/index.html

This is a breakthrough release. In fact, build 219 is very close to a release candidate version. It is extremely stable and includes virtually the final graphical user interface (GUI) of the PixInsight Standard platform.

This release is entirely based on our customized version of Qt 4.1.3 and PCL 1.0.1.77 RC4.

Basically, we only have to implement some relatively simple subsystems, like ImageContainers and a few explorer windows (View Explorer, Icon Explorer, Project Explorer), along with the project management system, and a few minor items. With these exceptions, the core PixInsight application is fully functional now.

New Auto-Hide Window System

You're going to see a very important change in the interface. Those nasty windows that were populating the workspace are no longer there. They are now managed through a sophisticated auto-hide windowing system, which you'll note immediately as a column of selectors at the main window's left side.

New DynamicAlignment Process

DynamicAlignment (DA) is a semi-automatic image registration system for deep-sky images. It works quite easily: you select two images: source and target. Target will be registered to match source. Then a set of alignment points are defined, which obviously are stars. DA's interface is quite sophisticated. It includes an adaptive star-searching algorithm, which is independent on brightness/contrast. This means that you can register linear raw CCD images without problems (using ScreenTransferFunctions to see the stars without altering the images). DA also includes a useful prediction system: starting from the second star, it will predict the target position of every source star.

To use DA, this is the step-by-step general procedure:

1. Open the DynamicAlignment interface (for example, from the Process Explorer).

2. Select a source image by clicking on it. Note that the cursor shape is now a checkmark with a "1" number.

3. Select a target image by clicking on it. Note that the cursor shape is now a checkmark with a "2" number.

4. Click close to the first alignment star on the source image. You don't have to click exactly on a star; DA will search for the brightest star in a small region (whose size is user-definable) around the point you clicked. Note that DA calculates the star's centroid automatically.

5. DA has placed a box on the target image. You have to move it to select the same star as on source. Again, DA will search the star into a searching box, so you don't have to bother with pointing at the star accurately.

6. Repeat the operation for more stars: first a source star, then move the target box if necessary to match the source star on the target image. Starting from the second star, DA will predict the positions of all target stars. Once you have defined two or three stars, DA will normally catch target stars correctly.

7. Watch the Ex and Ey numbers. These are the differences in pixels between DA-predicted target positions and found target star centroids. Normally, when three stars have been defined, Ex and Ey are very small, e.g. 0.01 pixels or something like that. When very small errors are consistently obtained over the whole image, adding more alignment stars doesn't make sense. Three or four stars are sufficient to accurately register most digital (CCD, DSLR) images. However, in presence of strong local distortions (i.e. with film images), more than 4 stars are often necessary. There is no specific limit in the number of alignment stars.

DynamicAlignment is able to register images with one, two, and three or more stars. With one and two stars, an affine transformation is applied. One-star registration can only correct for translation (simple shifts in the x and y directions). Two-star registration can correct for translation, rotation and scaling, when scaling ratios are identical on both axes. However, two stars cannot correct for arbitrary distortions.

With three or more stars, DA applies 2-D surface splines, which implement high-order arbitrary distortion models. I definitely recommend to use at least three stars in all cases.

DA has much more aspects to comment, but these are the basic things. Just a last point. Don't think that DA is worse than other solutions just because DA is a semi-manual process. DA will register your images with an accuracy that will probably surprise you 8)

DynamicAlignment is available under the ImageRegistration category, on the Process Explorer window.

Improved PixelMath Process

The PixelMath that we have included in build 219 is the final version of this essential process.

PixelMath is now an extremely powerful and versatile tool. It supports user-defined variables and multiple expressions. It allows you to access any pixel of any open image at arbitrary coordinates (Pixel() function). We have also added a couple of functions that provide polar coordinates (RDist() and PAngle()).

PixelMath's adaptability and suitability to solve complex problems are amazing now. This is a little but instructive example:

http://pleiades-astrophoto.com/tmp/pm-examples.psm

This is a PSM file (a process icon collection). When you load them, you'll see a few PixelMath icons. Each icon has a brief descriptive text about what it does and how. You can see that information by clicking on the small buttons that you'll see at the bottom-right corner of each icon (this is a new feature of the GUI, by the way), or by right-clicking on an icon and selecting "Edit Instance Information...".

Have a look at these icons to figure out what PixelMath can do now. Of course you can apply each icon to any image to see what happens.

New AutoHistogram process, by Carlos Milovic

Carlos Milovic has written a small but very interesting and useful process that you'll like: AutoHistogram. This process, along with a quick histogram clipping feature, applies an automatic gamma transform to each RGB channel to achieve prescribed median values.

Yes, you've read well: you can achieve automatic background neutralization with prescribed mean background values automatically, with a couple of clicks. Just enter the desired mean value for the background in the Median Target Values section, and apply the process. That's all.

Obviously, for AutoHistogram to work properly, the image's histograms must be dominated by the sky background, that is, there must be enough free sky areas on the image.

AutoHistogram is available under the TransferCurves category, on the Process Explorer window.

**************

Comments, suggestions, bug reports, etc, are always welcome.

The Pleiades Development Team