#Y0 running fred 3 generator#
For the same reason, an image identifier (including metasymbols such as $T) can only be used to define the source image of a generator, but not any of its other working parameters (because no generator can work on two or more images in the same function call, for now). If the above expression were legal, we would have a set of potentially infinite generated images during expression execution, which obviously does not make any sense. Gconv( MyImage, sigma, aspectRatio, rotationAngle )īecause now rotationAngle has been defined (and is being used) as a variable. Sigma = 2.5, aspectRatio = 0.2, rotationAngle However, constants can be used without problems. One of them is that variables cannot be used to define generator parameters. This has important consequences that must be taken into account from the syntactic and semantic points of view. This means that generators don't have the concept of current pixel, as the rest of PixelMath functions and operators do: a generator works on an entire image as a single step just once, not in a pixel-by-pixel fashion as normal functions. Generators are always executed just after all image references, metasymbols and invariant subexpressions have been evaluated, but before running PixelMath expressions. You have all of them well documented in the Expression Editor dialog. I'll describe gconv() and the rest of generators later in this post. This example is simple, since gconv() is capable of much more sophistication, such as applying elliptical filters with prescribed aspect ratios and rotation angles: The newly generated image plays exactly the role of an existing image in the PixelMath expression to which the generator belongs. It performs a convolution of the specified image with a Gaussian filter and generates a new image with the convolution result. Gconv() is one of the new PixelMath generators. The PixelMath expression used in the above example is: See it in action on the following figure: Now, wouldn't it be nice to have this functionality implemented directly in PixelMath, that is, without needing to duplicate images, apply other processes, and repeat the same sequence each time we want to try out different parameters, such as the Gaussian filter size in this particular example? Yes, it is extremely nice, and it is now a reality. This is similar to how a classical unsharp mask filter works. The 'A_minus_B' image is now a high-pass filtered version of 'A', since we have removed low-pass components. The image 'A_minus_B' is the result of subtracting B from A with PixelMath. Here we have an image 'A' and its duplicate 'B', to which we have applied a convolution with a Gaussian filter (which is a low-pass filter, as you know). An example is shown in the following figure: For example, a high-pass filter can be implemented by subtracting a low-pass filtered version of an original image. Or a transformed version of a transformed version, and so on. However, suppose that we want to use a transformed version of an existing image, instead of the image itself. So we can use existing images very easily, by just putting their names in a PixelMath expression. I'm sure all of you have seen expressions like this one many times, since this is standard practice to implement narrowband combinations. Images are specified by their identifiers in PixelMath expressions. For example, the following expression:Ĭombines the 'this' and 'other' images in a ratio of 3/4 to 1/4. As you know, images have always been native objects of the PixelMath language. To understand how generators work, let's see first how images play their role in PixelMath expressions. With this feature, PixelMath evolves to become an extremely powerful image processing and analysis tool.
This is a completely new feature that radically changes the applicability and meaning of PixelMath, which is from now on much more than a pixel-by-pixel image manipulation tool. Generators are special functions that can build new images on the fly when a PixelMath expression is executed.
#Y0 running fred 3 update#
In this post I'm going to show you some important new features implemented in version 1.8.0 of PixelMath, which is now available as a regular update to version 1.8.8-7 of PixInsight on all supported platforms. My intention with this tool has always been to realize the idea of flexibility with your imagination being the only limit, and this new version makes that idea come true as never before. PixelMath is one of those tools that really define PixInsight as a uniquely powerful image processing platform.
0 kommentar(er)
