This file lists out the changes done to the design of the program from Assignment 4 to Assignment 5.
All the features from Assignment 4 still work (i.e. load, save, red-greyscale, blue-greyscale, green-greyscale, value-greyscale, intensity-greyscale,luma-greyscale, brighten, darken, flip-horizontally, flip-vertically).
All the features required in this assingment are also completed; the program can save and load ppm, jpg, png, and bmp, the program can blur, sharpen, greyscale, and sepia the same way as the script commands from Assignment 4, the program can run with and without a script file, an example script file was written with all the operations, and a working JAR file of our program was made (see the example under "Running the Program" in the USEME.md for how to run the script file with the JAR).
With the introduction of the new filtering and color transformation operations, we realized the Pixel interface provide the flexibilty needed to implement these operations i.e. the two void methods, greyscaleWith and adjustBrightness, that we had were not flexible enough to be able to be used here. So we ended up writing setters for the RGB components.
With having to implement loading and saving for JPG, PNG, and BMP, we felt that we needed to save the alpha component (which our design had not previously done as PPM doesn't support the alpha channel). To avoid changing what the Pixel class represented, we extended the interface with a new interface that represented a pixel that supported the alpha component (TransparentPixel). We then could implement a pixel that supported RGBA by extending the the original implmentation of the Pixel interface (the RGBPixel class, previously known as SimplePixel but was changed to clarify what it represented) and implement the new TransparentPixel interface. This implementation is the RGBAPixel class. Since RGBAPixel is both a Pixel and a TransparentPixel, it's able to use the operations that are specific to Pixel (the operations that extend the PixelOperation interface) and able to support the old operations. However, if there were new operations in the future that required change on the alpha channel specifically, a new interface would have to be written (likely named TransparentPixelOperation) to be able to perform those operations. Additionally, the new interface was written to have a setter for the alpha component as we would likely encounter the same problem as not having setters for the RGB componenets of Pixel.
In order for images to be saved as PNGs, we realized that all images saved within the program would have to have the alpha channel and we had to change the design of MyImage (interface) and SimpleImage (class) to represent a collection of TransparentPixel objects rather than Pixel objects. Images loaded in without an alpha component would just be opaque (their alpha component set to 255 for 8-bit). When reading and writing images, the images would not be able to read or write the alpha component without changing the fact that a MyImage object should represnt a collection of TransparentPixel objects. Creating a new interface and class to represent images that support transparency would mean the operations that extend ImageOperation would not work for this new interface and class. So now all of the images loaded into the model are saved as SimpleImage objects which holds a 2D ArrayList of TransparentPixel objects and the images that are loaded in from file formats that don't support the alpha channel have all their pixel's alpha components set to 255 (a constructor for RGBAPixel was written for this case).
An extra sentence was added after the "Command successful!" message in the controller after successfully completing commands to better the user experience by confirming what command was just executed (it's especially helpful when running the program with a script file).
The reading and writing of files were moved to the controller so it would better follow the MVC pattern. Two methods (getMyImage and writeImage) were added to the controller interface and implemented in the controller class for this to work. Additionally, saveAs and loadFile were both deleted from the model interface and controller and a new method was added (loadImage) and an old method that was already there (getImageNamed) are used alongside the new methods in the controller to make this work. The Save and Load classes (both implmenet ImageProcessingCommand interface) were changed so that they would both have a fields of an ImageProcessingController so they could execute the commands properly; for load it uses the getMyImage method from the controller first and gives it to the model to use in loadImage and for save it uses the getImageNamed method from the model first and then uses the writeImage method from the controller to complete the writing of the file.