BufferedImage and pixel manipulation with transparency

Question

Essentially my question is 2 parts.

1. I want to know the preferred/fast method for being about to manipulate pixels and turn a certain color into a transparent pixel.

2. I want to know if I am able to use this "BufferedImage" without having to save it to a file format that supports transparency like "png."

---

I found a method for setting an individual pixel

here

BufferedImage layer = ImageIO.read(new File(file));
           System.out.println(layer.getWidth());
           BufferedImage image = new BufferedImage(layer.getWidth(), layer.getHeight(), BufferedImage.TYPE_INT_ARGB);
           WritableRaster raster = image.getRaster();


       int width = layer.getWidth();
       int height = layer.getHeight();

       // Slow method: scan all input (layer) image pixels, plotting only those which are not the transparency color
       int lPixel,red,green,blue;
       for(int w=0;w<width;w++)
         for(int h=0;h<height;h++)
           {
           lPixel = layer.getRGB(w,h); 
           if ((lPixel&0x00FFFFFF) != trans) //transparent color
             {
             red   = (int)((lPixel&0x00FF0000)>>>16); // Red level
             green = (int)((lPixel&0x0000FF00)>>>8);  // Green level
             blue  = (int) (lPixel&0x000000FF);       // Blue level
             // Set the pixel on the output image's raster.
             raster.setPixel(w,h,new int[]{red,green,blue,255});  

             }
           }

which as it mentions is a "slow method."

---

I had found this thread Java: Filling a BufferedImage with transparent pixels

Which one comment talks about the "int[]" and manipulating pixels there.

---

1. Essentially I want to take an image, remove the transparency of a certain color, and then be able to use that image.

I notice that I could just set each pixel with "setRGB" in the bufferedImage, but then there is "setpixel" in the "WritableRaster..."

Why exactly would you use the "WritableRaster's" "setpixel," instead of "setRGB" in the "BufferedImage?"

I'm also curious how I would access the "int[]" and if that is the preferred way to go about this?

Is there a way to search for certain colored pixels, instead of going through each and every pixel to find the right color myself?

Just curious what the preferred/fastest method would be.

The method with the int[] claimed it wouldn't use "hardware acceleration" so I'm curious if that is important as well?

---

2. I want to ask about this image in general. I know that to use a transparent image, you need a special file format like "gif," "png," or "tga," but I'm curious if I can just use this new bufferedImage that I created instead of saving it, and then re-reading it back in? I'm not sure if once it's buffered if I can do anything with it, and it will display, or if it has to be a certain format?

I would assume once it's buffered in, I could do anything to it, but I'm curious what other people think?

---

Thoughts?

Thanks a lot for any help!

Answer 1

Well about your second question: you just need to load the File into a BufferedImage with a right format that supports Transparency, for example:BufferedImage.TYPE_INT_ARGB. ARGB means "Alpha, Red , Green , Blue" where alpha stands for transparency. You can manipulate that BufferedImage however you want, File format is just how you intend to write it in the end.

As for your first question, I'm not sure how you would optimize this, as you are going to have to access each pixel(no matter which method you use), and change the value.I would go for the slow method, since (for me) seems unavoidable.

EDIT: The int[] method is that you basically take the values in the WritableRaster and put them in an array. So instead of calling setRGB() you would just call array[i][j]=someValue. This would theoretically increase performance as you would be doing low level things (accessing a value in some array) instead of a higher level method call like setRGB().But you would still have to convert that array back into a Raster and then back into a BufferedImage. To be honest, the profit of doing this seems negligeable as you are not making a high performance game.I just tried the "slow" method on a 4K picture ,the highest res you would usually encounter, and the program finished in 10 seconds (knowing that I made ALL the pixels transparent for a worst case scenario).A usual HD picture would take about 4 times less time.