Analyze and modify pixels of an image

[patrickulrich007]

Hi all

I'm looking for a replacement of the System.Drawing.Bitmap class, as we want to move to Linux.
I'm looking into the PhotoSauce library and so far it looks very promising.
One of my use cases is, to open an image(Jpg, Tiff, Png, ...) and analyze the image, for example look for all read pixels and replace them with a white pixel.
So far, I was able to open the image iterate over the pixels and check the color.
But now, I'm not able to save the modified byte[] back as a Jpg. Can anybody give me some hints?
This is my code so far:

int redPixels = 0;
long totalPixels = 0;

using (var imgStream = new FileStream(_SourceFileName, FileMode.Open, FileAccess.ReadWrite))
{
var settings = new ProcessImageSettings();
var pipeline = MagicImageProcessor.BuildPipeline(imgStream, settings);

IPixelSource bitmapSource = pipeline.PixelSource;
if (bitmapSource != null)
{
    int bytesPerPixel = 3;
    int width = bitmapSource.Width;
    int height = bitmapSource.Height;
    totalPixels = width * height;
    byte[] pixelBuffer = new byte[width * height * bytesPerPixel];
    bitmapSource.CopyPixels(new System.Drawing.Rectangle(0, 0, width, height), width * bytesPerPixel, pixelBuffer);

    for (int y = 0; y < height; y++)
    {
        for (int x = 0; x < width; x++)
        {
            int index = (y * width + x) * bytesPerPixel;
            byte blue = pixelBuffer[index];
            byte green = pixelBuffer[index + 1];
            byte red = pixelBuffer[index + 2];
            if (CheckColor(_TargetColor, red, green, blue))
            {
                redPixels++;
            }
        }
    }
}

}

My idea is, to modify the pixel in the if, where I count the red pixels. But then, how can I save that byte[] as an image?
I feel like this is a stupid question, but I spend already some time and now I try here in the forum.

Thx and greetings

[rickbrew]

PhotoSauce.MagicScaler isn't a general purpose imaging library. It isn't the right tool to use for this.

You should look at ImageSharp https://github.com/SixLabors/ImageSharp , it will likely be appropriate for your needs.

[patrickulrich007]

What a shame, I tried ImageSharp for same use case, with only reading pixels and it is so much slower then PhotoSauce.
But thx for your answer.

[saucecontrol]

It sounds like what you want to do is possible by creating an IPixelTransform implementation. Working in PhotoSauce requires a different mindset, because it uses a 'pull' model, kind of like LINQ for pixels.

Instead of thinking of your image as a bitmap, you want to think of it as a pipeline. The consumer of the pipeline is the image encoder if you want e.g. JPEG output. Your IPixelTransform would respond to CopyPixels requests from the encoder by requesting the corresponding pixels from your upstream source (the decoder) and doing whatever manipulation you want.

I don't have time to work up an example right now, but I'll try to put something up in the next couple of days.