Microsoft’s System.Drawing.Common package is commonly used for cross-platform graphics in .NET Framework and .NET Core applications, but according to the dotnet roadmap System.Drawing will soon only support Windows. As Microsoft sunsets cross-platform support for System.Drawing they will be simultaneously developing Microsoft.Maui.Graphics, a cross-platform graphics library for iOS, Android, Windows, macOS, Tizen and Linux completely in C#.
The Maui.Graphics library can be used in any .NET application (not just MAUI applications). This page documents how I used the Maui.Drawing package to render graphics in memory (using a Skia back-end) and save them as static images from a console application.
I predict Maui.Graphics will eventually evolve to overtake System.Drawing in utilization. It has many advantages for performance and memory management (discussed extensively elsewhere on the internet), but it is still early in development. As of today (July 2021) the Maui.Graphics GitHub page warns “This is an experimental library … There is no official support. Use at your own Risk.”
This program will create an image, fill it with blue, add 1,000 random lines, then draw some text. It is written as a .NET 5 top-level console application and requires the Microsoft.Maui.Graphics and Microsoft.Maui.Graphics.Skia NuGet packages (both are currently in preview).
We use SkiaSharp to create a canvas, but importantly that canvas implements Microsoft.Maui.Graphics.ICanvas (it’s not Skia-specific) so all the methods that draw on it can be agnostic to which rendering system was used. This makes it easy to write generic rendering methods now and have the option to switch the rendering system later.
This page demonstrates how to continuously monitor microphone input using C#. Code here may be a helpful reference for developers interested in working with mono or stereo data captured from an audio device in real time. This project uses NAudio to provide simple access to the microphone on Windows platforms.
Mono
Stereo
Full source code is available on GitHub (Program.cs)
This program starts by creating a WaveInEvent with a WaveFormat that specifies the sample rate, bit depth, and number of channels (1 for mono, 2 for stereo).
We can create a function to handle incoming data and add it to the DataAvailable event handler:
var waveIn = new NAudio.Wave.WaveInEvent
{
DeviceNumber = 0, // customize this to select your microphone device WaveFormat = new NAudio.Wave.WaveFormat(rate: 44100, bits: 16, channels: 1),
BufferMilliseconds = 50};
waveIn.DataAvailable += ShowPeakMono;
waveIn.StartRecording();
This method is called when the incoming audio buffer is filled. One of the arguments gives you access to the raw bytes in the buffer, and it’s up to you to convert them to the appropriate data format.
This example is suitable for 16-bit (two bytes per sample) mono input.
privatestaticvoid ShowPeakMono(object sender, NAudio.Wave.WaveInEventArgs args)
{
float maxValue = 32767;
int peakValue = 0;
int bytesPerSample = 2;
for (int index = 0; index < args.BytesRecorded; index += bytesPerSample)
{
intvalue = BitConverter.ToInt16(args.Buffer, index);
peakValue = Math.Max(peakValue, value);
}
Console.WriteLine("L=" + GetBars(peakValue / maxValue));
}
This method converts a level (fraction) into bars suitable to display in the console:
privatestaticstring GetBars(double fraction, int barCount = 35)
{
int barsOn = (int)(barCount * fraction);
int barsOff = barCount - barsOn;
returnnewstring('#', barsOn) + newstring('-', barsOff);
}
When the WaveFormat is configured for 2 channels, bytes in the incoming audio buffer will have left and right channel values interleaved (2 bytes for left, two bytes for right, then repeat). Left and right channels must be treated separately to display independent levels for stereo audio inputs.
This example is suitable for 16-bit (two bytes per sample) stereo input.
privatestaticvoid ShowPeakStereo(object sender, NAudio.Wave.WaveInEventArgs args)
{
float maxValue = 32767;
int peakL = 0;
int peakR = 0;
int bytesPerSample = 4;
for (int index = 0; index < args.BytesRecorded; index += bytesPerSample)
{
int valueL = BitConverter.ToInt16(args.Buffer, index);
peakL = Math.Max(peakL, valueL);
int valueR = BitConverter.ToInt16(args.Buffer, index + 2);
peakR = Math.Max(peakR, valueR);
}
Console.Write("L=" + GetBars(peakL / maxValue));
Console.Write(" ");
Console.Write("R=" + GetBars(peakR / maxValue));
Console.Write("\n");
}
Scientific image analysis frequently involves working with 12-bit and 14-bit sensor data stored in 16-bit TIF files. Images commonly encountered on the internat are 24-bit or 32-bit RGB images (where each pixel is represented by 8 bits each for red, green, blue, and possibly alpha). Typical image analysis libraries and documentation often lack information about how to work with 16-bit image data.
This page summarizes how I work with 16-bit TIF file data in C#. I prefer Magick.NET (an ImageMagick wrapper) when working with many different file formats, and LibTiff.Net whenever I know my source files will all be identically-formatted TIFs or multidimensional TIFs (stacks).
ImageMagick is a free and open-source cross-platform software suite for displaying, creating, converting, modifying, and editing raster images. Although ImageMagick is commonly used at the command line, .NET wrappers exist to make it easy to use ImageMagick from within C# applications.
ImageMagick has many packages on NuGet and they are described on ImageMagick’s documentation GitHub page. TLDR: Install the Q16 package (not HDRI) to allow you to work with 16-bit data without losing precision.
ImageMagick is free and distributed under the Apache 2.0 license, so it can easily be used in commercial projects.
An advantage of loading images with ImageMagick is that it will work easily whether the source file is a JPG, PNG, GIF, TIF, or something different. ImageMagick supports over 100 file formats!
// Load pixel values from a 16-bit TIF using ImageMagick (Q16)MagickImage image = new MagickImage("16bit.tif");
ushort[] pixelValues = image.GetPixels().GetValues();
That’s it! The pixelValues array will contain one value per pixel from the original image. The length of this array will equal the image’s height times its width.
Since the Q16 package was installed, 2 bytes will be allocated for each pixel (16-bit) even if it only requires one byte (8-bit). In this case you must collect just the bytes you are interested in:
For this you’ll have to install the high dynamic range (HDRI) Q16 package, then your GetValues() method will return a float[] instead of a ushort[]. Convert these values to proper pixel intensity values by dividing by 2^16.
MagickImage image = new MagickImage("32bit.tif");
float[] pixels = image.GetPixels().GetValues();
for (int i = 0; i < pixels.Length; i++)
pixels[i] = (long)pixels[i] / 65535;
LibTiff is a pure C# (.NET Standard) TIF file reader. Although it doesn’t support all the image formats that ImageMagick does, it’s really good at working with TIFs. It has a more intuitive interface for working with TIF-specific features such as multi-dimensional images (color, Z position, time, etc.).
LibTiff gives you a lower-level access to the bytes that underlie image data, so it’s on you to perform the conversion from a byte array to the intended data type. Note that some TIFs are little-endian encoded and others are big-endian encoded, and endianness can be read from the header.
LibTiff is distributed under a BSD 3-clause license, so it too can be easily used in commercial projects.
// Load pixel values from a 16-bit TIF using LibTiffusingTiff image = Tiff.Open("16bit.tif", "r");
// get information from the headerint width = image.GetField(TiffTag.IMAGEWIDTH)[0].ToInt();
int height = image.GetField(TiffTag.IMAGELENGTH)[0].ToInt();
int bytesPerPixel = image.GetField(TiffTag.BITSPERSAMPLE)[0].ToInt() / 8;
// read the image data bytesint numberOfStrips = image.NumberOfStrips();
byte[] bytes = newbyte[numberOfStrips * image.StripSize()];
for (int i = 0; i < numberOfStrips; ++i)
image.ReadRawStrip(i, bytes, i * image.StripSize(), image.StripSize());
// convert the data bytes to a double arrayif (bytesPerPixel != 2)
thrownew NotImplementedException("this is only for 16-bit TIFs");
double[] data = newdouble[bytes.Length / bytesPerPixel];
for (int i = 0; i < data.Length; i++)
{
if (image.IsBigEndian())
data[i] = bytes[i * 2 + 1] + (bytes[i * 2] << 8);
else data[i] = bytes[i * 2] + (bytes[i * 2 + 1] << 8);
}
Routines for detecting and converting data from 8-bit, 24-bit, and 32-bit TIF files can be created by inspecting bytesPerPixel. LibTiff has documentation describing how to work with RGB TIF files and multi-frame TIFs.
I often prefer to work with scientific image data as a 2D arrays of double values. I write my analysis routines to pass double[,] between methods so the file I/O can be encapsulated in a static class.
// Load pixel values from a 16-bit TIF using ImageMagick (Q16)MagickImage image = new MagickImage("16bit.tif");
ushort[] pixelValues = image.GetPixels().GetValues();
// create a 2D array of pixel valuesdouble[,] imageData = newdouble[image.Height, image.Width];
for (int i = 0; i < image.Height; i++)
for (int j = 0; j < image.Width; j++)
imageData[i, j] = pixelValues[i * image.Width + j];
According to ImageProcessor’s GitHub page, “ImageProcessor is, and will only ever be supported on the .NET Framework running on a Windows OS” … it doesn’t appear to be actively maintained and is effectively labeled as deprecated, so I won’t spend much time looking further into it.
Although this library can save images at different depths, it can only load image files with 8-bit depths. System.Drawing does not support loading 16-bit TIFs, so another library must be used to work with these file types.
This page is a quick reference for programmers interested in working with image data in memory (byte arrays). This topic is straightforward overall, but there are a few traps that aren’t necessarily intuitive so I try my best to highlight those here.
This article assumes you have some programming experience working with byte arrays in a C-type language and have an understanding of what is meant by 32-bit, 24-bit, 16-bit, and 8-bit integers.
An image is composed of a 2D grid of square pixels, and the type of image greatly influences how much memory each pixel occupies and what format its data is in.
Bits per pixel (bpp) is the number of bits it takes to represent the value a single pixel. This is typically a multiple of 8 bits (1 byte).
Gray 8 - Specifies one of 2^8 (256) shades of gray.
Indexed 8 - The pixel data contains color-indexed values, which means the values are an index to colors in the system color table, as opposed to individual color values.
16-bit (2-byte) Pixel Formats
ARGB 1555 - Specifies one of 2^15 (32,768) shades of color (5 bits each for red, green, and blue) and 1 bit for alpha.
Gray 16 - Specifies one of 2^16 (65,536) shades of gray.
RGB 555 - 5 bits each are used for the red, green, and blue components. The remaining bit is not used.
RGB 565 - 5 bits are used for the red component, 6 bits are used for the green component, and 5 bits are used for the blue component. The additional green bit doubles the number of gradations and improves image perception in most humans.
24-bit (3 byte) Pixel Formats
RGB 888 - 8 bits each are used for the red, green, and blue components.
32-bit (4-byte) Pixel Formats
ARGB - 8 bits each are used for the alpha, red, green, and blue components. This is the most common pixel format.
There are others (e.g., 64-bit RGB images), but these are the most typically encountered pixel formats.
Endianness describes the order of bytes in a multi-byte value uses to store its data:
big-endian: the smallest address contains the most significant byte
little-endian: the smallest address contains the least significant byte
Assuming array index values ascend from left to right, 32-bit (4-byte) pixel data can be represented using either of these two formats in memory:
4 bpp little-endian: [A, B, G, R] (most common)
4 bpp big-endian: [R, G, B, A]
Bitmap images use little-endian integer format! New programmers may expect the bytes that contain “RGB” values to be sequenced in the order “R, G, B”, but this is not the case.
Premultiplication refers to the relationship between color (R, G, B) and transparency (alpha). In transparent images the alpha channel may be straight (unassociated) or premultiplied (associated).
With straight alpha, the RGB components represent the full-intensity color of the object or pixel, disregarding its opacity. Later R, G, and B will each be multiplied by the alpha to adjust intensity and transparency.
With premultiplied alpha, the RGB components represent the emission (color and intensity) of each pixel, and the alpha only represents transparency (occlusion of what is behind it). This reduces the computational performance for image processing if transparency isn’t actually used.
In C# using System.Drawing premultiplied alpha is not enabled by default. This must be defined when creating new Bitmap as seen here:
var bmp = new Bitmap(400, 300, PixelFormat.Format32bppPArgb);
Benchmarking reveals the performance enhancement of drawing on bitmaps in memory using premultiplied alpha pixel format. In this test I’m using .NET 5 with the System.Drawing.Common NuGet package. Anti-aliasing is off in this example, but similar results were obtained with it enabled.
Random rand = new(0);
int width = 600;
int height = 400;
var bmp = new Bitmap(600, 400, PixelFormat.Format32bppPArgb);
var gfx = Graphics.FromImage(bmp);
var pen = new Pen(Color.Black);
gfx.Clear(Color.Magenta);
var sw = Stopwatch.StartNew();
for (int i = 0; i < 1e6; i++)
{
pen.Color = Color.FromArgb(rand.Next());
gfx.DrawLine(pen, rand.Next(width), rand.Next(height), rand.Next(width), rand.Next(height));
}
Console.WriteLine(sw.Elapsed);
bmp.Save("benchmark.png", ImageFormat.Png);
A 2D image is composed of pixels, but addressing them in memory isn’t as trivial as it may seem. The dimensions of bitmaps are stored in their header, and the arrangement of pixels forms rows (left-to-right) then columns (top-to-bottom).
Width and height are the dimensions (in pixels) of the visible image, but…
⚠️ Image size in memory is not just width * height * bytesPerPixel
Because of old hardware limitations, bitmap widths in memory (also called the stride) must be multiplies of 4 bytes. This is effortless when using ARGB formats because each pixel is already 4 bytes, but when working with RGB images it’s possible to have images with an odd number of bytes in each row, requiring data to be padded such that the stride length is a multiple of 4.
// calculate stride length of a bitmap row in memoryint stride = 4 * ((imageWidth * bytesPerPixel + 3) / 4);
This example demonstrates how to convert a 3D array (X, Y, C) into a flat byte array ready for copying into a bitmap. Notice this code adds padding to the image width to ensure the stride is a multiple of 4 bytes. Notice also the integer encoding is little endian.
publicstaticbyte[] GetBitmapBytes(byte[,,] input)
{
int height = input.GetLength(0);
int width = input.GetLength(1);
int bpp = input.GetLength(2);
int stride = 4 * ((width * bpp + 3) / 4);
byte[] pixelsOutput = newbyte[height * stride];
byte[] output = newbyte[height * stride];
for (int y = 0; y < height; y++)
for (int x = 0; x < width; x++)
for (int z = 0; z < bpp; z++)
output[y * stride + x * bpp + (bpp - z - 1)] = input[y, x, z];
return output;
}
For completeness, here’s the complimentary code that converts a flat byte array from bitmap memory to a 3D array (assuming we know the image dimensions and bytes per pixel from reading the image header):
publicstaticbyte[,,] GetBitmapBytes3D(byte[] input, int width, int height, int bpp)
{
int stride = 4 * ((width * bpp + 3) / 4);
byte[,,] output = newbyte[height, width, bpp];
for (int y = 0; y < height; y++)
for (int x = 0; x < width; x++)
for (int z = 0; z < bpp; z++)
output[y, x, z] = input[stride * y + x * bpp + (bpp - z - 1)];
return output;
}
The code examples above are intentionally simple to focus on the location of pixels in memory and the endianness of their values. To actually convert between byte[] and System.Drawing.Bitmap you must use Marshall.Copy as shown:
The code examples above use System.Drawing.Common to create graphics, but creating bitmaps in a byte[] array is not difficult and can be done in any language. See the Creating Bitmaps from Scratch article for more information.
Python can be used to securely deploy website content using FTPS. Many people have used a FTP client like FileZilla to drag-and-drop content from their local computer to a web server, but this method requires manual clicking and is error-prone. If you write a script to accomplish this task it lowers the effort barrier for deployment (encouraging smaller iterations) and reduces the risk you’ll accidentally do something unintentional (like deleting an important folder by accident).
This post reviews how I use Python, keyring, and TLS to securely manage login credentials and deploy builds from my local computer to a remote server using FTP. The strategy discussed here will be most useful in servers that use the LAMP stack, and it’s worth noting that .NET and Node have their own deployment paradigms. I hope you find the code on this page useful, but you should carefully review your deployment script and create something specific to your needs. Just as you could accidentally delete an important folder using a graphical client, an incorrectly written deployment script could cause damage to your website or leak secrets.
Store your password using an interactive interpreter to ensure you don’t accidentally save it in a plain text file somewhere. This only needs to be done once.
FTP was not designed to be a secure - it transfers login credentials in plain text! Traditionally FTP in Python was achieved using ftplib.FTP from the standard library, but logging-in using this protocol allows anyone sniffing traffic on your network to capture your password. In Python 2.7 ftplib.FTP_TLS was added which adds transport layer security to FTP (FTPS), improving protection of your login credentials.
# ⚠️ This is insecure (password transferred in plain text)fromftplibimport FTP
with FTP(hostname, username, password) as ftp:
print(ftp.nlst())
# 👍 This is better (password is encrypted)fromftplibimport FTP_TLS
with FTP_TLS(hostname, username, password) as ftps:
print(ftps.nlst())
By default ftplib.FTP_TLS only encrypts the username and password. You can call prot_p() to encrypt all transferred data, but in this post I’m only interested in encrypting my login credentials.
FTP over SSL (FTPS) is different than FTP over SSH (SFTP), but both use encryption to transfer usernames and passwords, making them superior to traditional FTP which transfers these in plain text.
This method deletes each of the contents of a folder, then deletes the folder itself. If one of the contents is a subfolder, it calls itself. This example uses modern Python practices, favoring pathlib over os.path.
Note that I define the remote path using pathlib.PurePosixPath() to ensure it’s formatted as a unix-type path since my remote server is a Linux machine.
importftplibimportpathlibdefremoveRecursively(ftp: ftplib.FTP, remotePath: pathlib.PurePath):
"""
Remove a folder and all its contents from a FTP server
"""defremoveFile(remoteFile):
print(f"DELETING FILE {remoteFile}")
ftp.delete(str(remoteFile))
defremoveFolder(remoteFolder):
print(f"DELETING FOLDER {remoteFolder}/")
ftp.rmd(str(remoteFolder))
for (name, properties) in ftp.mlsd(remotePath):
fullpath = remotePath.joinpath(name)
if name =='.'or name =='..':
continueelif properties['type'] =='file':
removeFile(fullpath)
elif properties['type'] =='dir':
removeRecursively(ftp, fullpath)
removeFolder(remotePath)
if __name__ =="__main__":
remotePath = pathlib.PurePosixPath('/the/remote/folder')
with ftplib.FTP_TLS("swharden.com", "scott", "P455W0RD") as ftps:
removeFolder(ftps, remotePath)
This method recursively uploads a local folder tree to the FTP server. It first creates the folder tree, then uploads all files individually. This example uses modern Python practices, favoring pathlib over os.walk() and os.path.
Like before I define the remote path using pathlib.PurePosixPath() since the server is running Linux, but I can use pathlib.Path() for the local path and it will auto-detect how to format it based on which system I’m currently running on.
importftplibimportpathlibdefuploadRecursively(ftp: ftplib.FTP, remoteBase: pathlib.PurePath, localBase: pathlib.PurePath):
"""
Upload a local folder to a remote path on a FTP server
"""defremoteFromLocal(localPath: pathlib.PurePath):
pathParts = localPath.parts[len(localBase.parts):]
return remoteBase.joinpath(*pathParts)
defuploadFile(localFile: pathlib.PurePath):
remoteFilePath = remoteFromLocal(localFile)
print(f"UPLOADING FILE {remoteFilePath}")
withopen(localFile, 'rb') as localBinary:
ftp.storbinary(f"STOR {remoteFilePath}", localBinary)
defcreateFolder(localFolder: pathlib.PurePath):
remoteFolderPath = remoteFromLocal(localFolder)
print(f"CREATING FOLDER {remoteFolderPath}/")
ftp.mkd(str(remoteFolderPath))
createFolder(localBase)
for localFolder in [x for x in localBase.glob("**/*") if x.is_dir()]:
createFolder(localFolder)
for localFile in [x for x in localBase.glob("**/*") ifnot x.is_dir()]:
uploadFile(localFile)
if __name__ =="__main__":
localPath = pathlib.Path(R'C:\my\project\folder')
remotePath = pathlib.PurePosixPath('/the/remote/folder')
with ftplib.FTP_TLS("swharden.com", "scott", "P455W0RD") as ftps:
uploadRecursively(ftps, remotePath, localPath)
Because walking remote folder trees deleting and upload files can be slow, this process may be disruptive to a website with live traffic. For low-traffic websites this isn’t an issue, but as traffic increases (or the size of the deployment increases) it may be worth considering how to achieve the swap faster.
An improved method of deployment could involve uploading the new website to a temporary folder, switching the names of the folders, then deleting the old folder. There is brief downtime between the two FTP rename calls.
Speed could be improved by handling the renaming with a shell script that runs on the server. This would require some coordination to execute though, but is worth considering. It could be executed by a HTTP endpoint.
You can automate deployment of a React project using Python and FTPS. After creating a new React app add a deploy.py in the project folder that uses FTPS to upload the build folder to the server, then edit your project’s package.json to add predeploy and deploy commands.
This post focused on how to automate uploading local content to a remote server using FTP, but don’t overlook the possibility that this may not be the best method for deployment for your application.
You can maintain a website as a git repository and use git pull on the server to update it. GitHub Actions can be used to trigger the pull step automatically using an HTTP endpoint. If this method is available to you, it should be strongly considered.
This method is very popular, but it (1) requires git to be on the server and (2) requires all the build tools/languages to be available on the server if a build step is required. I’m reminded that only SiteGround’s most expensive shared hosting plan even has git available at all.