Miłosz Orzeł

.net, js, html, arduino, java... no rants or clickbaits.

Fast pixel operations in .NET (with and without unsafe)

Bitmap class has GetPixel and SetPixel methods that let you acquire and change color of chosen pixels. Those methods are very easy to use but are also extremely slow. My previous post gives detailed explanation on the topic, click here if you are interested.

Fortunately you don’t have to use external libraries (or resign from .NET altogether) to do fast image manipulation. The Framework contains class called ColorMatrix that lets you apply many changes to images in an efficient manner. Properties such as contrast or saturation can be modified this way. But what about manipulation of individual pixels? It can be done too, with the help from Bitmap.LockBits method and BitmapData class…

Good way to test individual pixel manipulation speed is color difference detection. The task is to find portions of an image that have color similar to some chosen color. How to check if colors are similar? Think about color as a point in three dimensional space, where axes are: red, green and blue. Two colors are two points. The difference between colors is described by the distance between two points in RGB space.

Colors as points in 3D space diff = sqrt((C1R-C2R)2+(C1G-C2G)2+(C1B-C2B)2)

This technique is very easy to implement and gives decent results. Color comparison is actually a pretty complex matter though. Different color spaces are better suited for the task than RGB and human color perception should be taken into account (e. g. our eyes are more keen to detect difference in shades of green that in shades of blue). But let’s keep things simple here…

Our test image will be this Ultra HD 8K (7680x4320, 33.1Mpx) picture (on this blog it’s of course scaled down to save bandwidth):

Color difference detection input image (scaled down for blog)

This is a method that may be used to look for R=255 G=161 B=71 pixels (car number "36"). It sets matching pixels as white (the rest will be black):

static void DetectColorWithGetSetPixel(Bitmap image, byte searchedR, byte searchedG, int searchedB, int tolerance)
{
    int toleranceSquared = tolerance * tolerance;
    
    for (int x = 0; x < image.Width; x++)
    {
        for (int y = 0; y < image.Height; y++)
        {
            Color pixel = image.GetPixel(x, y);

            int diffR = pixel.R - searchedR;
            int diffG = pixel.G - searchedG;
            int diffB = pixel.B - searchedB;

            int distance = diffR * diffR + diffG * diffG + diffB * diffB;

            image.SetPixel(x, y, distance > toleranceSquared ? Color.Black : Color.White);
        }
    }
}

Above code is our terribly slow Get/SetPixel baseline. If we call it this way (named parameters for clarity):

DetectColorWithGetSetPixel(image, searchedR: 255, searchedG: 161, searchedB: 71, tolerance: 60);

we will receive following outcome:

Color difference detection output image (scaled down)

Result may be ok but having to wait over 84300ms* is a complete disaster! 

Now check out this method:

static unsafe void DetectColorWithUnsafe(Bitmap image, byte searchedR, byte searchedG, int searchedB, int tolerance)
{
    BitmapData imageData = image.LockBits(new Rectangle(0, 0, image.Width, image.Height), ImageLockMode.ReadWrite, PixelFormat.Format24bppRgb);
    int bytesPerPixel = 3;

    byte* scan0 = (byte*)imageData.Scan0.ToPointer();
    int stride = imageData.Stride;

    byte unmatchingValue = 0;
    byte matchingValue = 255;
    int toleranceSquared = tolerance * tolerance;

    for (int y = 0; y < imageData.Height; y++)
    {
        byte* row = scan0 + (y * stride);

        for (int x = 0; x < imageData.Width; x++)
        {
            // Watch out for actual order (BGR)!
            int bIndex = x * bytesPerPixel;
            int gIndex = bIndex + 1;
            int rIndex = bIndex + 2;

            byte pixelR = row[rIndex];
            byte pixelG = row[gIndex];
            byte pixelB = row[bIndex];

            int diffR = pixelR - searchedR;
            int diffG = pixelG - searchedG;
            int diffB = pixelB - searchedB;

            int distance = diffR * diffR + diffG * diffG + diffB * diffB;

            row[rIndex] = row[bIndex] = row[gIndex] = distance > toleranceSquared ? unmatchingValue : matchingValue;
        }
    }

    image.UnlockBits(imageData);
}

It does exactly the same thing but runs for only 230ms over 360 times faster!

Above code makes use of Bitmap.LockBits method that is a wrapper for native GdipBitmapLockBits (GDI+, gdiplus.dll) function. LockBits creates a temporary buffer that contains pixel information in desired format (in our case RGB, 8 bits per color component). Any changes to this buffer are copied back to the bitmap upon UnlockBits call (therefore you should always use LockBits and UnlockBits as a pair). Bitmap.LockBits returns BitmapData object (System.Drawing.Imaging namespace) that has two interesting properties: Scan0 and Stride. Scan0 returns an address of the first pixel data. Stride is the width of single row of pixels (scan line) in bytes (with optional padding to make it dividable by 4). 

BitmapData layout

Please notice that I don’t use calls to Math.Pow and Math.Sqrt to calculate distance between colors. Writing code like this: 

double distance = Math.Sqrt(Math.Pow(pixelR - searchedR, 2) + Math.Pow(pixelG - searchedG, 2) + Math.Pow(pixelB - searchedB, 2));

to process millions of pixels is a terrible idea. Such line can make our optimized method about 25 times slower! Using Math.Pow with integer parameters is extremely wasteful and we don’t have to calculate square root to determine if distance is longer than specified tolerance.

Previously presented method uses code marked with unsafe keyword. It allows C# program to take advantage of pointer arithmetic. Unfortunately, unsafe mode has some important restrictions. Code must be compiled with \unsafe option and executed for fully trusted assembly. 

Luckily there is a Marshal.Copy method (from System.Runtime.InteropServices namespace) that can move data between managed and unmanaged memory. We can use it to copy image data into a byte array and manipulate pixels very efficiently. Look at this method:

static void DetectColorWithMarshal(Bitmap image, byte searchedR, byte searchedG, int searchedB, int tolerance)
{        
    BitmapData imageData = image.LockBits(new Rectangle(0, 0, image.Width, image.Height), ImageLockMode.ReadWrite, PixelFormat.Format24bppRgb);

    byte[] imageBytes = new byte[Math.Abs(imageData.Stride) * image.Height];
    IntPtr scan0 = imageData.Scan0;

    Marshal.Copy(scan0, imageBytes, 0, imageBytes.Length);
  
    byte unmatchingValue = 0;
    byte matchingValue = 255;
    int toleranceSquared = tolerance * tolerance;

    for (int i = 0; i < imageBytes.Length; i += 3)
    {
        byte pixelB = imageBytes[i];
        byte pixelR = imageBytes[i + 2];
        byte pixelG = imageBytes[i + 1];

        int diffR = pixelR - searchedR;
        int diffG = pixelG - searchedG;
        int diffB = pixelB - searchedB;

        int distance = diffR * diffR + diffG * diffG + diffB * diffB;

        imageBytes[i] = imageBytes[i + 1] = imageBytes[i + 2] = distance > toleranceSquared ? unmatchingValue : matchingValue;
    }

    Marshal.Copy(imageBytes, 0, scan0, imageBytes.Length);

    image.UnlockBits(imageData);
}

It runs for 280ms, so it is only slightly slower than unsafe version. It is CPU efficient but uses more memory then previous method – almost 100 megabytes for our test Ultra HD 8K image in RGB 24 format.

If you want to make pixel manipulation even faster you may process different parts of the image in parallel. You need to make some benchmarking first because for small images the cost of threading may be bigger than gains from concurrent execution. Here’s a quick sample of code that uses 4 threads to process 4 parts of the image simultaneously. It yields 30% time improvement on my machine. Treat is as a quick and dirty hint, this post is already to long…

static unsafe void DetectColorWithUnsafeParallel(Bitmap image, byte searchedR, byte searchedG, int searchedB, int tolerance)
{
    BitmapData imageData = image.LockBits(new Rectangle(0, 0, image.Width, image.Height), ImageLockMode.ReadWrite, PixelFormat.Format24bppRgb);
    int bytesPerPixel = 3;

    byte* scan0 = (byte*)imageData.Scan0.ToPointer();
    int stride = imageData.Stride;

    byte unmatchingValue = 0;
    byte matchingValue = 255;
    int toleranceSquared = tolerance * tolerance;

    Task[] tasks = new Task[4];
    for (int i = 0; i < tasks.Length; i++)
    {
        int ii = i;
        tasks[i] = Task.Factory.StartNew(() =>
            {
                int minY = ii < 2 ? 0 : imageData.Height / 2;
                int maxY = ii < 2 ? imageData.Height / 2 : imageData.Height;

                int minX = ii % 2 == 0 ? 0 : imageData.Width / 2;
                int maxX = ii % 2 == 0 ? imageData.Width / 2 : imageData.Width;                        
                
                for (int y = minY; y < maxY; y++)
                {
                    byte* row = scan0 + (y * stride);

                    for (int x = minX; x < maxX; x++)
                    {
                        int bIndex = x * bytesPerPixel;
                        int gIndex = bIndex + 1;
                        int rIndex = bIndex + 2;

                        byte pixelR = row[rIndex];
                        byte pixelG = row[gIndex];
                        byte pixelB = row[bIndex];

                        int diffR = pixelR - searchedR;
                        int diffG = pixelG - searchedG;
                        int diffB = pixelB - searchedB;

                        int distance = diffR * diffR + diffG * diffG + diffB * diffB;

                        row[rIndex] = row[bIndex] = row[gIndex] = distance > toleranceSquared ? unmatchingValue : matchingValue;
                    }
                }
            });
    }

    Task.WaitAll(tasks);

    image.UnlockBits(imageData);
}

Update 2018-01-08: If you really want to do some complex and efficient image processing then you should use specialized library like OpenCV. Few months ago I've written "Detecting a Drone - OpenCV in .NET for Beginners (Emgu CV 3.2, Visual Studio 2017)" blog post series that will help you do it...

* .NET 4 console app, executed  on MSI GE620 DX laptop: Intel Core i5-2430M 2.40GHz (2 cores, 4 threads), 4GB DDR3 RAM, NVIDIA GT 555M 2GB DDR3, HDD 500GB 7200RPM, Windows 7 Home Premium x64.

Radio buttons for list items in MVC 4 – problem with id uniqueness

Let's suppose that we have some model that has a list property and we want to render some radio buttons for items of that list. Take the following basic setup as an example.

Main model class with list:

using System.Collections.Generic;

public class Team
{
    public string Name { get; set; }
    public List<Player> Players { get; set; }
}

List item class:

public class Player
{
    public string Name { get; set; }
    public string Level { get; set; }
}

There are three accepted values for player’s skill Level property: BEG (Beginner), INT (Intermediate) and ADV (Advanced) so we want three radio buttons (with labels) for each player in a team. Yup, normally we would rather use enum instead of a string for Level property, but let’s skip it here for the sake of simplicity…

Controller action method that returns sample data:

public ActionResult Index()
{
    var team = new Team() {
        Name = "Some Team",
        Players = new List<Player> {
               new Player() {Name = "Player A", Level="BEG"},
               new Player() {Name = "Player B", Level="INT"},
               new Player() {Name = "Player C", Level="ADV"}
        }
    };

    return View(team);
}

Here is our Index.cshtml view:

@model Team

<section>
    <h1>@Model.Name</h1>        

    @Html.EditorFor(model => model.Players)            
</section>

Notice that markup for Players is not created inside a loop. Instead, EditorTemplate is used. It’s a good practice since it makes code more clear and maintainable. Framework is smart enough to use code from template for each player on a list, because Team.Players property implements IEnumerable interface...

And here is Players.cshtml EditorTemplate:

@model Player
<div>
    <strong>@Model.Name:</strong>

    @Html.RadioButtonFor(model => model.Level, "BEG")
    @Html.LabelFor(model => model.Level, "Beginner")

    @Html.RadioButtonFor(model => model.Level, "INT")
    @Html.LabelFor(model => model.Level, "Intermediate")

    @Html.RadioButtonFor(model => model.Level, "ADV")
    @Html.LabelFor(model => model.Level, "Advanced")        
</div>

The code looks fine, nice strongly typed helpers that relay on lambda expressions (for compile-time checking and easier refactoring)... but there’s a catch: HTML markup that is generated by such code is actually seriously flawed. Check this snipped of web page source generated for the first player:

<div>
    <strong>Player A:</strong>  
 
    <input name="Players[0].Level" id="Players_0__Level" type="radio" checked="checked" value="BEG">
    <label for="Players_0__Level">Beginner</label>
 
    <input name="Players[0].Level" id="Players_0__Level" type="radio" value="INT">
    <label for="Players_0__Level">Intermediate</label>
 
    <input name="Players[0].Level" id="Players_0__Level" type="radio" value="ADV">
    <label for="Players_0__Level">Advanced</label>        
</div>

Players_0__Level id is used for three different radio buttons! Lack of uniqueness not only violates HTML specification and makes scripting hard but also causes label tags to not work properly (clicking on them doesn’t check their corresponding input element). 

Fortunately MVC framework contains TemplateInfo class that has GetFullHtmlFieldId method. This method returns id for DOM element. That id is constructed by appending name provided as method argument to an automatically determined prefix. This prefix takes into account nesting level and list item's index. Internally, GetFullHtmlFieldId uses TemplateInfo.HtmlFieldPrefix property and TagBuilder.CreateSanitizedId method so even if you pass some illegal characters to id suffix they will be replaced.

Here is modified EditorTemplate
@model Player
<div>
    <strong>@Model.Name:</strong>

    @{string rbBeginnerId = ViewContext.ViewData.TemplateInfo.GetFullHtmlFieldId("rbBeginner"); }
    @Html.RadioButtonFor(model => model.Level, "BEG", new { id = rbBeginnerId })
    @Html.LabelFor(model => model.Level, "Beginner",  new { @for = rbBeginnerId} )

    @{string rbIntermediateId = ViewContext.ViewData.TemplateInfo.GetFullHtmlFieldId("rbIntermediate"); }
    @Html.RadioButtonFor(model => model.Level, "INT", new { id = rbIntermediateId })
    @Html.LabelFor(model => model.Level, "Intermediate",  new { @for = rbIntermediateId })

    @{string rbAdvancedId = ViewContext.ViewData.TemplateInfo.GetFullHtmlFieldId("rbAdvanced"); }
    @Html.RadioButtonFor(model => model.Level, "ADV", new { id = rbAdvancedId })
    @Html.LabelFor(model => model.Level, "Advanced",  new { @for = rbAdvancedId })
</div>

Calls to ViewContext.ViewData.TemplateInfo.GetFullHtmlFieldId method let us obtain ids for radio buttons which are also used to set for attributes of labels. In MVC 3 there was no overload of LabelFor extension method that accepted htmlAttributes object. Luckily version 4 has it build-in.

Above code produces such markup:

<div>
    <strong>Player A:</strong>
 
    <input name="Players[0].Level" id="Players_0__rbBeginner" type="radio" checked="checked" value="BEG">
    <label for="Players_0__rbBeginner">Beginner</label>
 
    <input name="Players[0].Level" id="Players_0__rbIntermediate" type="radio" value="INT">
    <label for="Players_0__rbIntermediate">Intermediate</label>
 
    <input name="Players[0].Level" id="Players_0__rbAdvanced" type="radio" value="ADV">
    <label for="Players_0__rbAdvanced">Advanced</label>
</div>

Now inputs ids are unique and labels properly reference radio buttons via for attribute. Alright :) 

BTW: the weird name “radio buttons” for mutually exclusive option elements comes from buttons on radio receivers that were used to switch between stations (pushing one in automatically pushed the others out).

Coordinate system in HTML5 Canvas, drawing with y-axis value increasing upwards

Coordinate system in HTML5 Canvas is set up in such a way that its origin (0,0) is in the upper-left corner. This solution is nothing new in the world of screen graphics (e.g. the same goes for Windows Forms and SVG). CRT monitors, which were standard in the past, displayed picture lines from top to bottom and image within a line was created from left to right. So locating origin (0,0) in the upper-left corner was intuitive and it made creating hardware and software for handling graphics easier.

Unfortunately sometimes default coordinate system in canvas is a bit impractical. Let’s assume that you want to create projectile motion animation. It seems natural that for ascending projectile, the value of y coordinate should increase. But it will result in a weird effect of inverted trajectory:

Default coordinate system (y value increases downwards)

You can get rid of this problem by modifying y value that is passed to drawing function:

context.fillRect(x, offsetY - y, size, size);

For y = 0, projectile will be placed in a location determined by offsetY (to make y = 0 be the very bottom of the canvas, set offsetY equal to height of the canvas). The bigger the value of y the higher a projectile will be drawn. The problem is that you can have hundreds of places in your code that use y coordinate. If you forget to use offsetY just once the whole image may get destroyed. 

Luckily canvas lets you make changes to coordinate system by means of transformations. Two transformation methods will be useful for us: translate(x ,y) and scale(x, y). The former allows us to move origin to an arbitrary place, the latter is for changing size of drawn objects, but it may also be used to invert coordinates.

Single execution of the following code will move origin of coordinate system to point (0, offsetY) and establish y-axis values as increasing towards the top of the screen:

context.translate(0, offsetY);
context.scale(1, -1);

Translation and scaling of coordinate system. Click to enlarge...

But there’s a catch: the result of providing -1 as scale’s method second argument is that the whole image is created for inverted y coordinate. This applies to text too (calling fillText will render letters upside-down). Therefore before writing any text, you have to restore default y-axis configuration. Because manual restoring of canvas state is awkward, methods save() and restore() exist. These methods are for pushing canvas state on the stack and popping canvas state from the stack, respectively. It is recommended to use save method before doing transformations. Canvas state includes not only transformations but also values such as fill style or line width...

context.save();
 
context.fillStyle = 'red';
context.scale(2, 2);
context.fillRect(0, 0, 10, 10);
 
context.restore();
 
context.fillRect(0, 0, 10, 10);

Above code draws 2 squares: 

First square is red and is drawn with 2x scale. Second square is drawn with default canvas settings (color black and 1x scale). This occurs because right before any changes to scale and color, canvas state was save on the stack, later on it was restored before second square drawing.

TortoiseSVN pre-commit hook in C# - save yourself some troubles!

Probably everyone who creates or debugs a program happens to make temporary changes to the code that make current task easier but should never get into the repository. And probably everyone has accidentally put such code into next revision. If you are lucky enough, mistake will be revealed quickly and the only result will be a bit of shame, if not...

If only there was a way to mark “uncommitable” code...

You can do it and it’s pretty simple!

TortoiseSVN lets you set so-called pre-commit hook. It’s a program (or script) that is run when user clicks “OK” button in “SVN Commit” window. Hook can for example check content of modified files and block commit when deemed appropriate. Tortoise hooks differ from Subversion hooks in that they are executed locally and not on the server that hosts the repository. You therefore don’t have to worry whether your hook will be accepted by the admin or if it works on the server (e.g. server may not have .NET installed), you also don’t affect the experience of other users of the repository. Client-side hooks are quicker too.

Detailed description of hooks can be found in „4.30.8. Client Side Hook Scripts” chapter of Tortoises help file.

Tortoise supports 7 kinds of hooks: start-commit, pre-commit, post-commit, start-update, pre-update, post-update and pre-connect. We are concerned with pre-commit action. The essence of the hook is to check whether one of added or modified files contains temporary code marker. Our marker may be a “NOT_FOR_REPO” text put into a comment placed above temporary code.

This is whole hook’s code – simple console application, that may save your ass :)

using System;
using System.IO;
using System.Text.RegularExpressions;

namespace NotForRepoPreCommitHook
{
    class Program
    {
        const string NotForRepoMarker = "NOT_FOR_REPO";

        static void Main(string[] args)
        {              
            string[] affectedPaths = File.ReadAllLines(args[0]);

            Regex fileExtensionPattern = new Regex(@"^.*\.(cs|js|xml|config)$", RegexOptions.IgnoreCase);

            foreach (string path in affectedPaths)
            {
                if (fileExtensionPattern.IsMatch(path) && File.Exists(path))
                {
                    if (ContainsNotForRepoMarker(path))
                    {
                        string errorMessage = string.Format("{0} marker found in {1}", NotForRepoMarker, path);
                        Console.Error.WriteLine(errorMessage);    
                        Environment.Exit(1);  
                    }
                }
            }             
        }

        static bool ContainsNotForRepoMarker(string path)
        {
            StreamReader reader = File.OpenText(path);

            try
            {
                string line = reader.ReadLine();

                while (line != null)
                {
                    if (line.Contains(NotForRepoMarker))
                    {
                        return true;
                    }

                    line = reader.ReadLine();
                }
            }
            finally
            {
                reader.Close();
            }  

            return false;
        }
    }
}

TSVN calls pre-commit hook with four parameters. We are interested only in the first one. It contains a path to *.tmp file. In this file there is a list of files affected by current commit. Each line is one path. After loading the list, files are filtered by extension (useful if you don’t want to process files of all types). Checking if file exists is also important – the list from *.tmp file contains paths for deleted files too! Detection of the marker represented by NotForRepoMarker constant is realized by ContainsNotForRepoMarker method. Despite its simplicity it provides good performance. On mine (middle range) laptop, 100 MB file takes less than a second to process. If marker is found, program exits with error code (value different than 0). Before quitting, information about which file contains the marker is sent to standard error output (via Console.Error). This message will get displayed in Tortoise window.

The code is simple, isn’t it? In addition, hook installation is also trivial!

To attach hook, choose “Settings” item from Tortoise’s context menu. Then select “Hook scripts” element and click “Add…” button. Such window will appear:

TSVN hooks configuration window

Set „Hook Type” to „Pre-Commit Hook”. Fill “Working Copy Path” field with a path to the directory that contains local copy of the repo (different folders can have different hooks). In “Command Line To Execute” field, set path to the application that implements the hook. Check “Wait for the script to finish” and “Hide the script while running” options (the latter will prevent console window from showing). Press “OK” button and voila, hook is installed!

Now mark some code with “NOT_FOR_REPO” comment and try to execute commit. You should see something like that:

Operation blocked by pre-commit hook

Notice the „Retry without hooks” button – it allows commit to be completed by ignoring hooks.

We now have a hook that prevents from temporary code submission. One may also want to create a hook that enforces log message to be filled, blocks *.log files commits etc. Your private hooks – you decide! And if some of the hooks will be usefull for the whole team, you can always remake them as Subversion hooks.

Tested on TortoiseSVN 1.7.8/Subversion 1.7.6.

Update 24.03.2014: Added emphasis to checking "Wait for the script to finish" option - without it hook will not block commits!

Update 17.09.2013: (additional info): You may set hook on a parent folder which contains multiple repositories checkouts. If you are willing to sacrifice a bit of performance for added protection you may resign from filtering files before checking for NotForRepoMarker marker.  

How to close pop-ups upon main window closure or logout?

Imagine you have to provide support for some really old web application. The app has one main window and pop-up windows that show some sensitive information (for example payroll list). Client wants to ensure that all pop-ups are closed when user leaves main window or clicks “logout” button in this window...

So... how to close all the windows opened with window.open?

On the web this question comes up very often. Unfortunately, most common answer is really naive. Proposed solution is based on keeping references to opened pop-ups and subsequent invocation of close method: 

var popups = []; 
 
function openPopup() {
    var wnd = window.open('Home/Popup', 'popup' + popups.length, 'height=300,width=300');
     
    popups.push(wnd);
}
 
function closePopups() {
    for (var i = 0; i < popups.length; i++) {
        popups[i].close();
    }
 
    popups = [];
}

In practice this doesn’t work because the array of references is cleared at full page reload (for example after clicking on a link or upon postback)...

Other suggested solution is to give the pop-up a unique name (using the second parameter of the open method) and later acquisition of a reference to the window:

var wnd = window.open('', 'popup0');
wnd.close();

This is based on the fact, that window.open method works in two modes:

  1. If a window with a given name doesn’t exist, it is created.
  2. If a window with a given name does exist, it will not be recreated, instead a reference to that window will be returned (if non empty URL is passed to the open method pop-up will be reloaded).

The problem lies at point no. 1. If pop-up window with given name wasn’t previously opened, the call to open and close methods will cause the pop-up to be briefly visible. It sucks…

But maybe a reference to pop-up can be retained between page reloads?

If there is no need to support older browsers (unlikely for the old application) we can try to put reference to the pop-up window into localStorage. However, this will not work:

var popup = window.open('http://morzel.net', 'test');
localStorage.setItem('key', JSON.stringify(popup)); 
 
TypeError: Converting circular structure to JSON

Old tricks for keeping page state between reloads that are based on cookies or window.name will not work too.

 

So… what to do?

Even if you can’t afford to have a major change such as introducing frames, don’t give up :)

Pop-up windows have opener property that points to parent window (that is the window in which the call to window.open was placed). Pop-ups can therefore periodically check whether the main window still remains open. Additionally, pop-ups can also access variables from parent window. This can be used to enforce pop-ups closure when main window is closed or when user clicks on “logout” button in parent window. When user is logged-in (and only then!), a marker variable (i.e. loggedIn) should be set in main window.

Here is the JS code that should be placed on a page displayed in a pup-up:

window.setInterval(function () {
    try {
        if (!window.opener || window.opener.closed === true || window.opener.loggedIn !== true) {
            window.close();
        }
    } catch (ex) {
        window.close(); // FF may throw security exception when you try to access loggedIn (for external site)
    }
}, 1000);

Checking variable from the opener window has another advantage. If user moves away from our application in main window (for example by clicking back button or a link to an external website), then the pop-up window will detect the lack of monitored variable in window.opener and close automatically.

Well, it's not the kind of code you enjoy to write but it achieves the desired result despite the painful gaps in the browsers API. If only they provide us with window.exists('name') method...