rotation

How to capture a photo in your Windows Phone 8.1 Runtime app – Part III: capturing and saving the photo

This is the third and last post of this series. In the first two posts I showed you how to start the preview of MediaCapture and some modifications we can apply to it. In this post, we are finally capturing and saving the photo – including the modifications we made before.

The easiest way – capture as is:

The easiest way to capture the photo is to use MediaCapture’s CapturePhotoToStorageFileAsync() method. This method shows you how to do it:

            //declare image format
            ImageEncodingProperties format = ImageEncodingProperties.CreateJpeg();

            //generate file in local folder:
            StorageFile capturefile = await ApplicationData.Current.LocalFolder.CreateFileAsync("photo_" + DateTime.Now.Ticks.ToString(), CreationCollisionOption.ReplaceExisting);

            ////take & save photo
            await captureManager.CapturePhotoToStorageFileAsync(format, capturefile);

            //show captured photo
            BitmapImage img = new BitmapImage(new Uri(capturefile.Path));
            takenImage.Source = img;
            takenImage.Visibility = Visibility.Visible;

This way however does not respect any modifications we made to the preview. The only thing that gets respected is the camera device we are using.

Respecting rotation in the captured photo:

In our ongoing sample, we are using a 90 degree rotation to display the preview element in portrait mode. Naturally, we want to port over this orientation in our captured image.

There are two ways to achieve this. We could capture the photo to a WriteableBitmap and manipulate it, or we could manipulate the image stream directly with the BitmapDecoder and  BitmapEncoder classes. We will do the latter one.

First, we need to open an InMemoryRandomAccessStream for our the captured photo. We are capturing the photo to the stream with MediaCapture’s CapturePhotoToStreamAsync() method, specifing the stream name and the image format.

The next step is to decode the stream with our BitmapDecoder. If we are performing only rotation, we can directly re-encode the InMemoryRandomAccessStream we are using. Rotating the captured photo is very simple with just setting the BitmapTransform.Rotation property to be rotated by 90 degrees, pretty much as easy as rotating the preview.

The last steps are generating a file in the storage, followed by copying the transcoded image stream into the file stream. Here is the complete code that does all this:

            //declare string for filename
            string captureFileName = string.Empty;
            //declare image format
            ImageEncodingProperties format = ImageEncodingProperties.CreateJpeg();

            //rotate and save the image
            using (var imageStream = new InMemoryRandomAccessStream())
            {
                //generate stream from MediaCapture
                await captureManager.CapturePhotoToStreamAsync(format, imageStream);

                //create decoder and encoder
                BitmapDecoder dec = await BitmapDecoder.CreateAsync(imageStream);
                BitmapEncoder enc = await BitmapEncoder.CreateForTranscodingAsync(imageStream, dec);

                //roate the image
                enc.BitmapTransform.Rotation = BitmapRotation.Clockwise90Degrees;

                //write changes to the image stream
                await enc.FlushAsync();

                //save the image
                StorageFolder folder = KnownFolders.SavedPictures;
                StorageFile capturefile = await folder.CreateFileAsync("photo_" + DateTime.Now.Ticks.ToString() + ".jpg", CreationCollisionOption.ReplaceExisting);
                captureFileName = capturefile.Name;

                //store stream in file
                using (var fileStream = await capturefile.OpenStreamForWriteAsync())
                {
                    try
                    {
                        //because of using statement stream will be closed automatically after copying finished
                        await RandomAccessStream.CopyAsync(imageStream, fileStream.AsOutputStream());
                    }
                    catch 
                    {

                    }
                }
            }

Of course, we need to stop the preview after we captured the photo. It also makes all sense to load the saved image and display it to the user. This is the code to stop the preview:

        private async void CleanCapture()
        {

            if (captureManager != null)
            {
                if (isPreviewing == true)
                {
                    await captureManager.StopPreviewAsync();
                    isPreviewing = false;
                }
                captureManager.Dispose();

                previewElement.Source = null;
                previewElement.Visibility = Visibility.Collapsed;
                takenImage.Source = null;
                takenImage.Visibility = Visibility.Collapsed;
                captureButton.Content = "capture";
            }

        }

The result of above mentioned code (screenshot of preview left, captured photo right):

16by9Photo

Cropping the captured photo

Not all Windows Phone devices have an aspect ratio of 16:9. In fact, most devices in the market have an aspect ratio of 15:9, due to the fact that they are WVGA or WXGA devices (I talked a bit about this already in my second post). If we are just capturing the photo with the method above, we will have the same black bands in our image as we have in our preview. To get around this and capture a photo that has a true 15:9 resolution (makes sense for photos that get reused in apps, but less for real life photos), additional code is needed.

As with getting the right camera solution, I generated an Enumeration that holds all possible values as well as a helper method to detect which aspect ratio the currently used device has:

        public enum DisplayAspectRatio
        {
            Unknown = -1,

            FifteenByNine = 0,

            SixteenByNine = 1
        }

        private DisplayAspectRatio GetDisplayAspectRatio()
        {
            DisplayAspectRatio result = DisplayAspectRatio.Unknown;

            //WP8.1 uses logical pixel dimensions, we need to convert this to raw pixel dimensions
            double logicalPixelWidth = Windows.UI.Xaml.Window.Current.Bounds.Width;
            double logicalPixelHeight = Windows.UI.Xaml.Window.Current.Bounds.Height;

            double rawPerViewPixels = DisplayInformation.GetForCurrentView().RawPixelsPerViewPixel;
            double rawPixelHeight = logicalPixelHeight * rawPerViewPixels;
            double rawPixelWidth = logicalPixelWidth * rawPerViewPixels;

            //calculate and return screen format
            double relation = Math.Max(rawPixelWidth, rawPixelHeight) / Math.Min(rawPixelWidth, rawPixelHeight);
            if (Math.Abs(relation - (15.0 / 9.0)) < 0.01)
            {
                result = DisplayAspectRatio.FifteenByNine;
            }
            else if (Math.Abs(relation - (16.0 / 9.0)) < 0.01)
            {
                result = DisplayAspectRatio.SixteenByNine;
            }

            return result;
        }

In Windows Phone 8.1, all Elements use logical pixel size. To get the values that most of us are used to, we need to calculate the raw pixels from the logical pixels. After that, we use the same math operations I used already for detecting the ratio of the camera resolution (see post 2). I tried to calculate the values with the logical pixels as well, but this ended up in some strange rounding behavior and not the results I wanted. That’s why I use the raw pixel sizes.

Before we continue with capturing the photo, we are going to add a border that is displayed and shows the area which is captured to the user in XAML:

            

When we are cropping our photo, we need to treaten the BitmapEncoder and the BitmapDecoder separately. To crop an image, we  need to set the Bounds and the new Width and Height of the photo via the BitmapTransform.Bounds property. We also need to read the PixelData via the GetPixelDataAsync() method, apply the changed Bounds to it and pass them to BitmapEncoder via the SetPixelData() method.

At the end, we are flushing the changed stream data directly into the file stream of our StorageFile. Here is how:

            //declare string for filename
            string captureFileName = string.Empty;
            //declare image format
            ImageEncodingProperties format = ImageEncodingProperties.CreateJpeg();

            using (var imageStream = new InMemoryRandomAccessStream())
            {
                //generate stream from MediaCapture
                await captureManager.CapturePhotoToStreamAsync(format, imageStream);

                //create decoder and transform
                BitmapDecoder dec = await BitmapDecoder.CreateAsync(imageStream);
                BitmapTransform transform = new BitmapTransform();

                //roate the image
                transform.Rotation = BitmapRotation.Clockwise90Degrees;
                transform.Bounds = GetFifteenByNineBounds();

                //get the conversion data that we need to save the cropped and rotated image
                BitmapPixelFormat pixelFormat = dec.BitmapPixelFormat;
                BitmapAlphaMode alpha = dec.BitmapAlphaMode;

                //read the PixelData
                PixelDataProvider pixelProvider = await dec.GetPixelDataAsync(
                    pixelFormat,
                    alpha,
                    transform,
                    ExifOrientationMode.RespectExifOrientation,
                    ColorManagementMode.ColorManageToSRgb
                    );
                byte[] pixels = pixelProvider.DetachPixelData();

                //generate the file
                StorageFolder folder = KnownFolders.SavedPictures;
                StorageFile capturefile = await folder.CreateFileAsync("photo_" + DateTime.Now.Ticks.ToString() + ".jpg", CreationCollisionOption.ReplaceExisting);
                captureFileName = capturefile.Name;

                //writing directly into the file stream
                using (IRandomAccessStream convertedImageStream = await capturefile.OpenAsync(FileAccessMode.ReadWrite))
                {
                    //write changes to the BitmapEncoder
                    BitmapEncoder enc = await BitmapEncoder.CreateAsync(BitmapEncoder.JpegEncoderId, convertedImageStream);
                    enc.SetPixelData(
                        pixelFormat,
                        alpha,
                        transform.Bounds.Width,
                        transform.Bounds.Height,
                        dec.DpiX,
                        dec.DpiY,
                        pixels
                        );

                    await enc.FlushAsync();
                }
            }

You may have notice the GetFifteenByNineBounds() method in the above code. As we need to calculate some values for cropping the image, I decided to separate them. They are not only providing values for the image to be cropped, but also size values for our earlier added Border that is used in my sample (download link at the end of the project) to show the size that the photo will have after our cropping (which is an automatic process in our case,). Here is the code:

        private BitmapBounds GetFifteenByNineBounds()
        {
            BitmapBounds bounds = new BitmapBounds();

            //image size is raw pixels, so we need also here raw pixels
            double logicalPixelWidth = Windows.UI.Xaml.Window.Current.Bounds.Width;
            double logicalPixelHeight = Windows.UI.Xaml.Window.Current.Bounds.Height;

            double rawPerViewPixels = DisplayInformation.GetForCurrentView().RawPixelsPerViewPixel;
            double rawPixelHeight = logicalPixelHeight * rawPerViewPixels;
            double rawPixelWidth = logicalPixelWidth * rawPerViewPixels;

            //calculate scale factor of UniformToFill Height (remember, we rotated the preview)
            double scaleFactorVisualHeight = maxResolution().Width / rawPixelHeight;

            //calculate the visual Width
            //(because UniFormToFill scaled the previewElement Width down to match the previewElement Height)
            double visualWidth = maxResolution().Height / scaleFactorVisualHeight;
            
            //calculate cropping area for 15:9
            uint scaledBoundsWidth = maxResolution().Height;
            uint scaledBoundsHeight = (scaledBoundsWidth / 9) * 15;

            //we are starting at the top of the image
            bounds.Y = 0;
            //cropping the image width
            bounds.X = 0;
            bounds.Height = scaledBoundsHeight;
            bounds.Width = scaledBoundsWidth;

            //set finalPhotoAreaBorder values that shows the user the area that is captured
            finalPhotoAreaBorder.Width = (scaledBoundsWidth / scaleFactorVisualHeight) / rawPerViewPixels;
            finalPhotoAreaBorder.Height = (scaledBoundsHeight / scaleFactorVisualHeight) / rawPerViewPixels;
            finalPhotoAreaBorder.Margin = new Thickness(
                                            Math.Floor(((rawPixelWidth - visualWidth) / 2) / rawPerViewPixels), 
                                            0,
                                            Math.Floor(((rawPixelWidth - visualWidth) / 2) / rawPerViewPixels), 
                                            0);
            finalPhotoAreaBorder.Visibility = Visibility.Visible;

            return bounds;
        }

Again, we need to apply raw pixels to achieve the best results here (I just pasted those lines in for this sample). To calculate the correct values for our Border, we need the scale factor between the screen and the preview resolution we used (which is the scaleFactorVisualHeight double).  Before we’re calculating the border values, we are setting the Width to resolution’s Height (we rotated, remember?) and calculate the matching 15:9 Height.

The Border values are based on the Width and Height of the cropped image, but scaled down by scaleFactorVisualHeight’s value and converted in raw pixel. The Margin positions the border accordingly on top of the preview element.

This is the result of above mentioned code (screenshot of preview left, captured photo right):

15by9Photo

That’s all you need to know to get started with basic photo capturing from within your Windows Phone 8.1 Runtime app. Of course, there are also other modifications that you can apply, and I mentioned already most of the classes that lead you to the matching methods and properties (click on the links to get to the documentation)

By the way, most of the code can be adapted in a Windows 8.1 app as well (with some differences, of course).

Sample project

As promised, you can download the sample here. It contains all code snippets I showed you and is able to run as you build and deploy it.

As always, feedback is welcome and I hope this post is helpful for some of  you.

Until the next time, happy coding!

Posted by msicc in Dev Stories, windev, 21 comments

How to capture a photo in your Windows Phone 8.1 Runtime app-Part II: some common modifications

Like promised in my first post about photo capturing, I will provide some common modification scenarios when using the MediaCapture API. This is what this post is about.

Choosing a camera

If you read my first post, you probably remember that the MediaCapture API automatically selected the front camera of my Lumia 1020. Like often, we have to write some additional code to switch between the cameras.

The cameras are listed in the Panels in the Windows.Devices.Enumeration Namespace. This namespace contains all “devices” that are connected to the phone and has different properties to detect the correct panel. We are going to use the DeviceClass to detect all video capture devices (which are normally also the photo capture devices on Windows Phone, but can be different on a PC/Tablet). As we want to switch between Front and Back, we are also detecting the EnclosureLocation. Here is how I implemented it:

        private static async Task<DeviceInformation> GetCameraID(Windows.Devices.Enumeration.Panel camera)
        {
            DeviceInformation deviceID = (await DeviceInformation.FindAllAsync(DeviceClass.VideoCapture))
                .FirstOrDefault(x => x.EnclosureLocation != null && x.EnclosureLocation.Panel == camera);

            return deviceID;
        }

To make this Task actually useful, we are also updating the InitializePreview() method from the first part:

        private async void InitializePreview()
        {
            captureManager = new MediaCapture();

            var cameraID = await GetCameraID(Windows.Devices.Enumeration.Panel.Back);

            await captureManager.InitializeAsync(new MediaCaptureInitializationSettings
            {
                StreamingCaptureMode = StreamingCaptureMode.Video,
                PhotoCaptureSource = PhotoCaptureSource.Photo,
                AudioDeviceId = string.Empty,
                VideoDeviceId = cameraID.Id,
            });

            StartPreview();
        }

In this case, we  selected the back camera. To make the MediaCapture API actually use this device, we need to generate a new instance of MediaCaptureInitializationSettings, where we select the cameras Id as VideDeviceId. If you now start capturing, this is an exemplary result:

wp_ss_20141115_0001

Rotating the preview

However, this not quite satisfying, because the preview automatically uses the landscape orientation. Luckily, this can be changed with just one single line of code (that needs to be added before actually starting the preview):

captureManager.SetPreviewRotation(VideoRotation.Clockwise90Degrees);

Now the result looks like this:

wp_ss_20141115_0002

Note: the black bands on both sides may happen due to the fact that most devices have a 15:9 ratio (WXGA, WVGA). On Devices like the Lumia 830 or 930, which have a 16:9 ratio, the preview will use the full screen in portrait mode. I tried a lot of things to get rid of those bands already, sadly without success. Once I found a proper solution, I will write another blog post and link it here on how to do it (any tips are welcome).

Limiting resolution

Sometimes, we need to limit resolutions (for example resolution limits on other parts in our app). This is possible by detecting the supported solutions and matching them to the screen ratio. As we are using the whole screen for previewing, of course we want to get our captured photo to use the same space, too.

My way to do this is to calculate the screen ratio, and return an enumeration value. This is the easiest way, and can be easily used in the further code to limit the resolution. The enumeration looks like this:

public enum CameraResolutionFormat
{
    Unknown = -1,

    FourByThree = 0,

    SixteenByNine = 1
}

And this is my helper to match the screen format (which is always wide screen on Windows Phone):

        private CameraResolutionFormat MatchScreenFormat(Size resolution)
        {
            CameraResolutionFormat result = CameraResolutionFormat.Unknown;

            double relation = Math.Max(resolution.Width, resolution.Height) / Math.Min(resolution.Width, resolution.Height);
            if (Math.Abs(relation - (4.0 / 3.0)) < 0.01)
            {
                result = CameraResolutionFormat.FourByThree;
            }
            else if (Math.Abs(relation - (16.0 / 9.0)) < 0.01)
            {
                result = CameraResolutionFormat.SixteenByNine;
            }

            return result;
        }

We could easily extend the calculation to 15:9, too. However, as the most camera resolutions are 4:3 or 16:9, this makes no sense in our use case (as 15:9 is still a widescreen format). The next thing we need to add is another helper to get the highest possible resolution for our photo and the preview. We are achieving this by generating a new object of type VideoEncodingProperties:

        private VideoEncodingProperties maxResolution()
        {
            VideoEncodingProperties resolutionMax = null;

            //get all photo properties
            var resolutions = captureManager.VideoDeviceController.GetAvailableMediaStreamProperties(MediaStreamType.Photo);

            //generate new list to work with
            List<VideoEncodingProperties> vidProps = new List<VideoEncodingProperties>();

            //add only those properties that are 16:9 to our own list
            for (var i = 0; i < resolutions.Count; i++)
            {
                VideoEncodingProperties res = (VideoEncodingProperties)resolutions[i];

                if (MatchScreenFormat(new Size(res.Width, res.Height)) != CameraResolutionFormat.FourByThree)
                {
                    vidProps.Add(res);
                }
            }

            //order the list, and select the highest resolution that fits our limit
            if (vidProps.Count != 0)
            {
                vidProps = vidProps.OrderByDescending(r => r.Width).ToList();

                resolutionMax = vidProps.Where(r => r.Width < 2600).First();                
            }

            return resolutionMax;
        }

What I am doing here: I read all available VideoEncodingProperties for the MediaStreamType Photo. As I mentioned before, we need only wide screen resolution for Windows Phone, that’s why I add only those that have not a 4:3 ratio to my list. Then I am using LINQ to order the list and select the highest resolution from that list.

Using this helper is also very easy, done with one line of code before starting the preview and best also before rotating the preview:

await captureManager.VideoDeviceController.SetMediaStreamPropertiesAsync(MediaStreamType.Photo, maxResolution());

This way, we are able to respect any resolution limits that we might face while developing our app, while keeping the photo quality as high as possible.

        private CameraResolutionFormat MatchScreenFormat(Size resolution)
        {
            CameraResolutionFormat result = CameraResolutionFormat.Unknown;

            double relation = Math.Max(resolution.Width, resolution.Height) / Math.Min(resolution.Width, resolution.Height);
            if (Math.Abs(relation - (4.0 / 3.0)) < 0.01)
            {
                result = CameraResolutionFormat.FourByThree;
            }
            else if (Math.Abs(relation - (16.0 / 9.0)) < 0.01)
            {
                result = CameraResolutionFormat.SixteenByNine;
            }

            return result;
        }

Focus

Focusing on objects in your photo is quite important. Sadly, it seems that currently we are not able to have a one solution fits all devices solution for using AutoFocus. I experimented a lot with it, and finally I got aware of known issues with Nokia drivers and the new MediaCapture API’s, as described here. Microsoft is working with Nokia (or their devices department) to fix this problem.

The only solution I got working for an Runtime app is to use manual focus. All other attempts gave me one Exception after the other, be it on cancelling the preview or be it on while previewing itself.  I’ll write another post on how to use the AutoFocus as soon as it is working like it should. In the meantime, here is my solution for manual focusing.

First, add a Slider control in your XAML page:

<Slider x:Name="FocusValueSlider" Maximum="1000" Minimum="0" Grid.Row="0" Margin="12,0,15,0" Header="adjust focus:" ValueChanged="FocusValueSlider_ValueChanged" Value="500" SmallChange="25" LargeChange="100" ></Slider>

Notice that as with any slider, you need to follow the order: Set Maximum first, then Minimum. If you do not, you will likely get an unusable Slider in return. If the VideoDeviceController.Focus property would work (seems like it is also affected by the above mentioned driver problems), we could read and set the Slider values from its MediaDeviceControl.Capabilities property. I tried to read them at any stage of previewing, but their values are always 0.0, null and false. The range up to 1000 fits in very well on all devices I tested (Lumia 920, 930 and 1020).

Ok, enough of whining. Let’s have a look at my solution. First, we need to generate a small helper that allows us to adjust the focus based on the slider values:

        private async void SetFocus(uint? focusValue = null)
        {
            //try catch used to avoid app crash at startup when no CaptureElement is active
            try
            {
                //setting default value
                if (!focusValue.HasValue)
                {
                    focusValue = 500;
                }

                //check if the devices camera supports focus control
                if (captureManager.VideoDeviceController.FocusControl.Supported)
                {
                    //disable flash assist for focus control
                    captureManager.VideoDeviceController.FlashControl.AssistantLightEnabled = false;

                    //configure the FocusControl to manual mode
                    captureManager.VideoDeviceController.FocusControl.Configure(new FocusSettings() { Mode = FocusMode.Manual, Value = focusValue, DisableDriverFallback = true });
                    //update the focus on our MediaCapture
                    await captureManager.VideoDeviceController.FocusControl.FocusAsync();
                }
            }
            catch { }
        }

This methods checks if the current camera supports Focus, and sets its value according to the slider. The AssistantLight is disabled in this case. Its default is enabled (true).

To add the possibility to adjust the focus, we need to configure our own FocusSettings that tell the camera that we are focusing manually based on the slider’s value. Finally, we need to perform the focusing action by calling the FocusControl’s FocusAsync method.

The next step is to hook up to changes in the slider values within the FocusValueSlider_ValueChanged event:

        private void FocusValueSlider_ValueChanged(object sender, RangeBaseValueChangedEventArgs e)
        {
            try
            {
                //convert double e.NewValue to uint and call SetFocus()
                uint focus = Convert.ToUInt32(e.NewValue);
                SetFocus(focus);
            }
            catch 
            {
                
            }
        }

Now every move of the slider will change the focus of the preview and of course also of the captured photo (which we will learn more about in the third post of this series).  To initialize our Focus correctly with the value of 500 we set in XAML, just call SetFocus(); before you start the preview. Here is the result:

focus screenshot

 

Disclaimer: I do not know if this follows best practices, but it works. If you have feedback for the above mentioned code snippets, feel free to leave a comment below.

In the third and last post I’ll show you how to save the images (also in different folders or only within the app).

Until then, happy coding!

Posted by msicc in Dev Stories, windev, 9 comments