QImage::pixel returns the same RGB values for entire image on iOS

[JRohner]

Hi!

I have a QVideoWidget from which i grab a pixmap which i convert to a QImage. When looking at the rgb values compiling on my desktop mac i get reasonable values (matching the actual image). When compiling to iOS however all rgb values are the same, eg ( 239 , 235 , 231 , 255 ), for each pixel.

Anybody knows whats going on?

image = videoWidget->grab().toImage();
...
for ( int row = 0; row < image.height(); ++row )
        for ( int col = 0; col < image.width(); ++col )
        {
            QColor clrCurrent( image.pixel( col, row ) );


            qDebug()<< "Pixel at [" << col << "," << row << "] contains color ("
                    << clrCurrent.red() << ", "
                    << clrCurrent.green() << ", "
                    << clrCurrent.blue() << ", "
                    << clrCurrent.alpha() << ").";


        }

[SGaist]

Hi,

Silly question but: are you sure that image is a valid image ?

[JRohner]

@SGaist Hi and thanks for responding :-) Yes the image seems valid since i can show it as a pixmap on a QLabel on the iPhone. Everything looks fine on screen (when i fixed QSize policy as you suggested in another post): in the upper part of the screen i have the video widget showing a live camera feed, in the bottom the image grab from the video widget is shown as a still picture; the live video widget and the grabbed image look exactly the same and have the same dimensions.

Since i made the changes you suggested concerning QSize policy, the situation has improved: RGB values for the middle portion of the image (on the y axis) are now reasonable, mirroring what the camera is pointed at. But RGB values (from the output of clrCurrent(image.pixel(col, row));) for the upper and lower part of the image matrix (on the y axis) are all the same.

Is
image = videoWidget->grab().toImage(); grabbing more than the actual video widget?

On the other hand the RGB values I'm getting might actually make sense... The actual video shown occupies a smaller portion of the upper 50% part of the screen (to preserve aspect ratio); the rest of the area is white'ish. This fits well with the kind of RGB values I'm getting (e.g. 239 , 235 , 231 , 255 ) above and below the RGB values that are reasonable and correspond to what the camera is pointed at. So in summary my understanding is that videoWidget->grab takes 50% of the screen (because i added two things to the layout), the actual video is smaller, maybe 30%. Thats why I'm getting white in the upper and lower part when calling videoWidget->grab.

Thanks!

Qt code:

#include "mainwindow.h"
#include "ui_mainwindow.h"

MainWindow::MainWindow(QWidget *parent) :
    QMainWindow(parent),
    ui(new Ui::MainWindow)
{

    ui->setupUi(this);
    
    camera = new QCamera(this);
    videoWidget = new QVideoWidget(this);
    imageLabel = new QLabel(this);
    
    camera->setViewfinder(videoWidget);
    ui->verticalLayout->addWidget(videoWidget);
    ui->verticalLayout->addWidget(imageLabel);
    videoWidget->setSizePolicy(QSizePolicy::Fixed,QSizePolicy::Fixed);
    videoWidget->show();
    camera->start();

    QTimer::singleShot(3000, this, SLOT(captureFrame()));

}

void MainWindow::captureFrame(){

   image = videoWidget->grab().toImage();
   imageLabel->setPixmap(QPixmap::fromImage(image));
   imageLabel->setSizePolicy(QSizePolicy::Fixed,QSizePolicy::Fixed);
   
   QTimer::singleShot(3000, this, SLOT(outputInfo()));

}

void MainWindow::outputInfo(){

    qDebug()<<image;
    for (int row = 0; row < image.height(); row = row + 10){
        for (int col = 0; col < image.width(); col = col + 10){
             QColor clrCurrent(image.pixel(col, row));
             qDebug()<< "Pixel at [" << col << "," << row << "] contains color ("
                     << clrCurrent.red() << ", "
                     << clrCurrent.green() << ", "
                     << clrCurrent.blue() << ", "
                     << clrCurrent.alpha() << ").";
         }
    }
    camera->stop();
}

MainWindow::~MainWindow()
{
    delete ui;
}

[SGaist]

Why are you using grab to get the image from the camera ?

[JRohner]

@SGaist I want to use pixel information in an object recognition app, passing RGB values to a feature detection algorithm. Grabbing a QImage was what i thought of first...

Is there a better way? I saw somewhere that you can set capture destination for QCamera to a byte buffer. Problem is that i have no idea how to get RGB values from that point (I'm kind of new to C++ and Qt).

Thanks :-)

[SGaist]

QVideoProbe comes to mind for that

[JRohner]

@SGaist Thanks! Yes this seems like a good idea.