QImage::loadFromData() behaviour

[Caius Aérobus]

Hello,
below a piece of code, which consists in loading image data in false color mode (or at least this is what I expected to do!) and retrieving them immediatly. I have deleted additionnal code, which includes a color table construction, to be attached to the image.
I obviously make a mistake somewhere or more likely I have not understood the way QImage interprets the raw data. I have not found any response to this specific problem in reading the various posts on loadFromData method, including http://www.qtcentre.org/forum/f-qt-p...t=loadFromData so this is the reason I create a new topic.
Thanks in advance for your help.

Code:

this->Print("size is (%u,%u)", size[0], size[1]);
  uchar *data;
  if ((data = (uchar *)malloc(size[0]*size[1]*sizeof(uchar))) == NULL) {
    this->Error("memory allocation error (%d)", size[0]*size[1]*sizeof(uchar));
    return;
  }
  QImage *image = new QImage(QSize(size[0], size[1]), QImage::Format_Indexed8);
  for (int j=0 ; j<size[1] ; j++)
    for (int i=0 ; i<size[0] ; i++)
      data[j*size[0]+i] = i==0 && j==0 ? 255 : 0;
  for (int j=0 ; j<size[1] ; j++) {
    for (int i=0 ; i<size[0] ; i++) {
      this->Print(" %u\r", data[j*size[0]+i]);
    }
    this->Print("\n");
  }
  image->loadFromData(data, field.size());
  for (int j=0 ; j<size[1] ; j++) {
    for (int i=0 ; i<size[0] ; i++) {
      QRgb rgb = image->pixel(i, j);
      this->Print(" %u\r", image->bits()[j*size[0]+i]);
    }
    this->Print("\n");
  }

this->Print("size is (%u,%u)", size[0], size[1]);
  uchar *data;
  if ((data = (uchar *)malloc(size[0]*size[1]*sizeof(uchar))) == NULL) {
    this->Error("memory allocation error (%d)", size[0]*size[1]*sizeof(uchar));
    return;
  }
  QImage *image = new QImage(QSize(size[0], size[1]), QImage::Format_Indexed8);
  for (int j=0 ; j<size[1] ; j++)
    for (int i=0 ; i<size[0] ; i++)
      data[j*size[0]+i] = i==0 && j==0 ? 255 : 0;
  for (int j=0 ; j<size[1] ; j++) {
    for (int i=0 ; i<size[0] ; i++) {
      this->Print(" %u\r", data[j*size[0]+i]);
    }
    this->Print("\n");
  }
  image->loadFromData(data, field.size());
  for (int j=0 ; j<size[1] ; j++) {
    for (int i=0 ; i<size[0] ; i++) {
      QRgb rgb = image->pixel(i, j);
      this->Print(" %u\r", image->bits()[j*size[0]+i]);
    }
    this->Print("\n");
  }

To copy to clipboard, switch view to plain text mode 

Output:

wlFieldImage: size is (4,4)
wlFieldImage: Original values:
wlFieldImage:  255   0   0   0
wlFieldImage:    0   0   0   0
wlFieldImage:    0   0   0   0
wlFieldImage:    0   0   0   0
wlFieldImage: Extracted values:
wlFieldImage:   40  59  19   8
wlFieldImage:    0   0   0   0
wlFieldImage:  216 103  12   8
wlFieldImage:  184 193   9   8

wlFieldImage: size is (4,4)
wlFieldImage: Original values:
wlFieldImage:  255   0   0   0
wlFieldImage:    0   0   0   0
wlFieldImage:    0   0   0   0
wlFieldImage:    0   0   0   0
wlFieldImage: Extracted values:
wlFieldImage:   40  59  19   8
wlFieldImage:    0   0   0   0
wlFieldImage:  216 103  12   8
wlFieldImage:  184 193   9   8

To copy to clipboard, switch view to plain text mode 

[^NyAw^]

Hi,

Maybe you have a problem setting the data using integers. "data" is defined as "uchar*" and you are indexing it using "ints".
The first pixel must be "data", second pixel "data+1 (byte)", and so on. One integer is 32 bytes so you are jumping to a different position.

Is this your problem? I had been using QImage like you one year ago and I made some similar error.

[Caius Aérobus]

Hi,

Maybe you have a problem setting the data using integers. "data" is defined as "uchar*" and you are indexing it using "ints".
The first pixel must be "data", second pixel "data+1 (byte)", and so on. One integer is 32 bytes so you are jumping to a different position.

Is this your problem? I had been using QImage like you one year ago and I made some similar error.

indexing cannot be done but with ints!
Your error was probably more likely something like this:
int *p = data;
for (...)
*p++ = some value;
or int *data then loadFromImage(data, ....)
which is of course a mistake.
But nothing wrong in adressing the data in my code...unfortunately because my error remains

[^NyAw^]

Hi,

But nothing wrong in adressing the data in my code...unfortunately because my error remains

So nothing is wrong but it doesn't works. What are you expecting the code do?

[Caius Aérobus]

Hi,



So nothing is wrong but it doesn't works. What are you expecting the code do?

I am expecting to get the same data I entered before as I extract them using QImage::bits()!

[Caius Aérobus]

Hi,



So nothing is wrong but it doesn't works. What are you expecting the code do?

I am expecting to get the same data I entered before as I extract them using QImage::bits()!
Actually I build a pixmap from this image and obviously the pixmap is wrong!
That is the reason I checked which data I get from the image and discovered data have been altered during their storage in QImage. So probably this is a normal behaviour but at least:
- I would like to understand this behaviour,
- I would like to know how to create a false color image from a list of color table indexes.

[mchara]

Hi,
loadFromData method isn't meant to read image from raw pixels matrix, but to read binary data in a specific format, i.e. when you read contents of *.png file and pass it to loadFromData method.

You shall create an empty image with given size & use setPixel to fill it.

If you really want to generate data outside the image, see PPM format - it's matrix of pixel values (with color depth and dimensions at the beginning) but in textual format.

[Caius Aérobus]

Ok, thanks for these explainations, I will use setPixel() instead.
Regards.

[anarsynd]

In many situations it is quite slow to loop through the image using setPixel. What I do is to use a constructor that accepts raw data, I cite:

###
QImage::QImage ( uchar * data, int width, int height, Format format )

Constructs an image with the given width, height and format, that uses an existing memory buffer, data. The width and height must be specified in pixels, data must be 32-bit aligned, and each scanline of data in the image must also be 32-bit aligned.
### ref: http://doc.trolltech.com/4.2/qimage.html#QImage-4

[mchara]

Yeah,
anarsynd is right: constructors data is a raw data, it's not the same as data in loadFromData which is binary image coded to some format, so this is what you need.

[Caius Aérobus]

Yeah,
anarsynd is right: constructors data is a raw data, it's not the same as data in loadFromData which is binary image coded to some format, so this is what you need.

not exactly because I have to assign new content to the image after it has been constructed, in a slot, and I cannot create a new image because I would have to take care of the image state, which would be complicated, so setPixel() is ok for me, although not so efficient.