Child classes that share Parent class member - c++

Question

I have an awkward question. I think it's impossible but I need to know for sure. It's a bit of an odd request, but I need my child classes to share the SAME Pixels vector from the parent class.
Basically, I want to create an instance of Image class. That Image class will hold Pixels for both Bitmap and Png Class so that if I need to convert from Bitmap To PNG and Vice-versa, they use the same vector rather than me creating both Bitmap and PNG classes.
class Image
{
private:
std::vector<RGB> Pixels;
};
class Bitmap : Image
{
public:
Bitmap() : Image() {};
};
class Png : Image
{
public:
Png() : Image() {};
};
Such that when I do:
int main()
{
Image Img();
Img.GetBitmapPixels(); //This
Img.GetPngPixels(); //And this, return the same Pixels Vector.
Bitmap Foo = Img.ToPng();
Png Moo = Img.ToBitmap();
//Such that both Foo and Moo have the exact same underlying Pixels Vector.
}
Currently my classes look like:
class Bitmap
{
private:
std::vector<RGB> Pixels;
public:
Bitmap();
std::vector<RGB> GetPixels() {return Pixels;}
void SetPixels(std::vector<RGB> Pixels) {this->Pixels = Pixels;}
};
class Png
{
private:
std::vector<RGB> Pixels;
public:
Png();
std::vector<RGB> GetPixels() {return Pixels;}
void SetPixels(std::vector<RGB> Pixels) {this->Pixels = Pixels;}
};
And to convert from one to the other, I have to do:
int main()
{
Bitmap Bmp();
Png PNG();
PNG.SetPixels(BMP.GetPixels); //BMP has to COPY PNG's Pixels and vice-versa..
}
It's kind of a stupid question. I just don't want to copy pixels. I just want to be able to convert between the two classes without any copying as both classes hold an std::vector Pixels member and the data is aligned the same.
I guess I'm trying to be able to do: PNG.SaveAsBitmap(...); Or BMP.SaveAsPNG(...); without creating a new instance of the other.
How can I avoid copying/creating and making a new instance of the other class I'm converting to? Can I have classes that inherit from eachother?

Answer 1

I may be oversimplyfying this. How are you loading your data? If it is the same for each image type could you have a load in the image class and just overload the Translation functions. I am not an expert on image data but could you do something similiar to:
public Image
{
private Pixels data;
public void GetPixels();
public virtual Pixels ToFormat(FormatEnum format);
}
public Bitmap:Image
{
public override Pixels ToFormat(FormatEnum format)
{
switch(format){
case FormatEnum.Bitmap:return data;break;
case FormatEnum.Png:return doSomethingElse();break;
}
}
}
int main()
{
Image Img();
Img.GetPixels();
Bitmap Foo = Img.ToFormat(Format.BitMap);
Png Moo = Img.ToFormat(Format.Png);
}

Answer 2

Since they share a parent as a common factor, you can elect to use pointers (this is considered a shallow copy), but this is usually not wise a deep copy is usually better. Perhaps you can create two functions one to create a shallow copy and one for a deep copy. Since a Bitmap and a Png are separate entities (they are siblings). Perhaps you should allow BMP to be the parent of png?
You may also elect to decouple the encoding altogether for saving purposes and just deal with the raw image data.