GetPixel and getting a screenshot.

I got the following code in a DLL , also i got another function wrapping getPixel. But i got 2 problems with result i'm getting.

- One is that the image looks more stretched than it should be

- Secondly it doesn't seem to be a 24 bit image, but rather 256 colors

I searched the internet for a solution but couldn't find one. I'm also a bit puzzled on what selectObject really does, msdn is rather vague about it.

Anyone that can help me out with this ?

Code:

---

HDC doBitmap(HWND hwnd) {
        int result;
        int height = getHeight(hwnd);
        int width = getWidth(hwnd);
        HDC hdc = GetWindowDC(hwnd);
        if (hdc)
        {
                HDC hdcMem = CreateCompatibleDC(hdc);
                if (hdcMem)
                {
                        HBITMAP hbitmap = CreateCompatibleBitmap(hdc, width,height);
                        if (hbitmap)
                        {
                                SelectObject(hdcMem, hbitmap);
                                result = PrintWindow(hwnd, hdcMem, 0);
                                return hdcMem;
                        }
                        DeleteObject(hdcMem);
                }
                ReleaseDC(hwnd, hdc);
        }
        return 0;
}

---

What's GetPixel() got to do with making a screenshot?

I'm using the DLL in an different programming language, and i need to get a screen of a window in there. GetPixel is an easy way to retreive the color values.
And getpixel can use the HDC pointer.

Don't you think that copying the image to your bitmap using BitBlt() could be a little more effective?

Where whould you put the bitblt line... ?


i just tried to use bitblt , but then i'm not getting the window capture but just a part of it or the pixels on the same location of some other window that is on top of it.

Quote:

---

Originally Posted by pur

I'm also a bit puzzled on what selectObject really does, msdn is rather vague about it.

---

SelectObject() 'pushes' one GDI object into a DC, 'pop'ing the current GDI object of the same type out.

Your current code has lots of GDI leaks.
If using a .NET era IDE this _may_ not be too much of an issue, but it is bad code.
If in MSVC 6 this code will mean the PC needs a reboot after a few calls to thei function.

EDIT: use BitBlt()
After you have the Screen DC (GetDC(NULL))
After you have a compatible DC,
with a screen sized compatible BMP selected into the compatible DC

BitBlt() from the Screen DC to the compatible DC, the size of the screen.

then clean up

Code:

---


//select in created compatible BMP
HBITMAP hOrigBmp=(HBITMAP*)SelectObject(hdc,hCompatibleBmp);
//use

//return original GDI objects before clean up or we will leak the very limited GDI objects.
//or the object (mainly bmps) will fail to delete while 'in' a DC

SelectObject(hdc,hOrigBmp);//no need to catch return as we have the object.

//now we can clean up the bitmap
DelectObject(hCompatibleBmp);

//and the DC
//etc.....

---

If i do it like this , with use of bitbtl , i get a part of the screen back instead of the window. And it still seems to be in 256 colors. :/

Code:

---

HDC doBitmap(HWND hwnd) {
        int result;
        int height = getHeight(hwnd);
        int width = getWidth(hwnd);
        HDC hdc = GetWindowDC(hwnd);
        if (hdc)
        {
                HDC hdcMem = CreateCompatibleDC(hdc);
                if (hdcMem)
                {
                        HBITMAP hbitmap = CreateCompatibleBitmap(hdc, width,height);
                        if (hbitmap) {

                                SelectObject(hdcMem, hbitmap);

                                BitBlt(hdcMem, 0, 0, width, height, hdc, 0, 0, SRCCOPY);
                               
                                HBITMAP* hOrigBmp=(HBITMAP*)SelectObject(hdc,hbitmap);

                                SelectObject(hdc,hOrigBmp);

                                DeleteObject(hbitmap);


                                return hdcMem;
                        }
                        DeleteObject(hdcMem);
                }
                ReleaseDC(hwnd, hdc);
        }
        return 0;
}

---

Code:

---

                                SelectObject(hdcMem, hbitmap);

                                BitBlt(hdcMem, 0, 0, width, height, hdc, 0, 0, SRCCOPY);
                               
                                HBITMAP* hOrigBmp=(HBITMAP*)SelectObject(hdc,hbitmap);

---

Surely, you want to store the original bitmap when you FIRST select the hbitmap, not after you've already selected it. I'm pretty sure that hOrigBmp is actually going to be equal to hbitmap in this case.

How are you inspecting/saving/displaying the information gathered in your bitmap - or, how do you know it's an 8bpp image?

This link is an example of how to do a screen copy in a standalone application:
http://msdn.microsoft.com/en-us/libr...02(VS.85).aspx

--
Mats

I'm using getPixel on hdcMem . I'm getting more and more confused about the code you guys supply-ing and even if it's all needed.

Quote:

---

Originally Posted by pur

I'm using getPixel on hdcMem . I'm getting more and more confused about the code you guys supply-ing and even if it's all needed.

---

All the previous posts should have helped you get to the point of BitBlting. Your problem probably lies in what you do after the BitBlting. I've used the above posts to get to the point of BitBlting. At that point I wrote two additional functions. One to build the bitmap header and the second to write the desktop snapshot out to a bitmap file. Herein probably lies your problem. Incorrectly building the bitmap header structure and/or incorrectly writing the data out to a bitmap file will provide strange (incorrect) results.

Quote:

---

Originally Posted by pur

If i do it like this , with use of bitbtl , i get a part of the screen back instead of the window. And it still seems to be in 256 colors. :/

---

As you delete the bitmap you have drawn to (hbitmap) prior to any chance to use it, I would assume you get a totally black screen.

A HDC is a holder for GDI objects.

The GDI object you want is the BMP (which you can get from the HBITMAP), not the HDC.

If in doubt do a search, I posted code to do screen capture years ago.

Watch 24bit colour depth (if doing conversions to a consistent colour depth) where the colours are reversed (compared to 16 or 32 bit).

Ah ok, but there isn't any equivalant of getpixel on HBitmap.

Why do you need GetPixel()?

You need to write the BMP parts.

GetDIBits() (all at once) will be faster than individually getting each pixel.




here is some code for the header

http://cboard.cprogramming.com/showthread.php?t=40807

rest should be around here somewhere......

Quote:

---

Why do you need GetPixel()?

You need to write the BMP parts.

GetDIBits() (all at once) will be faster than individually getting each pixel.

---

I got the functions in a win32 DLL for wrapping. Because i'm using it in an different programming language. In that language i only have a setPixel function i can use.

Is there maby a possiblity to retreive color values and position with use of the HBITMAP pointer?

Yes...

GetDIBits()


Which language are you using that can access some of the WIN32 API (in a DLL) but not all of it?

You do realize that you do not need either GetPixel or SetPixel calls. Both can be accomplished by the use of bit shifting to get or set the appropriate RGB values.

Code:

---

        HDC hdcc = CreateCompatibleDC(hdc);
//hdc - is your dc which you copy from

        HBITMAP hBm = CreateCompatibleBitmap(hdc, iSrcW, iSrcH);
//iSrcW, iSrcH - size of you sourse hdc
        HBITMAP hOldBm = (HBITMAP)SelectObject(hdcc, hBm);
        BitBlt(hdcc, 0, 0, iSrcW, iSrcH, hdc, iX, iY, SRCCOPY);

// now, get info from copied bitmap
        BITMAP bm;
        memset(&bm, 0, sizeof BITMAP);
        int iErr = GetObject(hBm, sizeof(BITMAP), &bm);
        iErr = GetLastError();
        iBPP = bm.bmBitsPixel;
        DIBSECTION dis;
        iErr = GetObject(hBm, sizeof(DIBSECTION), &dis);

        int length = (bm.bmWidth * bm.bmHeight) * (bm.bmBitsPixel>>3);
        if(0 == length)
                return NULL;// nothing to do

// Ok, bitmap info now
        BITMAPINFO bi;
        memset(&bi, 0, sizeof BITMAPINFO);
        BYTE *pBits = new BYTE[length];

        bi.bmiHeader.biSize          = sizeof(BITMAPINFOHEADER);
        bi.bmiHeader.biWidth        = bm.bmWidth;
        bi.bmiHeader.biHeight        = bm.bmHeight;
        bi.bmiHeader.biPlanes        = 1;
        bi.bmiHeader.biBitCount      = bm.bmPlanes * bm.bmBitsPixel;
        bi.bmiHeader.biCompression  = 0;
        bi.bmiHeader.biSizeImage    = 0;
        bi.bmiHeader.biXPelsPerMeter = 0;
        bi.bmiHeader.biYPelsPerMeter = 0;
        bi.bmiHeader.biClrUsed      = 0;
        bi.bmiHeader.biClrImportant  = 0;

        LPBYTE lpBits = (LPBYTE) GlobalAlloc(GMEM_FIXED, length);

        int iBits = GetDIBits(hdc, hBm, 0, bm.bmHeight, lpBits, &bi, DIB_RGB_COLORS);

//.......  now you got bits of your pixels in lpBits
// remember, that y-axis begun from bottom of screen
// and don't forget:

        GlobalFree(lpBits);
        SelectObject(hdcc, hOldBm);
        DeleteObject(hBm);
        DeleteDC(hdcc);

---

Good luck!

Quote:

---

Originally Posted by DeskEx

Good luck!

---

See MSDN code.

Seriously, forget that SetPixel and GetPixel exist. They are completely useless. Using them is like walking down the road to the store to buy a single paperclip every time you need one, even though you go through about 200+ per day.
Bitblt on the other hand would be like buying a bag full of them.