为什么绘制到隐藏的 HDC 会给出与绘制到 GetDC(HWND) 返回的 HDC 时不同的结果

Question

I have two methods, paintDoubleBuffered and paint. They are both supposed to draw this image on the screen:

The image is made up of approximately 12 smaller images each sized 256x256 tiled together.

My standard painting method works as expected. Here it is:

void MainWindow::paint(HWND hwnd) {

    HDC hdc{GetDC(hwnd)};
    paint(hwnd, hdc);
    ReleaseDC(hwnd, hdc);

}

void MainWindow::paint(HWND hwnd, HDC hdc) {
    constexpr INT img_width{ MDNR_Map::pannel_width };
    constexpr INT img_height{ MDNR_Map::pannel_height };

    const INT width{ GetDeviceCaps(hdc, HORZRES) };
    const INT height{ GetDeviceCaps(hdc, VERTRES) };

    const INT num_width_pannels{ (width / img_width) + 1 };
    const INT num_height_pannels{ (height / img_height) + 1 };

    Gdiplus::Graphics g(hdc);

    g.SetCompositingMode(CompositingMode::CompositingModeSourceCopy);

    g.SetInterpolationMode(InterpolationMode::InterpolationModeNearestNeighbor);

    for (INT y = 0; y < num_height_pannels; y++) {
        for (INT x = 0; x < num_width_pannels; x++) {

            Location_t get_loaction(x + map_location.x, y + map_location.y, map_location.layer);
            const IMG_t v{ mdnr_map.get(get_loaction) };

            const Point drawPoint((INT)(img_width * x), (INT)(img_height * y));

            Status stat{ g.DrawImage(v, drawPoint) };

            if (stat != Status::Ok)
            {
                throw std::runtime_error(":(");
            }


        }
    }
}

The issue with that paint method is that mdnr_map.get is an io bound call and may take several micro seconds. Because I need to call it about 12 times, it can lead to flickering.

To solve this, I attempted to write a double-buffered paint method, which is as follows:

void MainWindow::paintDoubleBuffered(HWND hwnd) {

    // Get DC for window
    HDC hdc{ GetDC(hwnd) };

    const INT win_width{ GetDeviceCaps(hdc, HORZRES) };
    const INT win_height{ GetDeviceCaps(hdc, VERTRES) };

    // Create an off-screen DC for double-buffering
    HDC hdcMem{ CreateCompatibleDC(hdc) };

    HBITMAP hbmMem{ CreateCompatibleBitmap(hdc, win_width, win_height) };

    HANDLE hOld{ SelectObject(hdcMem, hbmMem) };

    // Draw into hdcMem here
    paint(hwnd, hdcMem);

    // Transfer the off-screen DC to the screen
    BitBlt(hdc, 0, 0, win_width, win_height, hdcMem, 0, 0, SRCCOPY);

    // Free-up the off-screen DC
    SelectObject(hdcMem, hOld);

    DeleteObject(hbmMem);
    DeleteDC(hdcMem);
}

However, this does not work and instead produces this abombination of an image:

With a little poking and proding, I was able to discover that if I changed my double buffered paint method by multiplying the image size by 1.5, the image was no longer so garbled, but was now zoomed in by a factor of 1.5

void MainWindow::paintDoubleBuffered(HWND hwnd) {

    // Get DC for window
    HDC hdc{ GetDC(hwnd) };

    const INT win_width{ GetDeviceCaps(hdc, HORZRES) };
    const INT win_height{ GetDeviceCaps(hdc, VERTRES) };

    // Create an off-screen DC for double-buffering
    HDC hdcMem{ CreateCompatibleDC(hdc) };

    HBITMAP hbmMem{ CreateCompatibleBitmap(hdc, win_width, win_height) };

    HANDLE hOld{ SelectObject(hdcMem, hbmMem) };

    // Draw into hdcMem here
    
    constexpr INT img_width{ MDNR_Map::pannel_width + 128 }; // MDNR_Map::pannel_width is 256
    constexpr INT img_height{ MDNR_Map::pannel_height + 128}; // MDNR_Map::pannel_height is 256

    const INT num_width_pannels{ (win_width / img_width) + 1 };
    const INT num_height_pannels{ (win_height / img_height) + 1 };

    Gdiplus::Graphics g(hdcMem);

    g.SetCompositingMode(CompositingMode::CompositingModeSourceCopy);

    g.SetInterpolationMode(InterpolationMode::InterpolationModeNearestNeighbor);

    for (INT y = 0; y < num_height_pannels; y++) {
        for (INT x = 0; x < num_width_pannels; x++) {

            Location_t get_loaction(x + map_location.x, y + map_location.y, map_location.layer);
            Gdiplus::Bitmap* pannel{ mdnr_map.get(get_loaction) };

            const Point drawPoint((INT)(img_width * x), (INT)(img_height * y));

            Status stat{ g.DrawImage(pannel, drawPoint) };
            if (stat != Status::Ok)
            {
                throw std::runtime_error(":(");
            }

        }
    }

    // Transfer the off-screen DC to the screen
    BitBlt(hdc, 0, 0, win_width, win_height, hdcMem, 0, 0, SRCCOPY);

    // Free-up the off-screen DC
    SelectObject(hdcMem, hOld);

    DeleteObject(hbmMem);
    DeleteDC(hdcMem);
}

My question is why does drawing to the HDC returned by CreateCompatibleBitmap produce a different result than drawing to the HDC returned by GetDC?

I have tried: All raster-operation codes for BltBlt. I have checked that the temporary HDC is the same size as the window. I have tried replacing the code snippet

const INT win_width{ GetDeviceCaps(hdc, HORZRES) };
const INT win_height{ GetDeviceCaps(hdc, VERTRES) };

with

RECT rect;
GetWindowRect(hwnd, &rect);

const INT win_width{ rect.right - rect.left };
const INT win_height{ rect.bottom - rect.top };

I have also called SetProcessDPIAware() before drawing.

Upon feedback from @Paul Sanders, I rewrote my paintDoubleBuffered method as follows, NOTE, I have called BufferedPaintInit in the object constructor:

void MainWindow::paintDoubleBuffered(HWND hwnd) {

    PAINTSTRUCT ps;
    HDC hdc{ BeginPaint(hwnd, &ps)};

    RECT sz;
    GetWindowRect(hwnd, &sz);

    BP_PAINTPARAMS paintParams = { 0 };

    paintParams.cbSize = sizeof(paintParams);
    paintParams.dwFlags = BPPF_ERASE;
    paintParams.pBlendFunction = NULL;
    paintParams.prcExclude = NULL;

    HDC hdcBuffer;

    HPAINTBUFFER hBufferedPaint = BeginBufferedPaint(hdc, &sz, BPBF_COMPATIBLEBITMAP, &paintParams, &hdcBuffer);

    if (hBufferedPaint && this->bufferedInitResult == Ok) {
        // Application specific painting code
        paint(hwnd, hdcBuffer);
        EndBufferedPaint(hBufferedPaint, TRUE);
    }
    else{
        paint(hwnd, hdc);
    }

    ReleaseDC(hwnd, hdc);
}

Unfortunately, this did not work and the resultant screen looks like this:

Answer 1

The issue ended up being not in the approach to double buffering, but in the call to Graphics::DrawImage(Image*, Gdiplus::Point) in my paint method. Changing to DrawImage(Image* image, INT x, INT y, INT width, INT height) fixed the scaling issue.