I'm currently using the QWidget::grab() function to acquire a QFrame's pixmap (and all of its children), but the function doesn't seem to take into account if the widget doesn't have any background.
You see, my QFrame is set to "setAutoFillBackground(false)", but when its pixmap is grabbed, it seems to paint the default light-pinkish background instead of full transparency.
Replacing the pixmap with a picture containing an alpha channel works fine.
The situation I'm using this in is with QGL, so the pixmap is getting rendered later on as a texture.
I changed the frame's palette's background to contain 0 alpha. This fixed the program.
Although I still believe that the grab function should take into account the bool that was set for filling the background or not - since not autofilling the background equates to the same net visual effect normally, just not when grabbed.
Related
I'm trying to do something like what Auslogics Disk Defrag does with its custom window:
As can be seen, the blurred semi transparent shadow surrounding the window is much darker than the standard one, so the program must be drawing it by itself. The problem is, I can't find a way to paint anything transparent around a window.
In an answer to a similiar question, someone suggested creating a slightly bigger transparent window (using WS_EX_LAYERED + SetLayeredWindowAttributes()) behind the actual application window, and then do the translucent drawing on the transparent one. Not only does it sound like an ugly hack, it doesn't actually work. If, for example, one tries to draw a semi transparent black rectangle on a transparent window via GDI+, alpha blending is applied to the shape's color over the window background color (which would also be the transparency color) and then the shape is drawn with the calculated color, which obviously is not the window transparency, resulting in an opaque rectangle.
The semi transparent shadow is actually done by Gaussian Blur of the black square.
You can use this effect to create glows and drop shadows and use the
composite effect to apply the result to the original image. It is
useful in photo processing for filters like highlights and shadows.
You can use the output of this effect for input into lighting effects,
like the Specular Lighting or Diffuse Lighting effects, because the
alpha channel is blurred, too and lighting effects use the alpha
channel to determine surface geometry as a height map.
This effect is used by the built-in Shadow effect.
Refer: Gaussian blur effect
Then remove the standard frame, the entire window is your client area, so you can draw shadow in the extended frame.
Refer: Drawing in the Extended Frame Window
I think I've found a solution that works for me. I was hoping I wouldn't have to create an extra window just for the shadow, but every method I could find or think of would require me to draw all the controls by myself and/or somehow correct their alpha values.
So, I'm using a layered window with per pixel alpha. I paint over it with Direct2D, or, alternatively, I use some PNGs with transparency for the edges and corners of the shadow and copy them on a memory DC. Either way I simply recreate the shadow with the right dimensions when the window is resized, and call UpdateLayeredWindowIndirect. Both methods seem to be working well on Windows 7 and 10, and so far I haven't found any glitches.
Preventing it from showing on the taskbar was a bit tricky. All the ways I know have drawbacks. What worked best for me was making the layered window owned by the main one. At least that way it will only be visible on the desktop where the program is actually running, unlike other alternatives which would force it to show on every virtual desktop. Finally, because I disable that window, I interact with it by processing WM_SETCURSOR.
I've put two label of the same dimension overlapped on my GUI. On the first, that is on the back there is one image set by the pixmap of the label. The second label has another image set by pixmap that is trasparent in some pixel. During design i can see the back image via the front image. When I execute the program, the trasparent pixel of the second image are transformed into white pixel, so i cant see the back image. I'm using QT 5.10. The image are in .png format.
I've resolved putting autoFillBackground on and i've changed the color of window in palette putting a transparent color
I have an app written in Qt5.10 using QOpenGLWidget. I use raw OpenGL calls to draw on the widget.
If I have some viewport area that is drawn with alpha <1.0 this area is not drawn anew, but rather it looks like it's drawn over the previous state of the widget.
When I switch between different Windows apps, it's drawing over previous contents of the screen rendered by a different apps.
I don't change the default update behaviour of the widget.
I set the default SurfaceFormat as follows:
QSurfaceFormat surfaceFormat = QSurfaceFormat::defaultFormat();
surfaceFormat.setRedBufferSize(8);
surfaceFormat.setGreenBufferSize(8);
surfaceFormat.setBlueBufferSize(8);
surfaceFormat.setAlphaBufferSize(8);
surfaceFormat.setStencilBufferSize(8);
surfaceFormat.setDepthBufferSize(16);
surfaceFormat.setSamples(4);
surfaceFormat.setRenderableType(QSurfaceFormat::RenderableType::OpenGL);
surfaceFormat.setProfile(QSurfaceFormat::OpenGLContextProfile::CompatibilityProfile);
surfaceFormat.setSwapBehavior(QSurfaceFormat::DoubleBuffer);
QSurfaceFormat::setDefaultFormat(surfaceFormat);
I was trying to add calls to glClear and glInvalidateFramebuffer in random places in the code, but it didn't seem to help.
What should I do to force redrawing of the widget?
I am using Qt5.6, I have developed several widgets that render the content to an off-screen bitmap then copy the final image to the visible area.
I have an area on the visible display that shows a video feed, I want to copy the images over the video without overwriting the background and avoiding flicker.
I am currently creating the off-screen image using a 'QPixmap', I then create a painter using the Pixmap and draw as to the off-screen image. When the image is ready I then call the 'toImage' function to return a 'QImage' and then copy this to the visible display.
A lot of the widget contains lines and circles, a lot of which are not filled.
I've seen other posts not using a QPixmap, just using a 'QImage', should I be using a 'QPixmap' at all?
Question is how to copy the image from the off-screen area to the visible area without overwriting the background?
The key to transparency is that the overlay image has got an alpha channel. QPixmap uses the graphics format of the underlying graphics system which should include an alpha channel. For QImage, the format can be explicitly specified and it should be QImage::Format_ARGB32_Premultiplied, see [1]: http://doc.qt.io/qt-5/qimage.html#Format-enum
To get a a fully transparent QImage/QPixmap in the first place, call QPixmap/QImage::fill(QColor(0, 0, 0, 0)); before creating the QPainter.
The 4th parameter is the alpha channel which is 255 (full opacity) by default).
Unfortunately I can't give advice whether QPixmap or QImage is faster for your setup.
Provided the compositing operation with the videofeed considers the alpha-channel, this should solve your problem.
I'm creating a non-intrusive popup window to notify the user when processing a time-consuming operation. At the moment I'm setting its transparency by calling SetLayeredWindowAttributes which gives me a reasonable result:
alt text http://img6.imageshack.us/img6/3144/transparentn.jpg
However I'd like the text and close button to appear opaque (it doesn't quite look right with white text) while keeping the background transparent - is there a way of doing this?
In order to do "proper" alpha in a layered window you need to supply the window manager with a PARGB bitmap by a call to UpdateLayeredWindow.
The cleanest way to achieve this that I know of is the following:
Create a GDI+ Bitmap object with the PixelFormat32bppPARGB pixel format.
Create a Graphics object to draw in this Bitmap object.
Do all your drawing into this object using GDI+.
Destroy the Graphics object created in step 2.
Call the GetHBITMAP method on the Bitmap object to get a Windows HBITMAP.
Destroy the Bitmap object.
Create a memory DC using CreateCompatibleDC and select the HBITMAP from step 5 into it.
Call UpdateLayeredWindow using the memory DC as a source.
Select previous bitmap and delete the memory DC.
Destroy the HBITMAP created in step 5.
This method should allow you to control the alpha channel of everything that is drawn: transparent for the background, opaque for the text and button.
Also, since you are going to be outputting text, I recommend that you call SystemParametersInfo to get the default antialiasing setting (SPI_GETFONTSMOOTHING), and then the SetTextRenderingHint on the Graphics object to set the antialiasing type to the same type that is configured by the user, for a nicer look.
I suspect you'll need two top level windows rather than one - one that has the alpha blend and a second that is display above the first with the opaque text and button but with a transparent background. To accomplish this with a single window you'll need to use the UpdateLayeredWindow API call, but using this will cause your buttons to not redraw when they are interacted with (hover highlights, focus etc.)
It is possible that if this application is for Vista only there is a new API call that you can use, but I do not believe it is available in XP or earlier.
I can't say for sure, you'll need to try it, but since everything is a window, you could try setting the layered attributes for your button to make it opaque.
As for the text, you may be able to put that in its own frame with a set background and foreground color, and modify its layered attributes to make the background color transparent...
But since these are child windows and not the top-level window, I really don't know that it'll work.