I am writing an app which draws text and shapes in a ClientDC of a Frame. When I run the app under my Fedora 16(Gnome 3) nothing is drawn in the Frame, but if I run it under Windows all drawings display normally.
I've tried using WindowDC to do the drawing on, but it is not different to ClientDC under Fedora. I can only get a successful drawing when using PaintDC.
Am I doing something wrong(or missing something), or is it just Linux/Fedora/Gnome 3?
I think the recommended way to paint is with the newer wx.GCDC with a fallback to the wx.PaintDC. If you need advanced drawing, see FloatCanvas, cairo or the glcanvas.
Related
I have a rich text label that works fine under a resolution of 1920x1080. However, when scaled down to the resolution on my laptop, which has a 1366x768 resolution, the text on said label becomes janky and malformed.
Some lines are cut off at the top or the bottom, and others are squished (as you can see in the image at lines 3, 6, and 13).
I'm using Godot v3.5.1 and the text font is Noto Sans Regular from here
I tried enabling mipmaps, using the filter, disabling anti aliasing, disabling font oversampling and enabling GPU pixel snap under Project Settings > Rendering > 2D > Snapping. And out of all of those, only the pixel snap setting worked. Completely fixes the issue and the text is rendered properly.
However, this completely breaks an animation of a spinning circle that plays at basically all times during the actual gameplay. It becomes stuttered and shakes instead of the normally smooth animation it has otherwise. I realize this may possibly be fixed by using SVG instead of PNG sprites, but I feel like that's not an ideal solution in case other sprites get added. Especially since my game will also allow community members to add their own sprites for their own game play.
So is there any way to fix the text without breaking the animations, or make it so that the animations don't break with pixel snap enabled?
All my fonts are appearing pixelated, so I used AntiAliasing but it isn't helping out. As you can see the pixelated font in the image itself:
This is the code I am currently using:
butt1 = QtWidgets.QLabel("""Scrappr""")
font = QtGui.QFont()
font.setStyleStrategy(QtGui.QFont.PreferAntialias)
font.setPixelSize(22)
font.setFamily('Segoe UI Bold')
butt1.setFont(QtGui.QFont(font))
I tried different solutions on SO, qtforums etc but nothing works for me :(
I tried:
Different combinations of ClearType text but It didn't work out as, by default all the text appears good on windows and chrome but with Qt only, it becomes pixelated.
Changing windows aero theme to classic one...
But none of them helped.
Here are My PC Specs:
windows: 7 ultimate
PySide2 version: 5.14.2.1
Resolution: 1360 X 768
I'm using BrownPro font and the texts were blurry at all resolutions, but much more evident at low resolutions.
I was able to solve the issue by setting the hinting preference for the font to: PreferNoHinting. Applying it at the application level, fixes the issue everywhere.
Here is the documentation: https://doc.qt.io/qt-5/qfont.html#HintingPreference-enum
And here is the code I used:
QFontDatabase::addApplicationFont(":/fonts/BrownPro-Bold.ttf");
QFontDatabase::addApplicationFont(":/fonts/BrownPro-Regular.ttf");
QFontDatabase::addApplicationFont(":/fonts/BrownPro-Light.ttf");
QFont brown_pro_font("BrownPro");
brown_pro_font.setHintingPreference(QFont::HintingPreference::PreferNoHinting); //This line did the trick
QApplication::setFont(brown_pro_font);
Try to see the fonts used by PyQt5:
import PyQt5
from pyQt5 import QtGui
dir(QtGui.QFont)
the result will show all you need for QFond and the fonts can be used:
[..., 'Helvetica',...,'SansSerif',..., 'Serif',..., 'Times', ...
You can try to add your custom fonts but you need to test each font.
For example, the documentation tells us:
In Windows a request for the “Courier” font is automatically changed to “Courier New”, an improved version of Courier that allows for smooth scaling. The older “Courier” bitmap font can be selected by setting the PreferBitmap style strategy (see setStyleStrategy() ).
Once a font is found, the remaining attributes are matched in order of priority:
fixedPitch()
pointSize() (see below)
weight()
style()
I happen to work with Qt last year and i used qml for building the UI part of my application.
Qt itself prefers us to use qml for building UI, since they have written a UI engine that renders everything better compared to the old engines.
In case of PyQt you are using the python only approach which is only not usually recommended, i am not saying that the qml version is pixel perfect. it still works bad at drawing curves (but that is not the stuff we usually require). As far as your problem is concerned qml will work fine for you (it has much better text rendering).
You might struggle a bit finding the learning resource for qml. But at least give it a shot and yes it is easier much easier than Python only approach.
I am using Gimp python-fu plugin. The plug-in written correctly renders a text block:
(GIMP_UNIT_POINT = 3)
titleBlockLayer = pdb.gimp_text_layer_new(img, title, <font_name>, 40, GIMP_UNIT_POINT)
but only if gimp is invoked from the X window manager (or a terminal window) when logged in on the computer's console (Fedora, various releases, same behavior; various releases of GIMP, same behavior).
However, when logged in via ssh, (using X11 forwarding to display locally), the fonts are rendered HUGE by comparison.
I am confused because (supposedly) the same login .profile and .rc files have been run regardless.
Does anyone have a clue as to what/where/how GIMP determines what a POINT means, because it is obviously not coded in but dependent upon some resource file.
For Gimp a point is 1/72 of an inch, according to the doc of gimp_text_get_extents_fontname() in the PDB browser:
If you need to display a font in points, divide the size in points by 72.0 and multiply it by the vertical resolution of the image you are taking into account.
Experimentally, when using points in gimp_text_layer_new() the size depends on the image print definition. This value can depend on user settings (72DPI by default) but you can control/set it using gimp_image_get_resolution(...) and gimp_image_set_resolution(...). What bothers me though it that it seems that at least on my Ubuntu 16.04 gimp_text_layer_new() works in inches and not in points (at 72DPI, points and pixels should be equivalent), so I would stick to using pixels and to the point<->pixels conversions using the image resolution.
Btw, what are the versions of Gimp and the Pango and Cairo libs on your various systems?
Edit: Possible explanation: it appears that in recent versions (since 2.8 at least), gimp_text_layer_new() accepts more than units than just pixels and points. It also takes millimeters, inches, and picas. So it's no longer 0 for pixels and 1 for points, and you can use the predefined UNIT_* values. The catch: 1 is no longer the value for points but is now the value for inches:
UNIT_PIXEL=0
UNIT_INCH=1
UNIT_MM=2
UNIT_POINT=3
UNIT_PICA=4
I am a new to openCV and trying its examples from documentation. I am doing the trackbar as the color palette example.
But my UI renders in wrong order as well as crops the switch title: OFF.
where it is supposed to be like this.
I am using macbook pro with macOS 10.13.
OpenCV's Highgui module uses QT as a backend. QT renders elements differently depending on the OS that is using it. I believe that this is because QT uses UI elements that are native to the OS. So a frame in Windows will look like a Windows Frame, a frame in Linux will look like a Linux frame.
This is what the same example produces in Windows:
You can get the slider bars to fit the window by changing:
cv2.namedWindow('image')
to:
cv2.namedWindow('image',cv2.WND_PROP_AUTOSIZE)
Doing that produces this output (in Windows):
Is it possible in gtk3 to create a menu that is transparent? The underling window would use an image as it's background.
I can use a color for the window background but not a image.
I attempted to do what you said using an example from the gdk2 reference by adding a background image first and then porting it to gtk3. I'm no expert at gtk at all, but I did make it somehow compile:
http://pastebin.com/0XwUW5k3 (note that there has to be a "background.png" in the same folder)
The transparent dark rectangle holding the widgets is most likely the box; I tried settings its background color to full transparency as well, but it didn't work, and you'd probably have to do the composing/drawing of it yourself if you wanted it to be completely transparent, but that's not something I'd suggest because it seems too complex..
Also, you might want to create a background image with an already fitting resolution for the window, then you could skip the scaling part.
The scale function originally comes from this mailling-list thread