Unless I make "static CPaintDC dc(this);" the line won't draw? But this is not good as it will eventually error, also the graphics wont' keep on the screen.
Not sure what I am doing wrong
Note: I have a Timer that calls to this every 100ms(x and y are incremented)
thx
void CGraphicsDlg::OnPaint()
{
CString s;
CPaintDC dc(this);// device context for painting
if (IsIconic())
{
SendMessage(WM_ICONERASEBKGND, (WPARAM) dc.GetSafeHdc(), 0);
// Center icon in client rectangle
int cxIcon = GetSystemMetrics(SM_CXICON);
int cyIcon = GetSystemMetrics(SM_CYICON);
CRect rect;
GetClientRect(&rect);
int x = (rect.Width() - cxIcon + 1) / 2;
int y = (rect.Height() - cyIcon + 1) / 2;
// Draw the icon
dc.DrawIcon(x, y, m_hIcon);
}
else if(x==0)
{
s.Format("%d", x);
edXa->SetWindowText(s);
dc.MoveTo(20,400);
}
else if (x>0){
s.Format("%d", x);
edXb->SetWindowText(s);
dc.LineTo(20 + x, 40); // doesn't draw unless I make "static CPaintDC dc(this);" <- which will error out
}
CDialog::OnPaint();
}
void CGraphicsDlg::OnTimer(UINT nIDEvent)
{
if(nIDEvent==1){
srand( (unsigned)time( NULL ) );
//y = rand() % 100;
y++;
x++;
OnPaint();
}
}
LineTo draws a line from one point to another, using the selected pen. You need to use MoveTo to define the start of the line, and you need to select a pen into the DC.
The larger problem is how you're trying to use the DC. It isn't meant to be permanent; you're supposed to acquire it, draw everything to it, then shut it down. When you try to make CPaintDC static, Windows will eventually shut it down and every attempt to use it thereafter will return an error.
The proper way is to save any coordinates that you need for all of the drawing you need to do. Use a combination of MoveTo and LineTo to draw individual line segments, and every time you reenter OnPaint you need to start over.
I am not answering your question, but have you noticed that CDialog::OnPaint() will be called even if IsIconic() returns TRUE ?
I think you will need to use an extra pair of {} to solve this ;-)
Related
I have created a class where I define Shape objects for my program. Each of these Shapes has a transparent ellipse drawn around it (I defined that in my constructor) and if any other Shape moves into that circular ellipse area, I want that Shape to change it's direction so that it moves in a circular path.
Each Shape object has a defined radius attribute (because of the ellipse I draw around each object) and I want to use that value to determine how big of a circular pattern the Shape has to move in when it collides.
Please help! Anything is greatly appreciated!
EDIT:
As I said above, I want the shape to move into a circular path. HOWEVER, I want it only to move in a circular path once (meaning it moves around a circle once) and then I want it to continue on the original path it was programmed with.
The short answer is that you'll have to use basic trig to figure out the angle between the points, and then more basic trig to determine subsequent points on the circular path.
Check out the trigonometry section of the Processing reference for more info.
But basically, if you have two points, you can use the atan2() function to calculate the angle between them. You'd use this to find the starting angle from the center of your circle to the shape.
Once you have that angle, you can simply increment it, and then use cos() and sin() to figure out the x and y coordinates at that new angle.
Here is a basic sketch that does all of the above:
PVector center;
float angle;
float radius;
void setup() {
size(500, 500);
center = new PVector(width/2, height/2);
//get the initial point
//for you, this would be the initial location of the object
PVector point = new PVector(random(width), random(height));
//find the angle between the points
float deltaX = center.x - point.x;
float deltaY = center.y - point.y;
angle = atan2(deltaX, deltaY);
//find the radius of the circle
radius = dist(center.x, center.y, point.x, point.y);
ellipseMode(RADIUS);
}
void draw() {
background(0);
//draw the center point
ellipse(center.x, center.y, 10, 10);
//find the point based on the angle
float x = center.x + cos(angle)*radius;
float y = center.y + sin(angle)*radius;
//draw the traveling point
ellipse(x, y, 10, 10);
//increment the angle to move the point
angle += PI/120;
}
Well, before I saw Kevin's post, I did one also. Not using objects, just a simple procedural example. Posting anyway :)
PVector pos, speed, stored;
float diam = 40;
boolean wonder = false;
float angle = 0;
void setup() {
size(300, 300);
// arbitrary positioning and speeding
pos = new PVector(-20, height/2);
speed = new PVector(1, 0);
noStroke();
}
void draw() {
background(5);
// normally increment speed
if (!wonder) {
pos.add(speed);
} else {
// if is to wonder...
if (angle <= 360) {
//get circle path by trig
pos.x = stored.x + cos(radians(angle))*diam;
pos.y = stored.y + sin(radians(angle))*diam;
} else {
// if the circle is complete
// reset angle and stop wondering
wonder = false;
angle = 0;
}
// increment angle
angle++;
}
// draw
ellipse(pos.x, pos.y, diam, diam);
}
void mouseClicked() {
if (isOverCircle() ) {
// store position where it has being clicked
stored = pos.get();
// off set the diam
stored.x -= diam;
// trig wondering
wonder = true;
angle = 0;
}
}
boolean isOverCircle() {
float disX = pos.x - mouseX;
float disY = pos.y - mouseY;
return sqrt(sq(disX) + sq(disY)) < diam/2;
}
I am trying to simulate a mouse click on the CView window in a legacy code which I must say I don't fully understand. The idea is to search for a particular item in the CView, get its co-ordinates and then simulate a right mouse click on it using SendInput. I want to understand if the basic steps I am following are correct before I proceed digging further into the legacy code which has a bunch of transformations happening across co-ordinate systems :( Here are the steps I follow:
Get the position co-ordinates of the item displayed in CView. at this point the co-ordinates is in the internal co-ordinate system (lets call it CDPoint).
CDPoint gPosn = viewObj->m_point_a ;
Covert the co-ordinates to the client co-ordinate system i.e CDPoint to CPoint using the existing transformations in the code.
CPoint newPosn = GetTransform().Scale(gPosn);
//Note: The basis of arriving that this is the correct transformation to use is the below code with the exact reverse transform happening in the mouse click handler code to convert CPoint to CDPoint:
`CDesignView::OnLButtonDown(UINT nFlags, CPoint p) {
CDPoint np = GetTransform().DeScale(p);
}`
Is this thinking right that CPoint received in the OnLButtonDown() handler will always be in the client co-ordinates and hence the reverse transform should convert CDPoint (internal co-ordinates) to client coordinates (CPoint) ?
Convert client co-ordinates to screen co-ordinates:
ClientToScreen(&newPosn);
Pass these values to SendInput function after converting to pixel co-ordinates:
INPUT buffer[1];
MouseSetup(buffer);
MouseMoveAbsolute(buffer, newPos.x, newPos.y);
MouseClick(buffer);
The Mousexxx() functions are defined as below similar to the sample code in this post:
How to simulate a mouse movement
.
#define SCREEN_WIDTH (::GetSystemMetrics( SM_CXSCREEN )-1)
#define SCREEN_HEIGHT (::GetSystemMetrics( SM_CYSCREEN )-1)
static void inline makeAbsXY(double &x, double &y) {
x = (x * 0xFFFF) / SCREEN_WIDTH ;
y = (y * 0xFFFF) / SCREEN_HEIGHT ;
}
static void inline MouseSetup(INPUT *buffer)
{
buffer->type = INPUT_MOUSE;
buffer->mi.dx = (0 * (0xFFFF / SCREEN_WIDTH));
buffer->mi.dy = (0 * (0xFFFF / SCREEN_HEIGHT));
buffer->mi.mouseData = 0;
buffer->mi.dwFlags = MOUSEEVENTF_ABSOLUTE;
buffer->mi.time = 0;
buffer->mi.dwExtraInfo = 0;
}
static void inline MouseMoveAbsolute(INPUT *buffer, double x, double y)
{
makeAbsXY(x,y) ;
buffer->mi.dx = x ;
buffer->mi.dy = y ;
buffer->mi.dwFlags = (MOUSEEVENTF_ABSOLUTE | MOUSEEVENTF_MOVE);
SendInput(1, buffer, sizeof(INPUT));
}
static void inline MouseClick(INPUT *buffer)
{
buffer->mi.dwFlags = (MOUSEEVENTF_ABSOLUTE | MOUSEEVENTF_RIGHTDOWN);
SendInput(1, buffer, sizeof(INPUT));
Sleep(10);
buffer->mi.dwFlags = (MOUSEEVENTF_ABSOLUTE | MOUSEEVENTF_RIGHTUP);
SendInput(1, buffer, sizeof(INPUT));
}
Could anyone pls provide pointers on what might be going wrong in these steps since the simulated mosue click always seem to be shifted left by some factor which keeps increasing as x becoems larger. I have verified that is gPosn is pointing to (0,0) it always simulates a mouse click on the top right corner of the client screen.
Thanks for your time.
If you have x and y in client coordinates, you have to convert them to screen coordinates:
POINT point;
point.x = x;
point.y = y;
::ClientToScreen(m_hWnd, point);
Where m_hWnd is the window which owns the objects. x and y are relative to top-left of the client area of this window.
Assuming point.x and point.y are in screen coordinates, the rest of the conversion for SendInput is correct. You can also create INPUT array for SendInput, this will send the mouse messages without interruption.
INPUT input[3];
for (int i = 0; i < 3; i++)
{
memset(&input[i], 0, sizeof(INPUT));
input[i].type = INPUT_MOUSE;
}
input[0].mi.dx = (point.x * 0xFFFF) / (GetSystemMetrics(SM_CXSCREEN) - 1);
input[0].mi.dy = (point.y * 0xFFFF) / (GetSystemMetrics(SM_CYSCREEN) - 1);
input[0].mi.dwFlags = MOUSEEVENTF_ABSOLUTE | MOUSEEVENTF_MOVE;
input[1].mi.dwFlags = MOUSEEVENTF_RIGHTDOWN;
input[2].mi.dwFlags = MOUSEEVENTF_RIGHTUP;
SendInput(3, input, sizeof(INPUT));
Working in Processing, I am trying to build my first generative patch. What I want to have happen is start drawing a circle somewhere on screen (a point following the path of a circle), but after a random amount of time, the circle breaks, the line goes in a random direction for a random amount of time, and begins drawing a new circle elsewhere.
Right now I have the circle being drawn, and I have a toggle mechanism that turns on and off after a random period of time. I can't figure out how to get it "break" from that original circle, let alone get it to start a new circle elsewhere. Would anybody have some advice on how to accomplish this? I think it might have an interesting visual effect.
Rotor r;
float timer = 0;
boolean freeze = false;
void setup() {
size(1000,600);
smooth();
noFill();
frameRate(60);
background(255);
timeLimit();
r = new Rotor(random(width),random(height),random(40,100));
}
void draw() {
float t = frameCount / 100.0;
timer = timer + frameRate/1000;
r.drawRotor(t);
if(timer > timeLimit()){
timer = 0;
timeLimit();
if(freeze == true){
freeze = false;
}else{
freeze = true;
}
background(255);
}
}
float timeLimit(){
float timeLimit = random(200);
return timeLimit;
}
Rotor Class:
class Rotor {
color c;
int thickness;
float xPoint;
float yPoint;
float radius;
float angle = 0;
float centerX;
float centerY;
Rotor(float cX, float cY, float rad) {
c = color(0);
thickness = 1;
centerX = cX;
centerY = cY;
radius = rad;
}
void drawRotor(float t) {
stroke(c);
strokeWeight(thickness);
angle = angle + frameRate/1000;
xPoint = centerX + cos(angle) * radius;
yPoint = centerY + sin(angle) * radius;
ellipse(xPoint, yPoint,thickness,thickness);
}
}
First to answer your question about "breaking" circle: you need to create new rotor instance or just change its properties like center and radius. If I got your idea right you just need one instance of rotor so just change this values:
r.centerX = newX;
r.centerY = newY
r.radius = random(40,100) //as you have in setup
But how you can calculate new position? It could be random but you want to create path so you need to calculate it. And here comes the tricky part. So how to make connecting line and start new circle?
First you will need two mode. First will draw circle second will draw line. Simplest way to achieve that is by updating rotor draw method [You can pass mode variable as parameter of drawRotor function or as global variable]:
if(mode == 1){
angle += frameRate/1000;
}else{
radius += 2;
}
As you can see I just differ between increasing angle to draw circle and increasing radius to draw line (not in random direction but in way from center). Then we will need to calculate new position of circle's center. To do this we simple calculate how it would continue according to angle and substitute new radiusso whole part will looks like this:
if(mode != 1){
float newR = random(40,100);
float newX = r.centerX + cos(r.angle) * (r.radius - newR);
float newY = r.centerY + sin(r.angle) * (r.radius - newR);
r.newPos(newX, newY);
r.radius = newR; //we cant change it earlier because we need also old value
}
This will happen inside your "time handler" function only when you change mode back to drawing circle. Mode can be simple changed within handler
mode *= -1; //but need to be init to 1 inside setup()
If you want to have path always visible just delete background() function but if you want some cool effect add this at the begging of draw()
noStroke(); //No stroke needed and you turn it on again in drawRotor()
fill( 255,255,255, 10 ); //This will set transparency to 10%
rect(0,0,width,height); //You put layer after each "point" you draw
noFill(); //This will restore fill settings as you have before
Here I paste whole code just for demonstration and you should modify it according your own purpose. Better to code own version.
The call to background()usually comes as first thing in draw. That's because the draw only renders at the end of each loop (frame). So calling bg at the beginning will clear all stuff drawn in last frame. If you need to persist the draws trough frames can either remove the call to background() or draw your stuff every frame. Or yet draw stuff in a PGraphics and display it.
The other thing is each time the 'Rotor' stops you should give it new random coordinates.
If you go for removing the background() call this will do the trick:
Rotor r;
float timer = 0;
boolean freeze = false;
void setup() {
size(1000,600);
smooth();
noFill();
frameRate(60);
background(255);
timeLimit();
r = new Rotor(random(width),random(height),random(40,100));
}
void draw() {
float t = frameCount / 100.0;
timer = timer + frameRate/1000;
r.drawRotor(t);
if(timer > timeLimit()){
timer = 0;
timeLimit();
//***** here new coordinates!!
r = new Rotor(random(width),random(height),random(40,100));
//*****
if(freeze == true){
freeze = false;
}else{
freeze = true;
}
//***** no background()
// background(255);
}
}
float timeLimit(){
float timeLimit = random(200);
return timeLimit;
}
class Rotor {
color c;
int thickness;
float xPoint;
float yPoint;
float radius;
float angle = 0;
float centerX;
float centerY;
Rotor(float cX, float cY, float rad) {
c = color(0);
thickness = 1;
centerX = cX;
centerY = cY;
radius = rad;
}
void drawRotor(float t) {
stroke(c);
strokeWeight(thickness);
angle = angle + frameRate/1000;
xPoint = centerX + cos(angle) * radius;
yPoint = centerY + sin(angle) * radius;
ellipse(xPoint, yPoint,thickness,thickness);
}
}
now, if you need to clear the screen, You can make a List (ArrayList?) and add a new Rotor to it when the previous is done. But you need to manage the Rotor to be able to display it self without animating as well. So new created Rotor would animate, and old ones would just display its arc without animating. Or make a PGraphis with no call to bg and display it in main canvas that can have a bg call...
A side note, be aware that relying in frameRate to times stuff makes it dependable on the system performance. You can do the same thing using millis()to avoid that. Not an issue so far, as this is very light yet, but may become an issue if the project grows further.
I'm trying to display circles at a user accepted input (usually centers), using OpenCV 2.4.3 (VS 2010). On output image (displayed using 'namedWindow') circle seems to shift column-wise as one marks points along columns. Not sure how I should correct this.
Code:
struct OPTIONS{
OPTIONS(): X(-1), Y(-1), drawing_dot(false){}
int X;
int Y;
bool drawing_dot;
};
OPTIONS options;
void my_mouse_callback( int event, int x, int y, int flags, void* param ){
IplImage* image = (IplImage*) param;
switch( event ){
case CV_EVENT_LBUTTONDOWN:
options.X = x;
options.Y = y;
options.drawing_dot = true;
break;
default:
break;
}
}
int main( void ){
IplImage* image = cvLoadImage("Images/TestRealData/img1.bmp");
Mat frame = imread("Images/TestRealData/img1.bmp");
namedWindow("Test", CV_WINDOW_KEEPRATIO);
cvSetMouseCallback("Test", my_mouse_callback, (void*) image);
while( cvWaitKey(15) != 27 ){
if( options.drawing_dot ){
circle(frame, Point(options.X,options.Y), 3, CV_RGB(0,0,255), 2);
options.drawing_dot = false;
}
imshow("Test", frame);
waitKey(10);
}
cvReleaseImage(&image);
return 0;
}
I think the circle does not shift. The mouse cursor may trick our eyes. You may simply check it by increasing the radius and reduce the thickness of the circle outline like:
circle(frame, Point(options.X, options.Y/2), 15, CV_RGB(0, 0, 255), 1);
By the way, I think if you want to draw the circle at the point you click on, options.Y should not be divided by 2.
Found answer after much time lost - Make sure to specify flags in 'namedWindow'. Changing flag to CV_WINDOW_KEEPRATIO did the trick for me. Hope this helps somebody out there.
I've got a little text drawing puzzle under Win32. I'm trying to draw some instructions for users of my application at the top of the window.
Please refer to the following window (I've changed the background color on the text so you can see the boundaries)
(source: billy-oneal.com)
I'm currently using DrawTextEx to draw the text to my window, but the problem is that it does not fill the entire RECTangle that I give it. Not drawing that area is just fine, until the window resizes:
(source: billy-oneal.com)
When the text is re wrapped due to the window sizing, because DrawTextEx doesn't clear it's background, these artifacts are leftover.
I tried using FillRect to fill in the area behind the text drawing call, which does eliminate the visual artifacts, but then causes the text to flicker constantly, as it is completely erased and then completely redrawn to the display.
Any ideas on how one might get the area not containing text to be drawn with the background color?
EDIT: I'd like to avoid having to double buffer the form if at app possible.
EDIT2: I solved the problem by only redrawing the text when I detect that the wrapping changes during a resize.
Use double buffering?
Draw everything to a bitmap and draw the bitmap to the window. Flickering is commonly a double buffering issue.
There are many possible solutions and without seeing your code, it's hard to tell which method would be best so I'd suggest taking a look at this article on flicker free drawing
SetBkMode + SetBkColor ?
Well since nobody seems to know what to do about it, I implemented it this way:
std::vector<std::wstring> wrapString(HDC hDC, const std::wstring& text, const RECT& targetRect, HFONT font)
{
std::vector<std::wstring> result;
RECT targetRectangle;
CopyRect(&targetRectangle, &targetRect);
//Calculate the width of the bounding rectangle.
int maxWidth = targetRectangle.right - targetRectangle.left;
//Draw the lines one at a time
std::wstring currentLine;
for(std::wstring::const_iterator it = text.begin(); it != text.end(); currentLine.push_back(*it), it++)
{
if(*it == L'\r' || *it == L'\n')
{ //Hard return
while(it != text.end() && (*it == L'\r' || *it == L'\n')) it++;
result.push_back(currentLine);
currentLine.clear();
}
else
{ //Check for soft return
SIZE sizeStruct;
GetTextExtentPoint32(hDC, currentLine.c_str(), static_cast<int>(currentLine.length()), &sizeStruct);
if (sizeStruct.cx > maxWidth)
{
std::wstring::size_type lineLength = currentLine.find_last_of(L' ');
if (lineLength == currentLine.npos)
{ //Word is longer than a line.
for(;it != text.end() && !iswspace(*it);it++) currentLine.push_back(*it);
}
else
{ //Clip word to line.
//Backtrack our scan of the source text.
it -= currentLine.length() - lineLength - 1;
//Remove the clipped word
currentLine.erase(lineLength);
}
result.push_back(currentLine);
currentLine.clear();
}
}
}
//Last remaining text.
result.push_back(currentLine);
return result;
}
void DrawInstructionsWithFilledBackground(HDC hDC, const std::wstring& text, RECT& targetRectangle, HFONT font, COLORREF backgroundColor)
{
//Set up our background color.
int dcIdx = SaveDC(hDC);
HBRUSH backgroundBrush = CreateSolidBrush(backgroundColor);
SelectObject(hDC, backgroundBrush);
SelectObject(hDC, font);
SetBkColor(hDC, backgroundColor);
std::vector<std::wstring> lines(wrapString(hDC, text, targetRectangle, font));
for(std::vector<std::wstring>::const_iterator it = lines.begin(); it!=lines.end(); it++)
{
RECT backgroundRect = targetRectangle;
DrawText(hDC, const_cast<LPWSTR>(it->c_str()), static_cast<int>(it->length()), &backgroundRect, DT_CALCRECT | DT_NOCLIP | DT_SINGLELINE);
backgroundRect.left = backgroundRect.right;
backgroundRect.right = targetRectangle.right;
if (backgroundRect.right >= backgroundRect.left)
FillRect(hDC, &backgroundRect, backgroundBrush);
ExtTextOut(hDC, targetRectangle.left, targetRectangle.top, ETO_OPAQUE, NULL, it->c_str(), static_cast<UINT>(it->length()), NULL);
targetRectangle.top += backgroundRect.bottom - backgroundRect.top;
}
instructionsWrap = lines;
//Restore the DC to it's former glory.
RestoreDC(hDC, dcIdx);
DeleteObject(backgroundBrush);
}
Get/Calculate the rect used by the DrawText call and clip it with something like ExcludeClipRect before calling FillRect