How to convert shape coordinates in 3D space?

Question

How to correctly calculate the 3d coordinates of the object's vertices so that it is constantly in one position relative to the camera (in front of it) during any movements and turns of the camera and always faces the camera on one side?

Obviously, we are talking about the matrix of the model, which must be built according to the coordinates of the "Look-To" point and the direction of the vector ("Look-To" - "Look-From"). I assume that you can use a view matrix for this. How to build such a matrix model?

But I was fascinated by the task itself - mathematically: in theory, everything is beautiful, but in practice it does not work at all.
And I would also like to practically assess whether there will be a gain in the speed of a single pass render from the method described by you.
@ampawd look at the text of the question before the edit, there is mentioned HUD
Yes, I first compared the figure in front of the camera with the HUD, so that the essence of the question was more understandable.
But then he corrected - as they say, "The right question is half the answer."

Accepted Answer · 2017-01-24T12:38:03

First I want to say about the matrix. Completely confused the head with these transformations. So, the matrices are needed for the shader engine when calculating the scene. If we calculate the coordinates of a single figure in a fixed environment, the use of matrices will not bring an acceleration of the calculation and understanding of the code.

I knowingly wrote in the comments about the "correctly asked question." Indeed, it was the thoughtful and corrected wording of the question that helped find the correct answer (C ++):

// исходное направление камеры: вдоль оси Х в сторону + // Построим перед камерой плоский прямоугольник, немного // сместив его (x = t) навстречу камере. После "сдвига" координат // при помощи сложения с вектором ViewTo это смещение // компенсируется. double s = 0.01, t = -0.9; std::vector<glm::vec3> p { glm::vec3(t,-s,-s), glm::vec3(t,-s, s), glm::vec3(t, s, s), glm::vec3(t, s,-s) }; // каждую вершину поворачиваем на углы положения камеры и смещаем // на вектор ViewTo (к точке, на которую направлена камера). for (size_t i = 0; i < 4; i++) { p[i] = glm::rotateZ(p[i], look_zenith); // азимут отрицательный из-за реализации поворота в // в библиотеке GLM, не совпавшей с моей моделью вращения. p[i] = glm::rotateY(p[i], 0 - look_azimuth); p[i] += ViewTo; }

In the image, the white rectangle is the plane in front of the camera, which is always stationary relative to the camera.

How to convert shape coordinates in 3D space?

1 answer 1

More articles: