C++ Direct2D 矩阵变换镜像
我目前正在学习 DirectX（2010 年 6 月的 SDK）并正在将位图绘制到屏幕上。现在我希望能够对这些图像进行转换，并且无法沿 x 或 y 轴镜像图像。
在绘图之前，我可以设置一个将应用于绘图操作的变换矩阵。尽管 DirectX 使用 D2D1_MATRIX_3x2_F 类型，但根据 MSDN，只有该矩阵的 2x2 部分用于转换。例如，矩阵 { [1, 0], [0, 1], [0, 0] } 是使所有内容自然显示的单位矩阵。
据我了解，沿 x 轴镜像图像的矩阵变换应如下所示： { [1, 0], [0, -1], [0, 0] }，沿 y 轴镜像图像那么将是 { [-1, 0], [0, 1], [0, 0] }。但是，如果我使用这些矩阵进行转换，则会出现黑屏。
对我来说，问题是，我现在有点陷入困境，不知道这个问题可能是由什么引起的，或者更确切地说，它是否是由其背后的数学或实现本身的逻辑错误引起的。不幸的是，除了 Direct2D RenderTarget 头文件中的一行注释“镜像必须通过矩阵变换来完成”之外，我找不到任何相关材料。
我目前用于测试的源代码是：
D2D1_MATRIX_3X2_F m;
m._11 = 1; m._12 = 0;
m._21 = 0; m._22 = -1;
m._31 = 0; m._32 = 0;
m_pDX2DRenderTarget->SetTransform( m );
I am currently learning DirectX (SDK of June 2010) and am drawing bitmaps to screen. Now I want to be able to do transformations on these images, and fail to mirror the images along the x- or y- axis.
Before drawing, I am able to set a transformation matrix that will be applied to the drawing operations. Though DirectX is using the D2D1_MATRIX_3x2_F type, according to MSDN, only a 2x2 part of this matrix is used for transformations. For example, the matrix { [1, 0], [0, 1], [0, 0] } is the identity matrix that causes everything to be displayed naturally.
As I understand it, a matrix transformation to mirror my image along the x-axis should look like this: { [1, 0], [0, -1], [0, 0] }, and along the y-axis it would then be { [-1, 0], [0, 1], [0, 0] }. If I use these matrices for the transformation however, I get a black screen.
The problem for me is, that I am kind of stuck at this point without ideas what this problem could be caused by, or rather whether it is caused by a logical error in the mathematics behind this, or the implementation itself. Unfortunately, I could not find any material on this, besides a one line comment that "mirroring has to be done with matrix transformations" in the header file of the Direct2D RenderTarget.
The source code I am currently using for testing this is:
D2D1_MATRIX_3X2_F m;
m._11 = 1; m._12 = 0;
m._21 = 0; m._22 = -1;
m._31 = 0; m._32 = 0;
m_pDX2DRenderTarget->SetTransform( m );
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
我完全错误地认为矩阵的第三行不相关。它实际上是转换的
x-和y-偏移量。变形后我看不到任何东西的原因是，应用镜子后，不仅图像本身会被镜像，而且它在屏幕上的位置也会被镜像。例如，如果我显示图像，使其左上角边缘位于
(100, 100)，那么在应用 x 轴镜像的矩阵变换后，图像的右上角边缘将是在(-100, 100)。由于它始终位于渲染区域之外，因此不会显示。I was completely mistaken about the third row of the matrix not being relevant. It is in fact the
x-andy-offset for the transformation. The reason why I did not see anything after my transformation is that, after applying a mirror, not only the image itself will be mirrored, but also its position on screen.For example, if I display an image so its top left edge will be at
(100, 100), after applying the matrix transformation of a x-axis mirror, the top right edge of the image will be at(-100, 100). Since this is always out of the rendering area, it will not show.