如何将emgu图像转换为XNA Texture2D？

Question

Image<Bgr, Byte> video = cap.QueryFrame();
Texture2D t = new Texture2D(GraphicsDevice, video.Width, video.Height, false, SurfaceFormat.Color);
t.SetData<byte>(video.Bytes);

ArgumentException was unhandled

The size of the data passed in is too large or too small for this resource.

Answer 1

My perfered way is to 'save' the image into memory then load it with the Texture2D.FromStream function.

Texture2D t;

using(MemoryStream memStream = new MemoryStream())
{
    Image<Bgr, Byte> video = cap.QueryFrame();
    cap.save(memStream, ImageFormat.PNG);
    t = Texture2D.FromStream(GraphicsDevice, memStream, video.Width, video.Height, 1f)
}

This will work with the NonPremultiplied BlendStates but if you want to use Premultiplied alpha you should run the Texture2D though the following function. This function simply uses the GPU to quickly premultiply the texture's alpha the same as the content processor would.

    static public void PreMultiplyAlpha(this Texture2D texture) {            

        //Setup a render target to hold our final texture which will have premulitplied color values
        var result = new RenderTarget2D(texture.GraphicsDevice, texture.Width, texture.Height);

        texture.GraphicsDevice.SetRenderTarget(result);
        texture.GraphicsDevice.Clear(Color.Black);

        // Using default blending function
        // (source × Blend.SourceAlpha) + (destination × Blend.InvSourceAlpha)
        // Destination is zero so the reduces to
        // (source × Blend.SourceAlpha)
        // So this multiplies our color values by the alpha value and draws it to the RenderTarget
        var blendColor = new BlendState {
            ColorWriteChannels = ColorWriteChannels.Red | ColorWriteChannels.Green | ColorWriteChannels.Blue,
            AlphaDestinationBlend = Blend.Zero,
            ColorDestinationBlend = Blend.Zero,
            AlphaSourceBlend = Blend.SourceAlpha,
            ColorSourceBlend = Blend.SourceAlpha
        };

        var spriteBatch = new SpriteBatch(texture.GraphicsDevice);
        spriteBatch.Begin(SpriteSortMode.Immediate, blendColor);
        spriteBatch.Draw(texture, texture.Bounds, Color.White);
        spriteBatch.End();

        // Simply copy over the alpha channel
        var blendAlpha = new BlendState {
            ColorWriteChannels = ColorWriteChannels.Alpha,
            AlphaDestinationBlend = Blend.Zero,
            ColorDestinationBlend = Blend.Zero,
            AlphaSourceBlend = Blend.One,
            ColorSourceBlend = Blend.One
        };

        spriteBatch.Begin(SpriteSortMode.Immediate, blendAlpha);
        spriteBatch.Draw(texture, texture.Bounds, Color.White);
        spriteBatch.End();

        texture.GraphicsDevice.SetRenderTarget(null);

        var t = new Color[result.Width * result.Height];
        result.GetData(t);
        texture.SetData(t);
    }

Answer 2

the video stores images in (Blue, green, Red) format. Whereas texture2d expects an extra alpha pixel as well.

Image<Bgr, Byte> video = cap.QueryFrame();
Image<Bgra, Byte> video2 = video.Convert<Bgra, Byte>();
Texture2D t = new Texture2D(GraphicsDevice, video.Width, video.Height, false, SurfaceFormat.Color);
t.SetData<byte>(video2.Bytes);